Patent Application
20230332228
The present invention is in the field of Nucleic Acid (NA) editing, such as gene editing, and more specifically relates to techniques for determining the effects of an NA editing procedure on Nucleic Acids (NA) acquired from a given source.
References considered to be relevant as background to the presently disclosed subject matter are listed below:
Nucleic Acid (NA) editing techniques present powerful tools which may be used inter-alia for curing genetic illnesses, genome editing in mammalians, for enhancement as well as for treatment, in crop engineering and in many other applications3.
One widely investigated Nucleic Acid (NA) editing techniques is known as Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) genome/NA editing. CRISPR utilizes guide RNA (gRNA)-directed Cas nucleases (hereinafter referred to as CRISPR-Cas) to induce double-strand breaks (DSBs) in the treated genome, has shown promising preliminary results as an approach for definitively curing a variety of genetic disorders1. Additional genome editing techniques that are induced by engineered nucleases, including, but not limited to, are Zinc finger nucleases (ZFNs), transcription-activator like effector nucleases (TALEN), and meganucleases.
Several methodologies have been developed to detect off-target activity in an unbiased manner7. The unbiased methods can detect unintended cleavage sites on the whole genome level, without the need for predetermined focus. One such approach is GUIDE-seq (Genome-wide, Unbiased Identification of DSBs Enabled by Sequencing)8. GUIDE-seq relies on the introduction of short tag DNA sequence that is transfected into the cells and integrated at the Cas9-induced DSB via NHEJ. The subsequent sequencing of tag-adjacent regions identifies potential off-target sites. Other Unbiased approaches include CIRCLE-seq9, SITE-seg10, and DISCOVER-Seq11, among others. There are various bioinformatics analysis tools20, 21, 22, 23, 24, 25, 26, 27, 28, 29, previously suggested for estimation of genome-editing activity rates, based on sequencing data of the edited genome-sequences.
Other known in the art tools utilize a so called treatment (Tx) vs. mock (M) approach (also referred to herein interchangeably as Tx vs. M) in order to assess the rates of gene editing events/indels at given sites introduced by CRISPR to the genome treated sequences (Tx). This is achieved by processing/sequencing of treated genome sequences (Tx), which were treated by CRISPR as well as processing/sequencing of comparable/similar none treated genome sequences, mock (M), e.g. which are similar to the treated sequences prior to their treatment by CRISPR), and comparing the rates of indels found at the given sites in both the treated and mock specimens/sequences.
For example, CRISPResso (see e.g. CRISPResso1/231, 32) and AmpliCan33, are two widely used tools which utilize Tx vs M data to assess INDEL activity rates of the genome editing treatment by subtract the separately inferred INDEL rates from each of the treatment (Tx) and mock (M) specimens.
Conventional methods for de novo translocation detection, AMP-Seq17, 8 (Anchored Multiplex-PCR sequencing), High-Throughput, Genome-wide, Translocation Sequencing (HTGTS)18, and Uni-Directional Targeted Sequencing (UDiTaS)19, identify translocations which involve particularly selected DSB sites (on-target or off-target) which might be inflicted by genome editing using site-directed nucleases.
There is a need in the art for a technique capable of determination of editing activity, particularly off-target/adverse editing activity of Nucleic Acid (NA)/genome techniques. Further, there is a need in the art for a novel technique capable of quantifying the occurrence off-target activities/adverse effects over a broad spectrum of possible editing events (e.g. double strand breaks (DSBs) which might be introduced to various NAs subjected to editing by particular one or more NA editing procedures, indels of various types or classes). Also, there is a need in the art for a novel technique enabling identification and possibly quantification of translocations occurring due to NA editing processes/procedure.
Present NA editing techniques, such as CRISPR-based genome editing systems, generally lack the required genome editing specificity and accuracy, which is required for in various NA/genome editing applications. The lack of specificity is manifested for example by off-target activity of genome editing systems, which may in turn introduce unwanted and un-stable/predictable genome editing events at undesired sites/loci. This presents a major drawback for the usability of gene editing systems for various applications since the unwanted genome/NA editing events may result in unwanted mutations and unwanted genomic structural variations, such as translocations.
In this regard, it might be noted that NA editing systems, such as CRISPR-Cas endonucleases, did not naturally evolve to function as a highly specific gene-editing mechanism, certainly not in the context of mammalian genomes. Using these bacterial nucleases in mammalian, plant, and other types of cells often entails off-target activity, leading to unintended DNA breaks at other sites in the genome with only partial complementarity to the gRNA sequence2.
To bring CRISPR technology, or any other Nucleic Acid (NA) editing techniques (e.g. with engineered nuclease), to safe broader use in the clinic or for genome editing in mammalian, crop engineering, synthetic NA editing, and/or other applications3, 4, 5, 6, it must be highly active at the on-target site (which is the genome site/locus intended for nucleases by the CRISPR mechanism) and have minimal off-target editing adverse effects (i.e. having minimal editing effects at off-target genome sites/loci being sites/loci other than the on-target sites/loci).
Potential off-target activities NA editing procedures, such as CRISPR-based procedures, present major pitfalls for using this technique of genome editing, due to potentially unwanted NA editing events which may lead for example to genome instability such as resulting mutations (e.g. due to unwanted indels) and/or NA/genomic structural variations (e.g. due to unwanted translocations).
For example, the Cas9 endonuclease can create DSBs at undesired off-target locations, even in the presence of mismatches. This may lead to adverse effects such as indels and translocations.
It should be noted that in the description herein below the following terms/phrases should be understood to encompass at least as follows:
Considering the potential off-target activities NA editing procedures, it might be noted that it is possible to design different NA editing procedures (e.g., different CRISPR-based procedures) to targeting the same or similar on-target location, and having the same/equivalent required NA editing effect on the on-target location, but having different adverse effects or different off-targets' activities and/or different probabilities to induce these different off-targets' activities. This might be achieved for instance by different designs of the guide RNA of NA editing procedures, such as CRISPR-based procedures) to target/bind the similar on-target location, but with different likelihoods to bind to undesired off-target locations.
To this end, one approach to overcome these drawbacks (e.g. the lack of specificity) of general NA editing procedures, rely on the ability to accurately inspect the on/off target effects of one or more NA/gene editing procedures (e.g. having different guide RNAs) designed for a particular NA editing application (e.g. to a particular on-target activity) to determine their respective specificities to the required on-target location, or otherwise their respective off-target activities (adverse effects). This will enable selection of a proper NA/gene editing procedure, for a given NA editing application, which has sufficiently high specificity and/or, sufficiently negligible adverse effects, in terms of their probabilities of occurrence by the NA editing or in terms of the significance of the adverse effects introduce thereby (e.g. according to the expected effects of editing in the particular off-target loci affected thereby).
To achieve that there is a need for a technique to accurately and reliably assess the performance of NA/genome editing procedures, and particularly to be capable of accurately identifying and quantifying the resulting off-target activities of such NA/genome editing procedures, when applied to NA sequences of the required NA editing application.
However, the conventional techniques for assessing the performance of gene editing systems are lacking in various respects.
Conventional ‘unbiased’ techniques, such as GUIDE-seq8 CIRCLE-seq9, SITE-seq10, and DISCOVER-Seq11, have been established identifying potential genome-wide off-target activity caused by NA editing procedures such as CRISPR-Cas9 genome editing. These techniques are used to detect/determine potential off-target activity, and provide data indicative of potential off-target sites of a genome editing procedure. These techniques are however note capable of validating or quantifying the actual occurrence of adverse effects in those potential off-target sites by the gene editing procedure and require further validation and/or quantification of the actual occurrence of such potential effects by techniques (for example by utilizing targeted amplicon sequencing with primers designed for the reported genomic loci at which adverse effect is expected (e.g., rhAmpSeq12, 13). These techniques are not capable of identifying translocations and not quantitatively measuring any other editing activity of genome editing procedures, but can identify potential off-target locations of genome editing in an unbiased manner.
The off-target activity identification obtained from bioinformatics analysis tools20, 21, 22, 23, 24, 25, 26, 27, 28, 29, which rely solely on the sequencing treated/edited genome/DNA-sequences (e.g. NGS data of the edited genome), is also deficient and inaccurate. This is because the off-target activities identified by such tools may be obscured by sequencing errors introduced to the NGS data. More specifically, in the process of generating sequencing data for the edited genome/DNA-sequences, a number of errors are/maybe generally introduced by the sequencing it-self, and may then be represented in the raw sequencing data. These errors/events may come from various sources (e.g. the sequencing platform, polymerase errors, and library preparation buffers such as, base oxidation)30). These errors from the sequencing itself can often occur near or at the site of the CRISPR-induced DSB, thus limiting the accuracy of bioinformatics analysis tools aimed to quantify off-target events with low activity rates. This is because it is generally difficult to distinguish sequencing/NGS errors from actual edit events.
The conventional tools based on the so called treatment (Tx) vs. mock (M) approach, such as CRISPResso and ampliCan, are limited by their ability to only roughly assess adverse effects of only INDEL category. These methods are based on a simple subtraction of the inferred INDEL activity rates of the mock (M) from those in the treatment (Tx). These techniques are incapable of providing any statistical assessment/determination of whether an adverse effects of each certain type, which is observed in the treatment (Tx), actually occurred due to an edit event, and are also not adapted for segregating the INDEL activity rate inferred thereby into different indel types. Moreover these techniques also do not provide any statistical evaluation or confidence intervals for the inferred rates of INDEL activity.
CRISPResso is for example deficient in that it provides only a combined Tx vs M indel activity rates at a given site of the treated and mock sequences, while not being stratified into different indel types which may occur at said given site. This may yield a problematic and inaccurate assessment of the indel activity rates of the CRISPR, particularly for example in cases where a greater incidence of indel events of a certain type (e.g. which may be unique caused by the gene-editing/CRISPR treatment), are masked by other types of indel events/pseudo-events at the given site, which may appear in relative abundances at the given site, due for example to sequencing artifacts or other/natural processes.
Although, ampliCan does attempt to separately present the different modification/indel types produced by the genome editing, it performs a simple background subtraction for each modification type (Tx-M). However also this technique is incapable of providing any statistical assessment/determination of whether an adverse effects of each certain type, which is observed in the treatment (Tx), actually occurred due to an edit event. This technique, is also incapable of providing a statistical estimate of the inferred rates or of the obtained differences for each modification, and the simple background subtraction may often lead to abused results (such as negative inferred rates of certain indel types in the treatment. Furthermore, ampliCan accounts for different modifications as belong to the same indel type only if they are identical (same position in the reference sequence and with identical base pairs). Namely, ampliCan does not aggregate modifications with similar characteristics, such as insertions at the same reference position and with identical lengths, but with different base pairs. Thus, ampliCan is sensitive to ‘noisy’ indels which originated by amplification, PCR or sequencing errors.
