Patent Application
20190080045
The present disclosure concerns a method of determining the presence of a novel structural variation in a genome using long-read genome sequence fragments. By a process involving aligning, filtering ranking and linking long-read sequence fragments against a reference genome, presence of a novel structural variation in said genome can be determined when said linked alignment contains multiple linked fragments that are mapped to a single locus, referred to as a split-read.
Genomic structural variation is prevalent in the human genome and includes deletions, insertions, duplications, inversions, and translocations. Collectively, these structural variants (SVs) account for a significant portion of genome heterogeneity between individuals and human populations. The complexity and diversity of genome structure changes have been shown to influence genome evolution and genetic diversity among populations, while specific SVs are associated with disease susceptibility. Many cancer genomes have been found to harbor significant structural variation, and specific SVs are considered instrumental in promoting tumor progression by disrupting gene structures, dysregulating gene expression, creating fusing transcription units or increasing gene copy number. The detection of specific SVs can be used as the basis for tumor classification and potentially of prognostic value for tumor severity and therapeutic response. However, despite the prevalence of SVs and their particular relevance to cancer, the molecular organization of various SV classes and the mechanisms that generate them are not well understood. This is in large part due to the inability of current technologies to uncover the full spectrum of SVs with high specificity at nucleotide-level resolution.
Advances in sequencing technology coupled with improvements in computational algorithms have greatly enhanced our understanding of the abundance, diversity, and molecular features of SVs across human populations and disease. However, current sequencing approaches that generate high coverage, paired-end short-read sequencing data, combined with split read mapping methods lack the precision and sensitivity necessary for the detection of many types of SVs, particularly in regions of repetitive sequence. Specifically, paired-end short reads are not sufficiently sensitive to detect small SVs, which require a high depth of sequencing coverage to achieve high specificity, and lack the nucleotide-level of detail for analysis of the breakpoints that flank SVs. They are also unable to decipher complex SV patterns or provide haplotype phase of SVs in diploid genomes. Thus, there is an ongoing effort to develop effective long-read sequencing approaches for the characterization of SVs.
What is therefore needed is a method that harnesses the newly-developed capabilities of long-read sequencing techniques to detect a full spectrum of SVs with high specificity and sensitivity.
A method of determining the presence of a novel structural variation in a genome is disclosed. The method comprises: a) providing a plurality of long-read genome sequences derived from a genome; b) aligning said long-read genome sequences with a reference genome sequence to produce a plurality of alignments by using alignment parameters configured for low sequence similarity; c) filtering said plurality of alignments by removing low quality alignments to yield remaining alignments based on an alignment parameter; d) ranking said remaining alignments based on (i) a probability of random hit, and (ii) an alignment score; e) selecting a seed candidate alignment, wherein said seed candidate alignment has a highest rank as compared to the remaining alignments; f) linking said seed candidate alignment with said remaining alignments to create a linked alignment extension having a combined alignment score, said linking is performed by using read coordinates of said seed candidate alignment in vicinity to read coordinates of said remaining alignment to cover a maximal sequence length; g) repeating step e) and step f) to obtain a plurality of linked alignment extensions; h) selecting from step g) a best linked alignment extension that has a highest combined alignment score; and i) determining whether a novel structural variation is present in said genome, wherein when said best linked alignment extension contains multiple linked alignments that are mapped to a single locus is indicative of the presence of a novel structural variation.
These and other features can be appreciated from the accompanying description of certain embodiments which are discussed in relation to the accompanying drawing figures.
The following definitions of terms apply to the disclosure herein:
“About” is defined as at least close to and including a given value or state (preferably within 10%).
A “long-read genomic sequence” (also referred to as “long-read data” herein) is a sequence of nucleotides containing 10,000 or more bases for single-stranded DNA, or 50,000 or more base pairs for double-stranded DNA.
A “long read coordinate” is a specific numerical coordinate assigned to each nucleotide within a genomic sequence obtained by a long-read sequencing process.