Moreover, additional deficiency of ampliCan's technique is that it is only known/demonstrated to perform with singleplex PCR amplification. This means that the technique is limited to the detection of indels at only one site (typically the on target site) There are, in particular, no performance indications as related to detecting off-target activity by AmpliCan.
Translocations represent a group of possible adverse consequences of off-target editing, that can be particularly devastating even when occurring at low frequencies. For instance deleterious genomic structural variation events are structural variation that can lead to the onset of several human disease conditions including many types of cancer, infertility, and other acquired genomic disorders14, 15. Translocations, such as Chromosomal translocations and large deletions, are structural variation that can arise when on-target/off-target or off-target/off-target cut-sites (on the same or on different Chromosomes) fuse as a result of DSBs at both loci16.
Therefore, translocations and other adverse/undesired structural variation effects of NA editing require thorough investigation to understand their prevalence, characteristics, and the conditions of an NA editing procedure promoting or repressing their formation, prior to the administration of the NA editing procedure for any particular use.
Conventional techniques, such as those directed for de novo translocation detection (e.g. the above mentioned AMP-Seq17, 8, HTGTS18, and UDiTaS) address a fixed predetermined list of potential events on one side. For example, in UDiTas genomic sequences are amplified by one sequence-specific primer, targeted to a specific genomic locus, on one side, and a second tagmentation-mediated primer on the other side. Thus, this technique allows for identifying any edit events, including structural variations, as well as any translocation partners of a specific off-target site, at a particular predetermined genomic location.
DSBs are important components of the process leading to the translocation, and therefore there is a need in the art for a technique that is capable of investigate all the possible pairwise translocation events that can take place as a result of an NA editing procedure associated with set of potential off-target sites (e.g, potential off-target sites which may be identified for example by techniques such as GUIDE-seq, Circle-Seq, etc).
However, the conventional de novo translocation detection techniques are inherently deficient in this respect, due to their limited ability to identify only those translocations/genome-editing events, which involve DSB at one or more of the limited set of selected DSB sites which are monitored thereby. This is generally insufficient since off-target gene editing events may occur also at locations other that the particular limited set of DSB sites monitored by such techniques, and particularly translocations can involve various combinations of off-target loci/sites or may occur between the off-target loci and spontaneous breaks. It is also important to note that most or all of these existing techniques are also deficient in that they are designed or not suitable for processing sequencing results from multiplexed amplification, by which multiple potential on and off-target effects may be simultaneously observed.
The present invention provides a novel technique for identification/detection and/or quantification of adverse effects actually occurring due to an NA editing procedures, such as translocations and indel types. Accurate determining of the adverse effects of NA/genum editing is essential in the field of genetics and genomics to support the assessment of the effectiveness and side effects of a genome editing protocol/procedure. The techniques of the present invention solve the deficiencies of the conventional techniques and facilitates accurate statistical determination/assessment of the actual occurrence of various respective types of those adverse effects over the entire spectrum of NA sequences/sites that are expected to be potentially affected by each NA editing procedure. The invention thus provides the ability to assess the specificities and/or adverse-effects of various NA/genome editing procedure (e.g. assess their accuracy, precision, specificity and/or adverse effects in the context of a given genome editing application) and to thereby select the most suitable NA editing configuration/procedure for a given NA/genome editing application, having the high specificity and reduced or insignificant adverse effect for that application.
The technique of the present invention is based on the so called treatment (Tx) vs. mock (M) approach and utilizes advanced modeling for the analysis of Tx vs M multiplex-amplification/PCR data to obtain accurate measurement (determination and/or quantification) of off-target activity of an NA editing procedure with the ability to accurately determine indel types as well as translocations.
The technique of the present invention facilitates accurate statistical detection/determination of indel types occurring due to an NA edit procedure, as well as, statistical quantification thereof by utilizing a statistical model based on a comparative statistical model approach for the indel quantification. Alternatively, or additionally, the technique of the present invention facilitates accurate detection, quantification and statistical assessment of observed translocation. This is achieved by the detection of alternative cut-sites in off-target loci and by proper modeling of the identified alternative cut sites in the mock and treatment data. The technique of the present invention is suitable of analysis of multiplex-PCR and NGS data, based on which it can detect various types of indels and structural variations including deletions and insertions (indels) as well as translocation events occurring in an NA editing procedure, and to output data indicative of the types of adverse effect actually occurring due to the an NA edit procedure (as statistically determined) and possibly also the off-target activity rates of those actually occurring adverse effects as well as confidence intervals for the inferred off-target activity rates.
The technique of the present invention is advantageous inter-alia in that it enables for a comprehensive evaluation of all possible indels and/or translocations among the predicted off-target sites addressed by a single multiplex-PCR, while obviating a need to perform a multiple additional experiments or PCR amplifications. In other words, the technique of the present invention facilitates to determine various translocation events involving the potential off-target sites of an NA editing procedure with as little as a single multiplex amplification of the edited NA sequences.
Thus according to one broad aspect of the present invention there is provided a method for determining effects of Nucleic Acid (NA) editing procedure. The method includes:
In some implementations the said one or more types of adverse effect are classified to one or more classes of adverse effects, each class being characterized by the one or two involving sites [λm1,λm2], with which adverse effects of the class are involved, whereby each class belongs to one of two categories of adverse effects:
According to some embodiments of the present invention the method also includes one or more of the following preliminary operations:
The sequencing data may then be processed by a processor for constructing, per each particular type of adverse effect of one or more types of possible adverse effects of said NA editing procedure, a statistical model of occurrence of said type of adverse effect by the NA editing procedure, and applying said statistical model to said sequencing data to statistically determine actual occurrence of said type of adverse effect by the NA editing procedure. The data indicative of the of whether said each type of adverse effect by actually occurs due to the NA editing procedure, may be output enable determination of safety of the NA editing procedure.
According to some embodiments the processing further comprising utilizing the statistical model to quantify the types of adverse effects actually affected by the NA editing procedure by determining rates of occurrence thereof by the NA editing procedure, and a statistical confidence intervals for said rates.
According to some embodiments the one or more types of adverse effect are classified to one or more classes of adverse effects, each class being characterized by the one or two participating sites [λm1,λm2], with which adverse effects of the class are associated, whereby each class belongs to one of two categories of adverse effects:
According to some embodiments the processing said sequencing data for the particular type of adverse effect includes obtaining the statistical determination of the occurrence of said particular type of adverse effect by the NA editing procedure in the tested cell types or in related samples or clinical material.
According to some embodiments the multiplexed amplifications of the edited and control collections are conducted utilizing respective multiplex PCR processes with a similar selected set of primer molecule types{PRt}. The method includes providing the selected set of a plurality of primer molecule types {PRt} including primer molecule types selected according to said target data, such that the plurality of primer types {PRt} comprise, or constitutes of, matched pairs (PRM+m, PRM−m) of forward PRM+m and reverse PRM−m primer molecule types {(PRM+m, PRM−m)}1M ∈{PRt} suitable for amplification of said on-target and off-target sites {λm1}1M in the edited and control NA collections. These may for example be used in a One-Step-Amplification or in the 1st step of a Two-Step-Amplification.
In some implementations the processing includes a preliminary preprocessing of the sequencing data for adjusting said reads of said multiplexed amplifications' products/amplicons from each of the edited and control collections by carrying out at least one of the following:
According to some embodiments the method is adapted for determining at least one indel type T of said one or more of types of adverse effect of the NA editing procedure, which belong to the Category 1 of adverse effects that is associated with INDEL activity of said NA editing procedure, and which belong to at least one class of adverse effects associated with a respective site/locus of interest λm1.
According to some embodiments the processing of the sequencing data includes matching reads of said sequencing data to the site/locus of interest λm1 by carrying out the following:
The ‘collective’ count match condition of the template statistical model of Category 1 of adverse effects may for example be satisfied for a read in case the prefix and suffix and regions of the read match prefix PRS+m and suffix PRS-m primer sequences (PRS+m1, PRS−m1) of the respective site of interest λm1.
In some implementations the sizes of the respective collections LTx(λm1) and LM(λm1) are used as the ‘collective’ counts NTx and NMc of reads of amplicons which are associated with said at least one class of adverse effects involving the site λm1 observed in the sequencing data of the corresponding edited and control collections.
In some embodiments the processing of the sequencing data by includes segregating the collections LTx(λm1) and LM(λm1) of reads matching the site of interest λm1, to form at two sub-collections LTx(λm1,T) and LM(λm1,T) of reads presenting a certain type T of indel observed in the matched reads from the sequencing data of the edited and control collections respectively. Each indel type T is characterized by at least one of:
The segregating may include carrying out the following:
In some embodiments the indel type T is characterized by both said size/length τ of bases and said position i.
In some embodiments the sizes of said respective sub-collections LTx(λm1,T) and LM(λm1,T) are used as the ‘affected’ counts nTx and nMc of reads of amplicons, in which said particular type T of adverse effect is observed.
In some embodiments the template statistical model provided for the INDEL activity of said NA editing procedure comprises a statistical classifier comprising a Maximum A Posteriori (MAP) estimator. In this case the method may include provision of probability reference data indicative of prior probabilities P(edit) and P(no edit) of occurrence of edit and no-edit associated with observance of said indel type T. The prior probabilities P(edit) and P(no edit) may be complementary P(edit)=1−P(no edit).
In some embodiments the prior probabilities P(edit) and P(no edit) are provided as functions depending on a distance between a position i along the site of interest λm1 at which said indel of the type T is observed and to an expected cut-site position i0 of the NA editing procedure at the site of interest λm1. For example in some embodiments the prior probabilities the prior probabilities within a predetermined window of distances between the position i and the expected cut-site position i0 are set with fixed predetermined probabilities (e.g. trivial priors of 0.5), and outside said range decrease in order, according to distance |i−i0| between the position i and the expected cut-site position.
Alternatively, the prior probabilities P(edit) and P(no edit) of said MAP estimator may set as fixed trivial probabilities independent of position P(edit)=P(no edit)=0.5, and said MAP estimator thereby functions as maximum likelihood estimator (MLE).
In some implementations, the MAP estimator of said template statistical model is a Bayesian classifier and wherein said applying of said template statistical model to the respective ‘collective’ counts NTx and NMc of reads and the respective ‘affected’ counts nTx and nMc comprises computing said Bayesian classifier to determining that said NA editing procedure affected an edit causing the indels of type T in case the following is satisfied according to Bayes formula as follows:
P(edit|nTx,nM)>P(no edit|nTx,nM)⇔P(edit)·P(nTx,nM|edit)>P(no edit)·P(nTx,nM|no edit)
wherein P(edit|nTx,nM) and P(no edit|nTx,nM) are the respective probabilities that an edit hypothesis and a no-edit hypothesis are valid given the observed ‘affected’ counts nTx and nMc of indels of type T in the in the edited and control sequences; P(edit) is a prior probability that an observed indel type T was caused by an edit event, and P(no edit) is a complementary prior probability P(edit)=1−P(no edit); P(nTx,nM|no edit) is a probability of observation of the ‘affected’ counts nTx and nMc in the edited and control sequences under an assumption that there was no edit causing the ‘affected’ count nTx observed in the edited sequences; P(nTx,nM|edit) is a probability of observation of the under an assumption that there was an edit causing the ‘affected’ count nTx observed in the edited sequences.
In some embodiments the template statistical model includes utilizes a hyper-geometric distribution for computing the probability P(nTx,nM|no edit) of observation of the ‘affected’ counts nTx and nMc under the no edit assumption.
In some embodiments the template statistical model utilizes a binomial distribution for computing assessing the probability P(nTx,nM|edit) of observation of the ‘affected’ counts nTx and nMc under the edit assumption.
To this end in some implementations the method includes provision of reference probability parameter q of said binomial distribution. The reference probability parameter q is indicative of a probability that an observed indel of type T in the reads of the edit and control collections has occurred through an edit event.
The method may be alternatively or additionally be adapted for determining at least one particular type T of translocation (Category 2 of adverse effects) out of said one or more of types of adverse effect of the NA editing procedure. The at least one particular type T of translocation may be characterized as translocation involving both sites of a respective pair of sites/loci of interest λm1, λm2 where m1≠m2.
In such embodiments the processing of the sequencing data may include matching reads of said sequencing data to the pair of sites/loci of interest, λm1 and λm2, by carrying out the following:
In some implementations the ‘collective’ counts NTx and NMc are estimated based on respective sizes of the following pairs of single site partially matching collections [|CTX(λm1)|, |CTX(λm2)|], [|CMc(λm1)|, |CMc(λm2)|], and said ‘affected’ counts nTx and nMc are determined based on respective sizes of the dual site partially matching collections |CTX(λm1,λm2)| and |CMc(λm1,λm2)| obtained from the reads of the edit and control collections respectively. For example the ‘collective’ counts NTx and NMc may be estimated as respective averages of said respective sizes of the pairs of single site partially matching collections such that: NTx=|CTX(λm1)|, |CTX(λm2)|> and NMc=<|CMc(λm1)|, |CMc(λm2)|>; and the ‘affected’ counts nTx and nMc are determined as the respective sizes of the dual site partially matching collections such that nTx=|CTX(λm1,λm2)| and nMc=|CMc(λm1,λm2)|. In a particular example the ‘collective’ counts NTx and NMc are estimated as respective geometrical averages of said respective sizes of the pairs of single site partially matching collections.
In some embodiments the method includes associating each read r, which presents the particular type of translocation between sites λm1 and λm2, with one of the following four possible translocation species S according to the following translocation matching conditions {DS1 to DS4}:
According to some embodiments the primer molecules, which are used for amplification of a site A, in at least one amplification step, include primer molecules each having respective one of the forward and reverse binding NA sequences and a correspondingly respective one the forward and reverse-adapters, as well as primer molecules each having respective one of the forward and reverse binding NA sequences and correspondingly respective one the reverse and forward adapters. This thereby enables sequencing of reads of all four translocation species: A, B, C and D.
According to some embodiments the processing of the sequencing data includes carrying out said matching reads of said sequencing data to the pair of sites/loci of interest, λm1 and λm2, without segregation to the possible translocation species S∈{DS1 to DS4}, by carrying out the following:
Pfx(ri)F(λm) or Pfx(λm)
R(λm) or Sfx(ri)
Rev(F(λm)) or Sfx(ri)
Rev(R(λm));
In some embodiments the processing of said sequencing data includes carrying out the matching of the reads of said sequencing data to the pair of sites/loci of interest, λm1 and λm2, with segregation to translocation species S={S1 to S4}, by carrying out the following for at least specific translocation species Si∈{S1 to S4}:
According to some embodiments the template statistical model provided for assessing the TRANSLOCATION activity of said NA editing procedure includes a statistical classifier comprising a hyper geometric tail distribution for computing said probability of occurrence of the adverse effect, and wherein assessing a probability of TRANSLOCATION activity of at least one type T or species S of said translocation types or species is determined by computing the probability of said hyper geometric tail distribution based on the respective ‘collective’ counts NTx and NMc type’ counts nTx and nMc of reads of amplicons which are associated with the respective translocation type T or species S.
According to another broad aspect of the present invention there is provided a method for determining translocation adverse effects of Nucleic Acid (NA) editing procedure, the method includes:
The selected statistical distribution model may for example be a hyper geometric tail distribution.
In some implementations the double site partial match condition is elected according to the type or according to the specific species of translocation whose probability of occurrence due to edit is to be assessed.
As indicated above in some implementations the respective ‘collective’ counts NTx and NMc are estimated based on an average of two counts of reads of each of the respective edit and control collections, which satisfy said single site match condition for at least one of the two different sites [λm1,λm2].
According to yet another broad aspect of the present invention there is provided a system comprising a non-transitory computer readable medium storing instructions executable by a processor, for determining and outputting data indicative of the effects of Nucleic Acid (NA) editing procedure according to any of the methods of the present invention described above and in ore details herein below.
In some embodiments the system includes or is adapted to operate the following:
In some implementations the system is configured and operable to determine occurrence of one or more of said types of adverse effect, which are associated with at least one category of the two categories of adverse effects:
The processor may be adapted to process the sequencing data for each particular type of adverse effect of said one or more of said types of adverse effects of said at least one category by carrying out the following:
According to yet another broad aspect there is provided a system for determining and outputting data indicative of the effects of Nucleic Acid (NA) editing procedure. The system may include:
The processor of the system may be is adapted to process said sequencing data for said type of adverse effect by carrying out the above indicated operations a. to d.
In some embodiments the systems described above may be configured and operable for determining said probability of occurrence for adverse effects of one or both of the following categories:
In some implementations the systems described above may include a sequencing utility capable of sequencing of the multiplexed amplifications products/amplicons of the first and second collections of NA sequences. In such implementations the input may be connected/connectable to the sequencing utility for receiving said sequencing data therefrom.
In some implementations of the methods and systems of the invention the sequencing may be conducted utilizing NGS sequencing techniques.
According to further another broad aspect of the present invention there is provided a kit for determining effects of a NA editing procedure. The kit includes:
In some embodiments of the kit, the set of the plurality of primer molecule types {PRt} include matched pairs (PRM+m, PRM−m) of forward PRM+m and reverse PRM−m primer molecule types {(PRM+m, PRM−m)}1M∈{PRt} suitable for amplification of said on-target and off-target sites {λm}1M.
In some embodiments the primer molecules, which are used for amplification of a site λm in at least one amplification step, include primer molecules each having respective one of the forward and reverse binding NA sequences and a correspondingly respective one the forward and reverse-adapters, as well as primer molecules each having respective one of the forward and reverse binding NA sequences and correspondingly respective one the reverse and forward adapters thereby enabling sequencing reads of all four translocation species: A, B, C and D.
According to further yet another broad aspect of the present invention there is provided a kit for determining effects of a NA editing procedure. The kit includes a set of a plurality of primer molecule types {PRt} designed to provide amplification of expected editing sites {λm} of the NA editing procedure, whereby the expected editing sites {λm} include at least one on-target site λ1 and one or more off-target sites {λm}2, where λm represents an off-target or on-target site indexed m; and wherein said set of the plurality of primer molecule types {PRt} comprises pairs (PRM+, PRM−) of forward PRM+ and reverse PRM− primer molecule types (PRM+, PRM−)∈{PRt} suitable for amplification of said on-target and off-target sites {λm} such that each respective forward and revers primer molecule, PRM+ and PRM−, include at least one of a forward and revers adapters. The kit is characterized in that said plurality of primer types {PRt} include:
The kit thereby enables sequencing of all possible translocation species between at least one pair of the editing sites {λm}1M.
In some embodiments the primer molecule types {PRt} include pairs (PRM+m, PRM−m) of forward PRM+m and reverse PRM−m primer molecule types, per each site λm of the one or more sites {λm} that are to be amplified; and wherein the forward binding primer sequence PRS+m of the forward primer PRM+m of the site m includes an NA sequence complementary to the site's λm prefix sequence and the revers binding primer sequence PRS−m of the revers primer PRM−m of the site m includes an NA sequence complementary to the site's λm suffix.
For example, the Kit may be configured for use in a 1st step, PCR1, of a Two-Step-Multiplex-Amplification process. The forwards and reverse adapters are forwards and reverse amplification adapters facilitating that all said translocation species will be amplified in said 2nd step, PCR2, of the Two-Step-Multiplex-Amplification process, to produce amplicons thereof which have forwards and revers sequencing adapters from either side of the amplicon.
In another example, the Kit may be configured for use in One-Step-Multiplex Amplification process. The forwards and reverse adapters may be in this case forwards and reverse sequencing adapters.
Alternatively, or additionally, in some embodiments the Kit may be configured for use in a 2nd step, PCR2, of a Two-Step-Multiplex-Amplification process. The primer molecule types {PRt} of the kit may include pairs (PRM+, PRM−) of forward PRM+ and reverse PRM− primer molecule types comprising respective forward PRS+ and reverse PRS− binding primer sequences complementary to respective forwards and reverse amplification adapters of a 1st step, PCR1, of the Two-Step-Multiplex-Amplification process. The forwards and reverse adapters of the forward PRM+ and reverse PRM− primer molecule types may be in this case forwards and reverse sequencing adapters.
In some implementations the Kit may also include the system according to any embodiment of the present invention as described above and in more details below.
Thus, the technique of the present invention as described above and as will be described in more details below, provides for accurate detection, characterization, and quantification of off-target genome/NA-editing activity, such as indels and translocations, in pre-identified potential off-target and on-target sites, while enabling the use of multiplex amplification (multiplex PCR) followed by sequencing such as NGS, to achieve the same.
The potential off-target sites may be for example be pre-identified by unbiased discovery approaches such as GUIDE-seq8 or by in silico-based strategies that are homology dependent and computationally nominate potential off-target sites based on mismatch and gaps (termed “the editing distance”) between the gRNA spacer and sites in the genome of interest.
The technique of the present invention is advantageous for detecting the off-target activity with low false negative (FN) and false positive (FP) detection rates, and was proved specifically effective for detection of off-target activity at challenging loci associated with low editing rates by the genome/NA-editing procedure. Furthermore, as will be appreciated from the description below, the technique of the present invention enables the detection of alternative cut-position at off-target loci. Additionally, the technique of the present invention facilitates the use of multiplex PCR/amplification and NGS data to detect and quantify, with high sensitivity, adverse structural variations, and translocation events occurring in NA editing procedures.
Some implementations the technique of the present invention use tunable parameters that can be specified to balance between FP and FN. Accordingly the system and method of the present invention may be adjusted to follow a conservative approach, where most of the Tx reads classified as edits, count as edit events, to thereby reduce the FN rate in potential off-target loci.