“Nucleic acid” and “polynucleotide” are terms that can be used interchangeably herein, and refer to both RNA and DNA, including cDNA, genomic DNA, synthetic DNA, and DNA or RNA containing nucleic acid analogs. Polynucleotides can have any three-dimensional structure. A nucleic acid can be double-stranded or single-stranded (e.g., a sense strand or an antisense strand). Non-limiting examples of polynucleotides include chromosomes, chromosome fragments, genes, intergenic regions, gene fragments, exons, introns, messenger RNA (mRNA), transfer RNA, ribosomal RNA, siRNA, micro-RNA, ribozymes, cDNA, recombinant polynucleotides, branched polynucleotides, nucleic acid probes and nucleic acid primers. A polynucleotide may contain unconventional or modified nucleotides
A“nucleotide” is a molecule that when joined together form the structural basis of polynucleotides, e.g., ribonucleic acids (RNA) and deoxyribonucleic acids (DNA). A “nucleotide sequence” is the sequence of nucleotides in a given polynucleotide. A nucleotide sequence can also be the complete or partial sequence of an individual's genome and can therefore encompass the sequence of multiple, physically distinct polynucleotides (e.g., chromosomes).
An “individual” can be of any species of interest that comprises genetic information. The individual can be an eukaryote, a prokaryote, or a virus. The individual can be an animal or a plant. The individual can be a human or non-human animal.
A “genome” of an individual member of a species comprises that individual's complete set of chromosomes, including both coding and non-coding regions. Particular locations within the genome of a species are referred to as “loci”, “sites” or “features”. “Alleles” are varying forms of the genomic DNA located at a given site. In the case of a site where there are two distinct alleles in a species, referred to as “A” and “B”, each individual member of the species can have one of four combinations: AA; AB; BA; and BB. The first allele of each pair is inherited from one parent, and the second from the other.
“Alignment” or “aligned fragment”—the terms alignment and aligned fragment can be used interchangeably herein and both denote a genomic string or segment that has been matched to a locus of a reference genome.
A “structural variation” (“SV”) is any change in an individual nucleotide sequence compared to a reference sequence.
An “inversion” (“INV”) is a mutation in which a segment of DNA that is reversed in orientation with respect to the rest of the chromosome.
An “insertion” (“INS”) is a mutation in which a sequence of DNA is added with respect to the reference genome.
A “deletion” (“DEL”) is a mutation in which a sequence of DNA is lost with respect to the reference genome.
An “indel” (INDEL”) is a mutation involving both insertion and deletion of bases in the genome.
A “duplication” is a sequence alteration whereby the copy number of a given region is greater than the reference sequence.
A “tandem duplication” (“TD”) is a duplication consisting of two identical adjacent regions.
A “tandem duplication split-read” (“TDSR”) is a tandem duplication that does not align contiguously to the reference genome, and is therefore broken into two or more fragments, in which adjacent fragments align to non-adjacent locations in the reference genome.
A “tandem duplication complete” (“TDC”) is a tandem duplication in which a single fragment contains the duplication region.
A “translocation” (“TLC”) is a change in position of a chromosomal segment within a genome that involves no change to the total DNA content. Translocations can be intra-chromosomal (“CTLC”—cis-chromosomal translocation) or inter-chromosomal (“TTLC”-trans-chromosomal translocation).
The following terms are parameters used in the Smith-Waterman algorithm which determines the similarity between two strings of DNA.
A “gap penalty” is a scoring penalty for opening or extending gaps.
A “match score: (“r”) is a numerical value (e.g., 1) attributed to a match of two compared elements in the substitution matrix (i.e., for an entry s(α,β), when base a=base b).
A “mismatch cost” (“q”) is a numerical value (e.g., −1) attributed to a mismatch of two compared elements in the substitution matrix (i.e., for an entry s(α,β) when base a≠ base b).
“Wk” is a numerical value attributed as a penalty to a gap of length k.
A “gap existence cost” (“a”) is a numerical value (e.g., 0) used in the calculation of Wk according to the equation Wk=gap existence cost (a)+gap extension cost(b)*k.