Another important feature of the present invention is that it allows to infer editing rates without the need to accurately identify the NA editing procedure's (CRISPR) cut-site. This thereby facilitates identification of alternative cut-sites compared to the expected cut-site at any candidate locus. This is achieved by using a wide quantification window (e.g. in the range of 10 bases or even larger than that for each side of the DNA cut-site (e.g. instead of a narrow quantification window such as of sizes of 1-5 bases for each side of the DNA cut-site) centered at the predicted cut-site, which are may be used to avoid measuring errors resulting from PCR and sequencing). The technique of the present invention overcomes the need to use a narrow quantification window, by incorporating different prior probabilities in a larger window for each position in the reference sequence, thereby providing a more robust technique of edit events quantification.
In order to better understand the subject matter that is disclosed herein and to exemplify how it may be carried out in practice, embodiments will now be described, by way of non-limiting example only, with reference to the accompanying drawings, in which:
Reference is made together to
The optional operations 10 to 40 described herein the following are preliminary operations which are not necessarily included or implemented by the method/system 100/1000 of the present invention.
The optional preliminary operations 10 and 20 include provision of collections a first collection Tx and second collection Mc of NA sequences originated from the same NA source SRC, whereby the first collection Tx is an edited collection of NA sequences from the NA source SRC, to which a certain NA editing procedure 20 is applied, and the second collection Mc is a control (mock) collection of NA sequences of the NA source, to the NA editing procedure 20 is not applied. Preliminary operation 10 designates the extraction of NA sequences of the collections Tx and Mc from the NA source SRC, and operation 20 designates the NA editing procedure applied to the NA sequences of the collection Tx. The
The optional preliminary operation 30 includes applying amplifications to the edited Tx and control Mc collections respectively, to obtain respective amplified products/amplicons ATx and AMc corresponding to said edited Tx and control Mc collections respectively.
Preferably, the amplifications of operation 30 are conducted by applying multiplexed amplifications 30 with a plurality of primer types {PRt} selected to simultaneously amplify a plurality of desired sites of the NA sequences of the edited and control collections Tx and Mc. As will be appreciated by those versed in the art, provided that a technique for determining effects of Nucleic Acid (NA) editing procedure 20 is capable of handling the sequencing results of multiplexed amplification (which is the case for the technique of the present invention), the multiplexed amplification (e.g. such as multiplex PCR) is advantageous for this purpose over a singleplex amplification (e.g. as singleplex PCR) since multiplexed amplification provides for simultaneous amplification of multiple different NA sites of interest, thus enabling to efficiently amplify simultaneously different on target and of target sites of the NA editing in relatively short time.
In this regard, as would be readily appreciated by those versed in the art, after knowing the present invention, that for at least the purpose of detection of INDEL adverse effects of NA editing the technique of the present invention may also be carried out based on the singleplexed amplification. However in such embodiments employment of multiple singleplexed might be required, each per each NA site at which INDEL adverse effects of NA editing might be expected. In other words for observing INDELS in each of the M number of expected on-target and off-target sites of the NA editing procedure, in the order of M singleplexed amplification processes will be resequenced, thus yielding long and cumbersome amplification process (which may be dismissed/replaced by a single multiplex amplification).
Another advantage of the use of the multiplexed amplification over a singleplex amplification, as contemplated by the inventors of the present invention, is that such multiplexed amplification facilitates the amplification of NA sequences presenting TRANSLOCATIONs between the on-target and/or the off-target sites of the NA editing.
In this regard, as would be readily appreciated by those versed in the art, after knowing the present invention, use of singleplexed amplifications for at least the purpose of detection of TRANSLOCATION adverse effects of NA editing, is generally not practical. This is because each type of TRANSLOCATION adverse effects involves a combination at least two NA sites (e.g. two sides of the expected on-target and/or the off-target sites of the NA editing procedure). The number of such possible combination is thus in the order of M2 where M presents the number of the expected on-target and off-target sites of activity of the NA editing procedure. This (M2) is typically a too large number of singleplex multiplications for being practically conducted. On the other hand, the inventors of the present invention have understood that using a multiplexed amplification of the on-target and off-target sites, will inherently amplify also at least some of the translocations between said sites. This makes the use of multiplexed amplification particularly advantageous for observation/detection of TRANSLOCATION adverse effects.
For clarity, in the following description the term multiplexed-amplification is used to describe the type of NA amplification used by the technique of the present invention. However, in view of the above explanations, it should be appreciated that the present invention is not necessarily limited to multiplex amplification, and instead a singleplex amplification may be conducted without departing from the scope of the present invention, particularly when INDEL activity of the NA editing procedure is to be assessed in only one or few sites.
Preferably, for improve accuracy and reliability of the technique of the present invention, the multiplexed amplifications 30 of the edited and control collections Tx and Mc are conducted with similar primer types. The multiplexed amplifications 30 may be conducted utilizing any known or future multiplexed NA amplification techniques, some known none-limiting examples therefore are multiplexed NA amplification techniques that are based on multiplex Polymerase Chain Reaction (PCR). It would be appreciated however that the technique of the present invention is not limited to this specific type of NA amplification and other amplification or enrichment techniques might be used. It might also be appreciated that when the NA Editing effects on a single stranded NA sequences, such as RNA, are to be examined by the technique of the present invention, (i.e. in embodiments where the edited and control collections Tx and Mc are collections of single stranded NAs), a preliminary step of the NA amplification ay include transcription/conversion of the single stranded NAs of the edited and control collections Tx and Mc to corresponding double stranded NAs (such as DNAs), and this step may then be followed by double stranded amplification, such as multiplexed PCR.
In this regards, as generally known in the art, an NA editing technique which is directed to edit a certain one or more on-target site in the NA sequences, e.g. indicated here λ1, may potentially affect/edit one or more potential off-target sites, e.g. indicated here {λm}2M in the NA sequences. This may be for example due to possible complemental similarity of base sequences of the off-target sites to the base sequence of the guide RNA used in the NA editing procedure, which may lead the guide RNA to bind to the off-target sites. Preliminary target information TD about expected editing sites {λm}1M of the NA editing procedure (M designating a number of the expected on-target and off-target sites indexed m) may for example be assessed based, for instance, on the unbiased genome-wide approaches or by other technique as for example mentioned above.
Accordingly, for the purpose of the present invention, the applying multiplexed amplifications 30, that are applied to the edited Tx and control Mc collections respectively, are preferably designed such some, or more preferably all, of the expected editing sites {λm}1M of the NA editing procedure, of both the edited Tx and control Mc collections, would be amplified thereby. Accordingly, the amplified products/amplicons ATx and AMc corresponding to the edited Tx and control Mc collections respectively, include amplicons of some, and preferably all, of the expected editing sites {λm}1M of the edited Tx and control Mc collections respectively.
For example, this may be achieved with present NA amplification techniques, such as multiplexed PCR, by conducting the NA amplifications operations 30 using a selected set of a plurality of primer molecule types {PRt} that are suitable for amplification of said on-target and off-target sites {λm}1M in the edited and control NA collections. The primer molecules of the selected set {PRt} may be selected according to the target information TD about the expected editing sites {λm}1M of the NA editing procedure. The primer molecules of the selected set {PRt} may for instance include, or be constitutes of, matched pairs (PRM+m, PRM−m) of ∈{PRt} that respectively match to certain prefix-primer-sequence PRS+m and suffix-primer-sequence PRS−m primer molecule types {(PRM+m, PRM−m)}1M of each or at least some of the expected editing sites {λm}1M.
The optional preliminary operation 40 includes sequencing of the multiplexed amplifications products/amplicons ATx and AMc of the edited and control collections, Tx and Mc, by which sequencing data, ESD and MSD, including respective pluralities of reads, RTx={riTx} and RMc={rjMc}, of the multiplexed amplifications' products/amplicons ATx and AMc of each of the edited and control collections, is obtained. In this regard it should be understood that the sequencing operation 40 is not limited to any particular sequencing technique or sequencer system 400 and may be conducted with any suitable existing or future NA sequencing technique sequencer system 400, as known or will be known in the art.
As indicated above, the operations 10 to 40, are optional preliminary operations which may be performed in the scope of the method 100 of the present invention, or prior thereto. According to the method 100 of the present invention, the sequencing data ESD and MSD including the respective pluralities of reads, RTx={riTx} and RMc={rjMc}, of the amplicons ATx and AMc is received and processed for constructing, per each particular type of adverse effect of one or more types of possible adverse effects of said NA editing procedure (not necessarily for all possible adverse effects), a statistical model of occurrence of that type of adverse effect by the NA editing procedure. The statistical model generally employs a classifier to determine whether an adverse effect of that type occurs due-to/by the NA editing procedure, or otherwise the that type of adverts effect is observed due to errors/artifacts (generally referred to herein as NOISE), caused by either the Amplification or the Sequencing preliminary operations 30 and/or 40.
Accordingly, in the technique of the present invention the classifier of the statistical model, statistically determines a likelihood that the observed type of adverse effect is a type of adverse effect occurring due to the NA editing procedure based on a comparison between numbers of observed amplicons with NA sequences corresponding to that type of adverse effect of in each of the ATx and AMc amplicons of the edited and control/mock collections respectively.
As a result of this process, accurate estimation of the types of adverse effects obtained by the examined NA editing procedure is obtained. Data indicative of the assessment/quantification of the occurrence of the type(s) of adverse effect by the NA editing procedure, of one or more categories (translocations or indels), is then output, to enable determination of suitability safety of application/use of NA editing procedure (e.g. on the NA sequences of the source SRC). Alternatively or additionally, data about the efficiency/specificity of the NA editing procedure may also be determined and output, based on the identified types of adverse effects.
Further details and examples of the technique, system 1000, and method 100 according to an embodiment of the present invention, will now be described with reference to
The system 1000 may be used for determining and outputting data indicative of the effects of Nucleic Acid (NA) editing procedure, The system 1000 includes:
More specifically the input 1010 is adapted to receive sequencing data indicative of respective the read results ESD and MSD of the edit and control amplicons ATx and AMc. This includes pluralities of reads, RTx={riTx} and RMc={rjMc} [CAN YOU POVIDE THE ORDER OF NUMBER OREADS] obtained by sequencing of multiplexed amplifications products/amplicons ATx and AMc of edit and control/mock collections of NA sequences.
The memory 1020 or a section M1 thereof, stores the respective pluralities of reads, RTx={riTx} and RMc={rjMc}, of the sequencing data ESD and MSD. The memory 1020 or a section M2 thereof, also stores the reference data REF which is indicative of at least one reference NA sequence of at least one respective site λm of the expected editing sites {λm}1M of the NA editing procedure. The memory 1020 or a section M2 thereof, may also store sequences data PRS indicative of the primer sequences (e.g. prefix and suffix) of on-target and/or off-target sites. The primer sequences data PRS may be for example indicative of the set of the plurality of primer molecule types {PRt} that are used in the multiplexed amplifications 30 of the expected editing sites {λm}1M of the NA editing procedure 20.
It should be noted that according to some embodiments of the present invention system 1000 may optionally also include (or be connected to (e.g. directly)) of the sequencing utility 400 which is capable of sequencing of the multiplexed amplifications products/amplicons of the first and second collections of NA sequences. In such embodiments the input 1010 may be connectable to the sequencing utility 400 for receiving the sequencing data therefrom.
It should be noted that according to some embodiments of the present invention system 1000 may optionally also include a preprocessor 1160 adapted to apply one or more preprocessing operations to the received sequencing data ESD and MSD (although not specifically depicted in the figures, these preprocessing operations may be part of the operations of the method 100). The preprocessing may for instance include adjusting the reads, RTx={riTx} and RMc={rjMc} of the sequencing data ESD and MSD by carrying out at least one of the following:
The memory 1020, or a section thereof M3, may also store at least one template statistical model TM corresponding to at least one category of adverse effects, indels and/or translocations. The template statistical model TM may for example include a set of match conditions and a classifier (e.g. matrix). The match conditions are designated to be applied for comparison between the sequencing data ESD and MSD and a part of the reference data REF that corresponds to the type T of effect looked for, to determine properties of the sequencing data (e.g. the collective and affected counts) indicative of occurrence of this effect T. The classifier is designated to process the properties of the sequencing data determined by the match conditions, to determine/assess occurrence of that adverse effect T due to the NA editing procedure 20.
It should be noted that the list EFT of one or more types {T} or classes of adverse effects which are to be examined by the system 1000 may be received as input, or may be internally identified, e.g. based on the category or class of adverse effects to be processed by the system.
In the latter case the system may include an ETF identifier module 1170 capable of processing the reference data (e.g. the sequences of the on and off targets sites, and possibly also the sequencing data to identify a list of the possible adverse effects to be processed by the system. For instance, in embodiments directed to translocation detection, identifier module 1170 may include in the list EFT of types/classes {T} of adverse effects all the pair combinations of different on and off targets sites [λm1,λm2]. Alternatively or additionally, in embodiments directed to indel detection, per each class/site [λm] in which types of indels are to be detected, the identifier module 1170 may process the sequencing data ESD and MSD after being aligned to the site λm of the class [λm] (see alignment discussion below), to identify gaps in the aligned sequences and process/identify the types T of the indels to be included in the list based on the relevant properties of the gaps (e.g. their i position within the respective site λm; their sizes for example in terms of number of bases—these may be positive of insertion-indels and negative for deletion-indels; or the sequence of bases introduced therein, in case of insertions indels). Accordingly the identifier module 1170 may list EFT the types/classes {T} adverse effects that should be processed by the system 1000.
In some implementations the system also includes a looper/threader module which is configured and operable to process the EFT the types/classes {T} adverse effects that should be processed by the system 1000 and operate the processor, for instance sequentially in a loop, and/or in parallel (e.g. by multithreaded operation), to process the sequencing data ESD and MSD per each type T of adverse effect in the list EFT to determine the likelihood of occurrence for each said type T.
In turn, type T of adverse effect the processor 1100 is configured and operable for processing the sequencing data ESD and MSD according to the method 100 of the present invention, based on the reference data REF and the template statistical model TM, for applying the template statistical model TM suitable to the type T, to sequencing data ESD and MSD, and to thereby assess/determine a the occurrence of at least one type T of adverse effect by the NA editing procedure.
As indicated above the systems and methods according to various embodiments of the present invention may be configured and/or operable for determining the occurrence adverse effects of one or both of the following categories, by the NA edit procedure:
With reference to
Note that in embodiments where only translocations are of interest, only the operation b. is performed and in embodiments where only indels are of interest, only the operation a. is performed.
At the end of the operations a. or b. each categorized read is classified according to its class. Namely for indels each categorized read ri are assigned with a corresponding class: ri->[λm1]=[λm1=λm2]|m1=m2; and for translocations, each relevant read ri, is assigned with a corresponding class: ri->[λm1,λm2]|m1≠m2. Reads not matching to the one or both categories of interest may optionally be ignored in the further processing (e.g considered as none-relevant to the category of interest or as fragments/artifacts.
In the following description, examples of template (statistical) models for determining occurrence of various respective indel types or translocation types/classes will be exemplified and described in more detail. It should be understood that the below models are provided as none limiting examples and as will be appreciated by those versed in the art after knowing the invention, other suitable models may be devised without departing from the scope of the present invention as defined in the claims. Thus, in the following at least one of modeling methods I. and II. for indel and translocations categories respectively may be performed:
I. Statistical Model for Determining INDEL Activity of the NA Edit Procedure
Reference is made to
i. Matching Reads to Site
For indels the combinations of the preprocessing operations A. and B.a. above may be more specifically described as follows for a given site/locus of interest λm1 of which indels are to be assessed:
This results with obtaining, for the site of interest, λm1, respective collections LTx(λm1) and LM(λm1) of matched reads, obtained from the sequencing data ESD and MSD of the edited and control collections respectively.
In this connection it is noted that the ‘collective’ count match condition of the template statistical model of the INDEL Category 1 presents the combination of the matching conditions described above with reference to the preprocessing operations A. and B.a. This collective’ count match condition is satisfied for a read ri in case the prefix and suffix regions of the read ri match prefix PRS+m and suffix PRS−m primer sequences (PRS+m1, PRS−m1) of a respective site of interest λm1.
As will be described in more details below, the sizes of respective collections LTx(λm1) and LM(λm1) present the ‘collective’ counts NTx(λm1) and NMc(λm1) of reads of amplicons which are associated with Indel at least one class of adverse effects involving the site λm1 as observed in the sequencing data of the corresponding edited and control collections ESD and MSD.
ii. Aligning the Matched Reads to their Respectively Matched Site to Identify Indel Types {T}
After obtaining the respective collections LTx(λm1) and LM(λm1) of reads for each site of interest λm1 (typically conducted for all of the on and off target sites), the lists/collections LTx(λm) and LM(λm1) of each site of interest λm1 may be processed to identify indel types {T} appearing therein. This includes carrying out the following:
At the end of this operation, for each locus λm1 of interest, a list of aligned reads coming from the mock collection Mc, denoted LMc and a list of aligned reads coming from the edited collection Tx, denoted LTx, are obtained. The comparison of these two lists may be used basis for quantifying the indel activity detected at the loci λm1 of interest.
iii. Segregating Observed Indels According to Indel Types
Then, for each locus site λm1, Tx and M read lists, LTx(λm) and LM(λm1), are converted into several more focused lists pertaining to specific indel types T and positions, LTI(λm1,T) and LM(λm1,T). For a given indel type T (and a given locus λm1) the information may for example be summarized in the form of a table where each column represents a position i on the reference locus λm1.
Thus, in order to typify the observed indels, further processing of the sequencing data of the indel reads LTI(λm) and LM(λm1) may include segregating the collections LTx(λm1) and LM(λm1) of reads matching the site of interest λm1, to form respective two sub-collections LTx(λm1,T) and LM(λm1,T) of reads per each indel type T of interest which is observed in the reads (typically all types are of interest). It should be noted that the reads may be typified according one or more of the following characteristics:
The segregation/conversion to the sub-collections LTI(λm1,T) and LM(λm1,T) of reads may for example include identified gaps in the aligned matched reads with properties of a gap representing each certain type T of indel based on an ‘affected’ count match condition of the template statistical model of Indels. The ‘affected’ count match condition of the template statistical model of Indels may be a condition satisfied upon fulfillment of a predetermined set one or more of the following conditions:
To this end, in some embodiments all the indel reads LTx(λm) and LM(λm1) may be assign into indel types {T} (for instance per position and/or indel gap size) to obtain lists of reads LM(τ,i) and LTx(τ,i), where τ denotes the indel type, e.g. insertion of length 3, and i denotes the index of the position at the given reference sequence. Note that reads that are perfectly aligned to a reference site of interest λm1, are not members of any of the above lists of reads LM(τ,i) and LTx(τ,i) (for off-target sites these are typically a majority as only small editing effects are expected therein). Also note that reads may represent more than one type T of indel implicated several gaps identified by the alignment. The sizes of respective sub-collections/lists LTx(λm1,T) and LM(λm1,T) present the ‘affected’ counts nTx(λm1,T) and nMc(λm1,T) of reads of amplicons, in which said particular type T of adverse effect is observed at site λm1.
It should be noted that in some embodiments the indel type T is characterized by only one of the size/length τ of bases of the gap or the position i of the gap in the respective site's sequence. Alternatively, in some embodiments the type T is characterized by the combination of both the size/length τ of bases and the position i.
The following description exemplifies the application of a statistical model to the ‘affected’ counts nTx(λm1,T) and nMc(λm1,T) and ‘collective’ counts NTx(λm1) and NMc(λm1) indicated above to determine whether the NA edit procedure causes an adverse effect of indel of type T at the site λm1. This may be performed for each of a plurality of indel types T of each of a plurality of sites λm1. For clarity, in the following description which exemplifies the statistical modeling for a certain indel of type T at a certain site λm1, the indication of the parameters (λm1,T) are omitted from the ‘affected’ and ‘collective’ counts nTx,nMc, NTx and NMc.
iv. Applying a Statistical Classification Model to Determine Whether of Reads Associated with an Indel Type T Originate from and Actual NHEJ/Indel Editing Activity of the NA Editing Procedure
The technique of the present invention provides for distinguishing indel reads resulting from molecules edited by the NA edit procedure (edited reads) from indel reads derived from various sources of noise (e.g. from the amplification or sequencing). This distinction is based on the ability to assess whether a specific indel originates from edit event of the NA edit procedure or is a result of experimental background noise (such as sequencing artifacts or erroneous read assignments). The model used according to an embodiment of the present invention for identifying/distinguishing edited reads from noise, is based on the comparison of indel statistics in the amplicons ATx of the edited collection vs the amplicons of the control/mock collection AMc of the edited collection (e.g. utilizing comparative Probability/statistics as exemplified in
For each indel type, T, the system 1000 or method 100 according to an embodiment of the present invention, applies a statistical inference classifier to classify whether indel events of type T in site λm1 in the reads of amplicons ATx of the edited collection, are a result of an editing event introducing the indel of type T to the site λm1 or are noise (such as amplification/PCR noise or sequencing noise which also occur in the mock sample). As indicated above, the indels types T are identified with comparison to the reference NA sequence of the respective site λm1 (e.g. via the alignment operation). The type T indel events observed within the reads of the edit collection Tx, are classified as originating from an edit event or from background noise. All Tx reads with type T indels in, which are positively classified (i.e. as edit event), are considered as edited reads. This process may be repeated for all or some of a plurality of indel types T (e.g. for each pair of the identified gap's size and position (τ,i)). Eventually, all reads that are classified as positive in at least one indel type T are marked as edited reads.