A “gas extension cost” (“b”) is a numerical value (e.g., 2) used in the calculation of Wk according to the equation Wk=gap existence cost (a)+gap extension cost(b)*k.
Disclosed herein is a method that employs long-read genomic sequence data (hereinafter termed “long-read data”) to determine the presence of a novel structural variation in a genome and characterize their genomic breakpoints with high specificity and sensitivity. To exploit the value of long reads in SV detection, a computational analysis pipeline is provided, which optimally performs read alignments and logically defines the full spectrum of SVs, including complex SVs enriched in repetitive DNA elements. Embodiments of the computational analysis pipeline of the invention may also be referred to herein as “Picky”). The method of analyzing long-read data to determine SVs in a genomic sample as disclosed herein comprises an end-to-end analysis pipeline that includes three general phases: 1) alignment of long-read data obtained from a long-read sequencing process to a reference genome; 2) optimal alignment merge/selection; and 3) SV classification. The analysis is performed using computer-readable code (i.e., one or more programs or scripts) executed on a computing system.
The determination of structural variations begins with a long-read sequence procedure that generates a plurality of long-read genomic sequences. Several different long-read sequencing techniques that generate long-read data have been developed and any of these techniques can be used to produce the data used for the structural variation analysis disclosed herein. For example, instruments from both Pacific Biosciences™ and Oxford Nanopore™ platforms produce read lengths in the thousands of bases per read. Pacific Biosciences uses optical detection of a sequencing-by-synthesis reaction that occurs inside a waveguide, while Oxford Nanopore uses nanopores for detection.
Recent progress in nanopore single-molecule sequencing and other long-read sequencing techniques offers the ability to extend sequencing read length and throughput, features that can facilitate detection and analysis of SVs in large complex genome. For instance, the nanopore platform sequences individual DNA molecules by passing them through nanometer-scaled pores and measuring changes in the electric current across the pore caused by the transiting nucleotides. These current changes are then computationally deconvoluted to reveal the identity of the bases. Unlike conventional sequencing-by-synthesis technology platforms, nanopore sequencing can generate reads of 10-100 Kb in real-time. Other long-read techniques have generated similar results. Such long-read sequencing techniques improve the speed, read length and throughput necessary for comprehensive and unbiased SV profiling and would be particularly useful for resolving complex structural rearrangements in cancer genomes.
The output of the long-read sequencing process comprises a plurality of long-read genomic sequence fragments. These fragments, along with a reference genomic sequence, comprise the input data used in the initial alignment step. The fragments of long-read data are compared to the bases of a reference genome, matches and mismatches between the fragments of long-read data and the reference genome are determined. To perform the alignment, an alignment program, such as LAST (Local Alignment Search Tool) [Kielbasa, et al. Genome Res. 21, 487-493 (2011) and Frith, et al., BMC Bioinformatics 11, 80 (2010)] is accessed and executed to perform the genomic alignment. While long-read sequencing techniques such as the Oxford nanopore technique produce long sequences with high-throughput, their efficiency can come at the cost of a high error rate. To accommodate a high error rate for long-read sequencing, the alignment process was configured with parameters adapted for low similarity between the long-read data and the reference genomic data. For sequences in which a high degree of divergence (similarity on the order of 75% to 95%) is expected, the ratio between reward scores for matches and penalties for mismatches is set relatively high (e.g., >0.5) to boost sensitivity. In certain embodiments, the ratio can be set to different values, e.g., about 0.1 to about 0.2, about 0.2 to about 0.3, about 0.3 to about 0.4, about 0.4 to about 0.5, about 0.5 to about 0.6, about 0.6 to about 0.7, about 0.7 to about 0.8, or about 0.8 to about 0.9 depending on the known accuracy of the long read process used. Recommended settings for low similarity configurations are provided by the megaBLAST greedy alignment algorithm. Exemplary parameters that can be used for alignment according to the megaBLAST greedy algorithm are: match score (reward)=1; mismatch cost (penalty)=−1; gap existence cost (open)=0; and gap extension cost=2.