Optionally, in order to improve the efficiency of the process, only indel types T whose position i along the site of interest λm1 is within a predetermined a window around an expected cut-site/position i0 of the NA editing procedure at the site of interest λm1 are processed (i.e. for classification). For example the window size may be in the order of several tens of base-pairs (e.g. 20 bp) and indel types observed in other positions outside that window may be discarded. This is because (one may expect that an indel originating from an edit event will be a result of a double strand break at/near the cut-site/position i0 of the NA editing procedure.
Accordingly, for each indel type T of interest, a classifier (is used to determine whether reads that represent this indel T, originate from an editing event, or from background noise. This is based on the above indicated the ‘affected’ and ‘collective’ counts nTx,nMc NTx and NMc associated with the site λm1 and the type T of the indel.
In some embodiments the template statistical model provided according to the present invention for the assessment of INDEL activity of the NA editing procedure includes a statistical maximum likelihood classifier may be used for this purpose as the statistical inference classifier.
Alternatively, or additionally in some embodiment's prior probabilities P(edit) and P(no edit) may be obtained or estimated, indicative of the probabilities of occurrence or none-occurrence of an edit event causing the respective indel type T. In such embodiments, the template statistical model provided according to the present invention for the assessment of INDEL activity of the NA editing procedure may include a more accurate statistical classifier including a Maximum A Posteriori estimator (MAP), such as a Bayesian classifier may be used to obtain a more accurate statistical classification.
In this regard it should be noted that prior probabilities (priors) P(edit) and P(no edit) are complementary Prior Probabilities indicative of the probability of edit or no-edit occurrence of the indel type T where here complementary P(edit)=1−P(no edit). The prior probabilities P(edit) and P(no edit) may be part of the model and may be provided as functions or lookup tables, which depend on a distance between a position i along the site of interest λm1 at which the indel of the type T is observed and the expected cut-site position i0 of the NA editing procedure at the site of interest λm1.
The priors may be defined based on the basic principles of the NA editing procedure (e.g. based on the basic principles of CRISPR activity) or based on experimental data of the NA editing procedure. In some embodiments these priors are configurable. In some embodiments these Prior Probabilities are adjusted depending on the base-distance between the indel's position i and the cut site of the DNA editing procedure.
For example, the prior probability P(edit) may be set to a predetermined value (e.g. about 0.5) within a predetermined range for example of −6≤i≤6 when the index i corresponds to a position relative to the expected cut-site position i0. The priors can then be P(edit,i)=exp(−α|i−i0|) for any positive a. The priors will then be decreasing in order, according to distance |i−i0| between the position i and the expected cut-site position. In some embodiments these priors can be 0.5, 0.1, 0.01, 0.001, 0.0001. Note, that selection/configuration pf the system and/or method of the present invention with non-zero Prior Probabilities away from the cut-site i0 allows for the detection of alternative cut-sites in the classification window and are fully user-configurable.
Indeed, it should be understood that the maximum likelihood classifier is a private case of the Maximum A Posteriori estimator where the prior probabilities P(edit) and P(no edit) are not know or estimated. E.g. by trivially setting P(edit) and P(no edit) to the trivial values P(edit)=0.5 and P(no edit)=0.5 in the Maximum A Posteriori estimator, the statistical maximum likelihood estimator (MLE) classifier is obtained. To this end in some embodiments the prior probabilities P(edit) and P(no edit) of said MAP estimator may be set as fixed trivial probabilities independent of position P(edit)=P(no edit)=0.5, and the MAP estimator thereby functions as maximum likelihood estimator (MLE).
In the following description the application of the template statistical model for indels is described based on a specific non-limiting example in which Bayesian classifier is used as MAP estimator of the template statistical model. In this case the application of the template statistical model to the respective ‘collective’ counts NTx and NMc of reads and the respective ‘affected’ counts nTx and nMc includes computing the Bayesian classifier to determining whether the NA editing procedure has caused edit events resulting with the indels of the type T.
For example, for a given indel type T˜(τ,i) (gap size τ and a indel position i) nTx=|LTx(τ,i)| and nM=|LM(τ,i)|. Based on these observed numbers, the model determines whether the observation is more likely to have originated from an edit event—P(edit|nTx,nM), or to represent background noise—P(no edit|nTx,nM). MAP classifies an indel of type T as originating from an edit event when the posterior P(edit|nTx,nM) is higher.
To this end the MAP estimator probabilities may be compared using Bayes rule as follows:
P(edit|nTx,nM)>P(no edit|nTx,nM)⇔P(edit)·P(nTx,nM|edit)>P(no edit)·P(nTx,nM|no edit) (1)
wherein P(edit|nTx,nM) and P(no edit|nTx,nM) are the respective probabilities that an edit hypothesis and a no-edit hypothesis are valid given the observed ‘affected’ counts nTx and nMc of indels of type T in the in the reads of the amplicons ATx and AMc of the edit and control sequences; P(edit) and P(no edit) are the above indicated priors; P(nTx,nM|no edit) is a probability of observation of the ‘affected’ counts nTx and nMc in the edited and control sequences under an assumption that there was no edit causing the ‘affected’ count nTx observed in the edited sequences and P(nTx,nM|edit) is a probability of observation of the ‘affected’ counts nTx and nMc under an assumption that there was an edit causing the ‘affected’ count nTx observed in the edited sequences.
In some embodiment as exemplified herein the template statistical model includes utilizing hyper-geometric distribution for computing the probability P(nTx,nM|no edit) of observation of the ‘Affected’ counts nTx and nMc under the no edit assumption. In this regards, the hypergeometric distribution is a discrete probability distribution that describes the probability of b=nTx successes (random draws for which the object drawn has a specified indel feature) in n=NTx draws, without replacement, from a finite population of size N=NTx+NM (size of both the mock and the edit sequenced populations) that contains exactly B=nTx+nM objects with that indel feature, wherein each draw is either a success or a failure. This model means that all indels are equally likely to occur in a M read as in a Tx read. For the reasons explained below, using the hypergeometric distribution provides accurate statistical modeling for the probability of a scenario in which nTx+nM indel events (of type T) are observed in the edit sample and the mock sample in case the events nTx in the edit sample are not a result of DNA editing.
It should be however understood that for some implementations using different distributions for modeling the probability of P(nTx,nM|no edit), may also be possible, for instance using the Binomial Distribution (e.g. although using such other distribution may generally be less suited for this scenario and may yield less accurate results in various circumstances).
To this end, given collective counts NTx and NM, indicated above, without loss of generality (due to the symmetry of the hypergeometric distribution), the hypergeometric distribution HG(b; N, B, n) for P(nTx,nMc|edit) may be defined as:
Where N=NTx+NMc, b=nTx, B=nTx+nMc, n=NTx
In some embodiment as exemplified herein the template statistical model includes utilizing the binomial distribution for computing assessing the probability P(nTx,nM|edit) of observation of the ‘affected’ counts nTx and nMc under the edit assumption. P(nTx,nM|edit) represents the probability of seeing the observed number nTx of indels in the reads of the edit amplicons ATx, out of the total number of the observed indel events of type T, nTx+nMc in the reads of both the edit and mock amplicons ATx and AMc. In this case the template statistical model a reference probability parameter q for use in the binomial distribution. The reference probability parameter q is indicative of a probability that an observed indel of type T in the reads of the edit and control collections has occurred through an edit event. As noted below this reference probability parameter q is typically set as a number close to 1 (for Edit and Mock collections of similar sizes of the same order). Then the random variable nTx,nM|edit is modeled with the Binomial distribution, which describes the probability of nTx successes out of n=nTx+nM draws with replacement. Accordingly, without loss of generality, P(nTx,nM|edit) may be modeled as P(nTx,nM|edit)˜Binom(nTx;n,q), where n=nTx+nM, to obtain:
It should be noted that the choice of reference/model parameter q to be close to unity (e.g. q for example may be chosen to be within the range of [0.92 to 0.98] or a larger range for similar number of reads of the edit and control amplicons ATx and AMc—for instance q=0.95) is based on the assumption that most of the observed indels of type T in the reads of the edit amplicons ATx amplicons are caused by an edit event of the NA edit procedure, and only a small portion in the reads of the edit and control amplicons ATx and AMc is due to background noise. Practically, the parameter q may be inferred from the experimental data and provided as reference for the system and/or method of the present invention. In this regard, it is noted that q may be a configuration parameters where higher number will increase the chances of getting false positive indication that an indels of certain types occurred due to edit event, and lower number with increase the chances of false negative indication that indels of certain types are observed due to background noise.
v. Quantifying Indel Editing Activity of the NA Editing Procedure and Optionally Determine Confidence Interval for the Same
Optionally the indel editing activity is quantified as the frequency of the edited reads out of the total number of reads in Tx:
Note that this is a conservative approach as we count all reads in types classified as edits to actually represent edit events.
A confidence interval may optionally also be calculated based on the above quantification, for each potential target site using the statistical approach:
Where {circumflex over (p)} denote the inferred editing frequency and NTx denote the total number of reads in Tx. a is the desired confidence level (which is 0.05 in our demonstration data) and c is the CDF of the standard normal distribution.
II. Statistical Model for Determining TRANSLOCATION Activity of the NA Edit Procedure
Reference is made to
As noted above, an important advantage of the technique of the present invention is that it facilitates the detection of translocation events with fusions at on-target and off-target sites.
Since the multiplexer amplification, such as multiplex-PCR reaction, contains all primer pairs for the sites {λm} of interest, it is possible that amplicons will be formed based on fusion NA molecules, as the primers on both sides will be present. Accordingly, utilizing/providing reference data REF indicative of the reference NA sequences corresponding to the sites/loci of interest, λm1 and λm the translocation activity of the NA edit procedure between these sites can be determined.
For example, determining/assessing a particular type T or species S of translocation activity of the NA edit procedure between two loci λm1 and λm2 where m1≠m2, according to the method 100.3 may be performed as follows:
i. Identifying Reads that are Putatively Originating from Translocation λm1-λm2.
This includes processing of the sequencing data ESD and MSD to determine a match between the reads thereof and the pair of sites/loci of interest, This is achieved by obtaining an ‘affected’ count matching condition of the translocations' template model. More specifically, for each read ri of a plurality of the reads riTx∈RTx rjMc∈RMc of the respective amplicons ATx and AMc, determine whether said read ri satisfies the ‘affected’ count matching condition. The ‘affected’ count matching condition is indicative of whether a read is at least partially matching to both the reference NA sequence λm1 or λm2, of the pair of reference NA sequences associated with the pair of sites/loci of interest λm1 and λm2. The ‘affected’ count matching condition of the template model may include a combination of one or more of the following four possible translocation matching conditions DS1 to DS4, whereby each of those translocation matching conditions DS1 to DS4, is associated with a different one of four possible translocation species S:
More specifically, the translocations species A, B are single-centromeric formed respectively by fusion of (A) the left-part L of the site λm2 in chromosome CH10 and right-part R of the site λm1 in chromosome CH3 (i.e. 10-L⊕3-R), and vice versa (B) the left-part of the site λm1 in chromosome CH3 and right-part of the site λm2 in chromosome CH10 i.e. 3-L⊕10-R). The single-centromeric translocations species A, B can be identified by the respective translocation matching conditions DS1 and DS2 indicated above. More specifically, the translocations species C, D are centromere-free (C) and double-centromeric (D) formed respectively by fusion of (C) the left-part L of the site λm2 in chromosome CH10 and left-part L of the site λm1 in chromosome CH3 (i.e. 10-L⊕3-L), and (D) the right-part of the site λm1 in chromosome CH3 and right-part of the site λm2 in chromosome CH10 (i.e. 3-R⊕10-R). The centromere-free C and the double-centromeric D translocations species may be identified by the respective translocation matching conditions DS3 and DS4 indicated above (provided that a pairs of primers with suitable sequencing adapters, as described in the present invention are used in the amplification, see e.g.
To this end the ‘affected’ count matching condition of the template model may include any one DSj of the above translocation matching conditions DS1 to DS4, in case a selected species j of translocation is to be separately identified/assessed; or it may be composed as a combined condition combined from two or more of the above translocation matching conditions DS1 to DS4 in alternative form. In the latter case for example in case all species of translocations are to be assessed/determined without distinction, the combined condition DS will be satisfied by a read in case any of the translocation matching conditions DS1 or DS2 or DS3 or DS4 is satisfied.
The ‘affected’ count matching condition is used to identify dual site partially matching collections CTX(λm1,λm2), CMc(λm1,λm2) of reads of the amplicons ATc and AMc of the edit and mock collections respectively. These reads, which are found to match one or more of the translocation matching conditions for the pair of sites (λm1,λm2) are asserted as reads representing the putative amplicon that represents the λ1 to λ2 fusion at the cut site.
Optionally the dual site partially matching collections CTX(λm1,λm2), CMc(λm1,λm2) of reads are further filtered by performing an alignment of the reads to a putative reference amplicon sequence that represents the λ1 to λ2 fusion at their respective cut sites (as determined by the PAM) and retaining only those reads which have sufficiently strong alignment scores, e.g. above a certain alignment threshold.
The reads of the collections CTX(λm1,λm2), CMc(λm1,λm2) which are found to match one or more of the translocation matching conditions for the pair sites (λm1,λm2) and which are possibly also aligned with sufficiently high alignment score to the putative reference amplicon sequence representing the λ1-λ2 translocation, are considered as reads attesting to fusion of the pair of sites (λm1,λm2).
Accordingly the counts nTx and nMc of the number of reads in these respective collections CTX(λm1,λm2), CMc(λm1,λm2) that represent the respective ‘affected’ counts nTx and nMc of reads, in which a translocation involving fusion of both site [λm1,λm2], are observed. In other words, for translocations, the ‘Affected’ counts nTx and nMc are determined as the respective sizes of the dual site partially matching collections such that nTx=|CTX(λm1,λm2)| and nMc=|CMc(λm1,λm2)|.
ii. Identifying Reads that Represent Amplicons Originating from Either of the Sites λm1, λm2.
This includes processing the sequencing data ESD and MSD to determine the ‘collective’ read counts CTX(λm1) and CM(λm2). CTX(λm1) counts the reads that have either end matching a primer from the primer pair of [λm1,λm2]. To this end optionally the single site partially matching condition with a site λm will be considered as satisfied for each read that either end thereof (prefix or suffix) matches the prefix or suffix sequence of the respective site. CTX(λm2) counts the reads that have either end matching a primer from the primer pair of λm2. We thereby get the four counts CTX(λm1), CTX(λm2), CMc(λm1), CMc(λm2) of reads of the edit and mock amplicons ATx and AMc.
The ‘collective’ counts can be separately counted for the four different translocation species. These are preferably obtained from the above numbers, CTX(λm1), CTX(λm2), CMc(λm1), CMc(λm2), by dividing them by the total number or possible species, 4, and possibly multiplying by the number of species actually considered.
The sizes of the single site partially matching collections {CTX(λm1), CTX(λm2)}, {CMc(λm1),CMc(λm2)} provided the total number of relevant reads, in both Tx and Mc collections, and are indicative of the respective ‘collective’ counts NTx and NMc of reads of amplicons.
For example, the ‘collective’ counts NTx and NMc are estimated based on respective sizes of the following pairs of single site partially matching collections [|CTX(λm1)|, |CTX(λm2)|], [|CMc(λm1)|, |CMc(λm2)|]. For example the ‘collective’ counts NTX and NMc of for translocations may be estimated as respective averages of the respective sizes of the pairs of single site partially matching collections of each of the edited and mock collections as follows: NTx=<|CTX(λm1)|, |CTX(λm2)|> and NMc=<|CMc(λm1)|, |CMc(λm2)|> (where < > indicates average). In a particular example the ‘collective’ counts NTx and NMc are estimated as respective geometrical averages of the respective sizes of the pairs of single site partially matching collections.
iii. Applying Aa Statistical Classification Model to Determine Whether Reads Associated with the Translocation Type T Originate from an Actual Translocation Editing Activity of Translocation Type T or Species S
As indicated above, the technique of the present invention facilitates distinguishing reads resulting from edit events of the NA edit procedure from translocation reads derived from various sources of noise (e.g. from the amplification or sequencing). This distinction is based on the ability to assess whether a specific translocation type/species originates from edit event of the NA edit procedure or is a result of experimental background noise (such as sequencing artifacts or erroneous read assignments). The model used according to an embodiment of the present invention for this distinction, is based on Probability Distribution suitable for the statistics of translocation in the amplicons ATx of the edited collection vs the amplicons of the control/mock collection AMc of the edited collection. A person of ordinary skill in the art would readily appreciate after knowing the present invention, that other statistical models may also be suitable for the purpose of the present invention.
The template statistical model provided for assessing the TRANSLOCATION activity of the NA editing procedure includes a statistical classifier adapted to classify whether translocation events of certain types T or species S originate from NA editing.
For each one or more of translocation classes/types T=[λm1, λm2], and possibly for each species S of one or more species thereof, the system 1000 or method 100 according to an embodiment of the present invention, applies the classifier to classify whether translocation events of this type T and possibly species S, are a result of an editing event or are noise (such as amplification/PCR noise or sequencing noise which also occur in the mock sample).
The above ‘affected’ counts and ‘collective’ counts are computed for each such type T or species S of translocation which is to be determined. Using these counts the classifier is applied/computed to determine whether the translocation of this type/species occur due to edit events of the NA editing procedure (e.g. determine the probability of such occurrence).
In some embodiments the classifier used for translocations is a Probability Distribution function.
In a particular none limiting example, a hypergeometric tail distribution function HGT is used as the classifier for translocations.
The inventors of the present invention have found that hypergeometric tail may be used to determine with good accuracy, whether a translocation of this type T or species S is likely to have occurred due to the NA editing procedure, and more specifically optionally to statistically assess/determine whether this type T or species S, of translocation has occurred due to the NA editing procedure. The assessment of the TRANSLOCATION activity of at least one type T or species S may be determined by computing the probability of the hyper geometric tail distribution based on the respective combined ‘collective’ counts NTx and NMc, and combined ‘affected’ counts nTx and nMc of reads of amplicons which are associated with the respective translocation type T or species S. This may be performed as follows for each particular type/class, and possibly species of S of translocation of interest:
Consider a possible translocation type T between target sites λm1 and λm2. The parameters of the hypergeometric tail distribution function HGT may be set based on the above indicated collective’ counts NTx and NMc, and ‘Affected counts nTx and nMc determined for the particular type T or species S of the translocation, as follows:
b=n
Tx
=C
TX(λm1,λm2);
B=n
Tx
+n
Mc
=CTX(λm1,λm2)+CMc(λm1,λm2)
n=N
Tx≅√{square root over (|CTx(λm1)|·|CTx(λm2)|)}
N=N
Tx
+N
Mc≅√{square root over (|CTx(λm1)|·|CTx(λm2)|)}+√{square root over (|CMc(λm1)|·|CMc(λm2)|)}
Accordingly, the probability P-value of the translocation of this type T or species S of translocation occurring due to an edit even, may be determined based on the hyper geometric tail function in equation (5 above) based on these parameters.
iv. Quantifying the Actual TRANSLOCATION Activity of the NA Editing Procedure and Optionally Determine Confidence Interval for the Same:
As indicated above, according to various embodiments of the present invention the translocation classification in operations (i) to (iii) above may be carried out for one, several or all observed types/classes T and possibly species S of translocations observed in the reads. Accordingly, a list of p-values may be obtained for all considered translocation types/species.
Thus optimally, a rate and a confidence interval of the translocation activity of the NA editing procedure may be determined/quantified for a given translocation type T or, possibly, species S. by computing
Where NTx=√{square root over (|CTx(λm1)|·|CTx(λm2)|)} is the collective read counts average for the two pertinent sites.
For a single species translocation S the number
N
Tx=¼√{square root over (|CTx(λm1)|·|CTx(λm2)|)}
may be used.
And
Optionally, the list of p-values may then be FDR38 corrected (i.e. corrected for False Discovery Rate as would be appreciated by those versed in the art) to filter out translocations with FDR above a certain predetermined FDR threshold (e.g. FDR-threshold=0.05).
The list of p-values (e.g. those with FDR above the FDR threshold) may then be output as a lists of translocation types and possibly species thereof and the respective probabilities (P-values) of their occurrence of due to the NA edit procedure.
Reference is now made together to
To this end,
In the conventional techniques each respective forward primer PRM+A+, PRM+A+) includes a forward binding primer sequence and a forward adapter sequence A+, and each respective reverse primer includes a revers binding primer sequence and a reverse adapter sequence A−.
In this regard it should be appreciated that the term adapter is used herein to indicate either amplification adapter (as generally in a 2-step multiplex amplification process Two-Step-PCR including steps PCR1 and PCR2 briefly described below) or a sequencing adapter (as generally used in either the 1-step multiplex amplification process, One-Step-PCR, or in the second step PCR2 of the 2-step multiplex amplification process, in order to enable sequencing of the amplification products.
Conventional One-Step-PCR/Amplification
Briefly as generally known, the primer molecule types {PRt} used in a 1-step multiplex amplification process, include matched pairs (PRM+m, PRM−m) of forward PRM+m and reverse PRM−m primer molecule types per each site λm of the one or more sites {λm} that are to be amplified, the forward binding primer sequence PRS+m of the forward primer PRM+m of the site m includes a an NA sequence complementary to the site's λm prefix sequence and the revers binding primer sequence PRS−m of the revers primer PRM−m of the site m includes a NA sequence complementary to the site's λm suffix. The respective forwards and reverse adapters, A+ and A−, in this case are typically respectively forwards and reverse sequencing adapters, such as the generally known P5 and P7 adapters.
Conventional Two-Step-PCR/Amplification
In a Two-Step-PCR the primer molecule types {PRt} used in the 1st step PCR1, include, as in the One-Step-PCR, matched pairs (PRM+m, PRM−m) of forward PRM+m and reverse PRM−m primer molecule types per each site λm of the one or more sites {λm} that are to be amplified, the forward binding primer sequence PRS+m of the forward primer PRM+m of the site m in this case includes an NA sequence complementary to the site's λm prefix sequence and the revers binding primer sequence PRS−m of the revers primer PRM−m of the site m includes an NA sequence complementary to the site's λm suffix. However here, the respective forwards and reverse adapters, A+ and A−, are forwards and reverse amplification adapters, which are needed-for/used-in the 2nd step PCR2, and serve as site prefix and suffix for binding the primers of the 2nd step PCR2 (e.g. the forwards and reverse amplification adapters used in the 1st step may be universal for all, or a plurality, of the sites {λm}, so that the forwards and reverse primer molecule types used in the 2nd step PCR2 may be insensitive/non-specific to the particular sequences of the sites {λm}). Accordingly, the 2nd step PCR2 may be conducted with as little as a single type of matched pair (PRM+, PRM−) of forward PRM+ and reverse PRM− primer molecules (e.g. a universal matched pair). The forward primer molecule PRM+ of the 2nd step PCR2 includes a forwards sequencing adapter (e.g. P5) and a forward binding primer sequence PRS+ complementary for binding to the forward amplification adapter used in the 1st step (e.g. non-site-specific), and accordingly the reverse primer molecule PRM− of the 2nd step includes a revers sequencing adapter (e.g. P7) and a revers binding primer sequence PRS− complementary for binding to the reverse amplification adapter used in the 1st step (e.g. also non-site-specific).
In view of the above amplification's products of each site of interests λm, which are produced by either the above described conventional One-Step-PCR or Two-Step-PCR multiplex PCR amplifications, include amplicons of the site of interest with two match pairs of sequencing adapters (e.g. P5 and P7), from either side of the site's λm amplicon. This facilitates sequencing of the site's amplicons since such configuration of the sequencing adapters from either side thereof are required for the sequencing process, particularly for NGS sequencing.
However, in implementations of the Conventional multiplex PCR process (e.g. the One-Step-PCR or Two-Step-PCR processes described above) for detection of translocations, not all the translocation species will be produced/amplified with the amplicons having the suitable arrangement of forwards and revers sequencing adapters from either side thereof. Indeed, as illustrated in the figure only the single-centromeric translocations species A and B, which are described above will be amplified with the required forwards A+ and revers A− sequencing adapters from either side thereof, while the centromere-free translocations species C and double-centromeric translocations species D described above will be amplified with either the forward or the revers sequencing adapters appearing from both sides thereof. Accordingly, the centromere-free translocations species C and the double-centromeric translocations species D will not be sequenced with the conventional simplification techniques.
To overcome this deficiency of the conventional techniques, the present invention, in some embodiments thereof, provides a kit 300 for determining effects of a NA editing procedure, the kit 300 includes a set of a plurality of primer molecule types {PRt} designed to provide amplification of expected editing sites {λm}1M of the NA editing procedure, whereby the expected editing sites {λm}1M include at least one on-target site {λ1} and one or more off-target sites {λm}2M, where λm represents an off-target or on-target site indexed m, and M is a number of the expected on-target and off-target sites.
In this regard it should be understood that in the scope of
To this end, the set/kit 300 of the plurality of primer molecule types {PRt} according to an embodiment of the present invention includes pairs (PRM+, PRM−) of forward PRM+ and reverse PRM− primer molecule types {(PRM+, PRM−)∈{PRt} suitable for amplification of said on-target and off-target sites {λm}1M such that each respective forward and revers primer molecule, PRM+ and PRM−, include at least one of a forward and revers adapters, A+ and A− (e.g. P5 and P7 in case of sequencing adapters or other types of adapters of example forward and revers amplification adapters);
The set/kit 300 of the plurality of primer molecule types {PRt} is characterized in that the plurality of primer types {PRt} includes:
thereby enabling sequencing of all possible translocation species between at least one pair of the editing sites {λm}1M.
The Kit 300 may be configured and operable for use in any suitable Multiplex-Amplification process.
To this end, in some embodiments the kit may be configured for use in One-Step-Multiplex Amplification process or a 1st step, PCR1, of a Two-Step-Multiplex-Amplification process. The primer molecule types {PRt} of the Kit 300 in this case, include pairs (PRM+m, PRM−m) of forward PRM+m and reverse PRM−m primer molecule types, per each site λm of the one or more sites {λm} that are to be amplified. In this case the forward binding primer sequence PRS+m of the forward primer PRM+m of the site m includes an NA sequence complementary to the site's λm prefix sequence and the revers binding primer sequence PRS-m of the revers primer PRM−m of the site m includes an NA sequence complementary to the site's λm suffix.
Specifically in such embodiments, the Kit 300 may be configured for use in a 1st step, PCR1, of a Two-Step-Multiplex-Amplification process. The forward and reverse adapters, A+ and A−, in this case are forwards and reverse amplification adapters facilitating that all said translocation species will be amplified in said 2nd step, PCR2, of the Two-Step-Multiplex-Amplification process, to produce amplicons thereof which have forwards and revers sequencing adapters from either side of the amplicon.
Alternatively, or additionally, the Kit 300 may be configured for use in One-Step-Multiplex Amplification process. In this case, the forwards and reverse adapters A+ and A− are forwards and reverse sequencing adapters.
In some embodiments the Kit 300 may be configured for use in a 2nd step, PCR2, of a Two-Step-Multiplex-Amplification process. In such embodiments the primer molecule types {PRt} include pairs (PRM+, PRM−) of forward PRM+ and reverse PRM− primer molecule types including respective forward PRS+ and reverse PRS− binding primer sequences complementary to respective forwards and reverse amplification adapters (not shown) of a 1st step, PCR1, of the Two-Step-Multiplex-Amplification process. In this case the forwards and reverse adapters, A+ and A−, of the forward PRM+ and reverse PRM− primer molecule types, are forwards and reverse sequencing adapters.
Reference is now made to
As indicated above, the system 1000 may be include a non-transitory computer readable medium storing instructions executable by a processor, for utilizing the sequencing products of any of these kits to determine and output data indicative of the effects of Nucleic Acid (NA) editing procedure according to any of the methods described above according to the present invention.
Several tests were performed to assess the accuracy of the technology of the present invention in determining the occurrence of various adverse effects of NA editing procedures.
Some experiments use a generally known rhAmpSeq assay (IDT, Coralville, IA) for the PCR and were conducted with a Tx vs M design. These yielded a total of 1,161 instances. The results obtained from evaluating the indel and translocation activity by applying the technique of the present invention to the sequencing data from these experiments, shown that the technique of the present invention accurately estimates indel activity levels at off-target sites.
Indeed, the editing activity estimation according to the present invention is based on the ability to model the background noise while being blind/insensitive to the source of the noise, whether it comes from high NGS error rates (
(a-b)), false site assignments, or ambiguous alignments. The template classifier applied to the results of each experiment statistically models the background noise thereby quantify editing events.
The performance of technique of the present invention were tested on the challenging off-target scenarios, where high error rates occur at sites with low editing activity, as well as in scenarios where process related (NGS, PCR, etc.) error rates can lead to false-negative inferences. The results of these test showed the technique of the present invention can recover the true validated editing activity even when error and editing rates are near identical, where the true validated editing activity was obtained by human examination of the actual reads as well as by two lines of statistical evidence as explained above.
Accurate editing activity estimation may also depend on the detection of alternative cut-sites. Flawed identification of the off-target gRNA binding configuration, due to an ambiguous alignment or false interpretation of GUIDE-seq or other screening methods, can lead to a mis-inferred PAM position. Moreover, optimal read alignments, even if slightly better justified from a biochemical perspective, can place real edit events away from the expected cut-site. Finally, real editing activity can occur away from the expected cut-site due to the existence of an alternative PAM sequence or less frequent non-canonical DSB mechanisms. The technique of the present invention facilitates the detection of alternative cut-sites by incorporating different prior probabilities for each position in the reference sequence, as described above. As indicated above, one novel and inventive feature of the technique of the present invention lies in is its ability to detect translocations resulting from NA/CRISPR editing procedures, by analyzing NGS data produced by a multiplex PCR using locus-specific primers (for instance such as rhAmpSeq34). Using the multiplex PCR mechanism for target enrichment (with primers designed to span the potential cut-positions of the off-target sites), four species of translocation events can occur for every pair of potential partner loci. The technique of the present invention is capable of analyzing the mixed pairs of primer sequences that are detected on common reads in the NGS data. These reads represent putative fusion amplicons which may be due to translocations by NA edit procedure. A statistical model (e.g. hypergeometric) may be applied according to the present invention, to those reads to infer there statistical significance and determine those who most probably pertain to translocations with significant (FDR corrected) p-values (≤0.05, by default).
In several tests conducted with the technique of the present invention, significant translocations were detected. For example, for editing procedures conducted on RAG1 and RAG2 loci in HEK293-Cas9 cells12, the technique of the present invention revealed evidence for translocations in 20 and 19 unique pairs of sites, for RAG1 and RAG2, respectively. The most significant corrected p-values are 4.9*10−22 for on-target site 1 with off-target site 7 in RAG1, and 1.53*10−53 for off-target site 1 with off-target site 5 in RAG2.
These results were experimentally confirmed in an independent measurement using the singleplex droplet digital PCRS using primers designed to separately amplify individual potential translocation events.
It is noted that occurrences of all four possible configuration species of translocations for different pairs of loci were detected including centromere-free, double-centromeric, and two single-centromeric configurations. This was achieved by using a multiplex amplification with the primer kit 300 as described above, and particularly using PCR panel with both P5 and P7 on both the reverse and forward primers, to enable measuring all four types of translocations at every potential fusion site covered by the assay. In conclusion, the technique of the present invention provides significant improvement in the accuracy of determination of genome/NA-editing adverse effects to enhance and accelerate the sound and accurate broader use of NA editing in biotechnology and therapeutic applications. A person of ordinary skill in the art will readily appreciate the various modification which can be implemented to the above describes systems and methods without departing from the scope of the present invention as defined by the claims.
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/IL2021/050539 | 5/11/2021 | WO |
| Number | Date | Country | |
|---|---|---|---|
| 63022912 | May 2020 | US |