Reference is now made to
After the plurality of long-read data fragments have been aligned, and alignment scores and EG2 value have been assigned to each fragment, the plurality of fragments are evaluated for their quality. In some embodiments, low quality alignments are filtered out based on low alignment score or low percentage identity and EG2 criteria. For example, aligned fragments having an EG2 score greater than a threshold value or less than an identity % threshold value are considered to have low quality and are removed. In certain embodiments, the EG2 score threshold is 5.0e−13 to 8.0e−13, 8.0e−13 to 1.1e−12, 1.1e−12 to 1.4e−12, 1.4e−12 to 1.7e−12, 1.7e−12 to 2.0e−—, 2.0e−12 to 2.3e−12, or 2.3e−12 to 2.6e−12. In addition, in certain embodiments, the % identity threshold is about 30% to about 40%, about 40% to about 50%, about 50% to about 60%, or about 60% to about 70%.
In certain embodiments, the high quality aligned fragments remaining after the low quality aligned fragments have been removed are ranked in a cardinal order according to both EG2 value (lowest) and alignment score (highest). In some implementations, the EG2 criterion is weighed higher, so that, for example, if one particular aligned fragment has the lower EG2 score, while a different aligned fragment has the highest alignment sore, the aligned fragment with the lowest EG2 is ranked higher than the aligned fragment with the highest alignment score. Generally, however, the fragment with the lowest EG2 score will have the highest alignment score in the vast majority of cases. When a ranking is complete, the number “n” of remaining aligned fragments are ordered by rank in terms of a “best” fragment, “second best” fragment, and so on up to an nth best fragment. Referring to
The retained aligned fragments can be subdivided into 3 subgroups: i) an ‘exact’ subgroup comprised of fragments that share the exact same start and end long-read coordinates but having different start and end genomic coordinates; ii) a ‘subset” subgroup comprised of fragments having long-read coordinates that are an exact sub-span of the representative fragment of the subgroup (the fragment having a lowest EG2 score and highest alignment score or % identity with respect to the reference genome); and iii) a “similar” subgroup that comprised of fragments having 95% reciprocal overlap and % identity difference ≤5 with respect to the representative fragment of the subgroup. The subgroups are exemplary, and more or fewer subgroups can be used with different criteria. The main purpose of establishing the subgroups is to make the long read analysis more efficient by attempting to sort out and use alignments that have similar quality, which improves error rates and precision.
Once the non-filtered aligned fragments remaining have been ranked, a seed alignment is selected for a seed-and-extension algorithm that is used to link fragments together to maximize coverage for each long read. In certain embodiments, a “greedy” seed-and-extension algorithm can be used. The seed-and-extension algorithm starts with selection of a seed candidate group for each of the subgroups of remaining aligned segments. The seed group is a group of alignment fragments that is parsed for long-read coordinates to determine whether the members of the group can be linked together to increase the span of a continuous string of bases. For each subgroup, the first seed candidate selected is the aligned fragment with the highest rank. This seed candidate is then combined with any other fragments in the subgroup having the same or about the same EG2 value as the selected candidate, to produce a seed group. If the seed group produced contains greater than a threshold number (e.g., 5 to 10) aligned fragments, a selected number of the fragments of the seed group (e.g., the top 200) are selected as potential linkers. In certain embodiments, the threshold number of fragments is 4, 5, 6, 7, 8, 9 or 10 fragments. Additionally, the number of fragments of the seed group selected as potential linkers can be set to about 140 to 160, about 160 to 180, about 180 to about 200, about 200 to about 220, about 220 to about 240 or about 240 to about 260. If there are fewer than the set number of alignments in the seed group, the number of potential linker fragments can be iteratively increased by including successive fragments within EG2 value bins. In other words, fragments can be added starting with the EG2=0 bin, increasing to the 0<EG2<=1e−300 bin, and so on to the 1e−300<EG2<=1e−200 bin, etc. until the number of potential linker fragments reaches a selected number.
After seed groups have been assembled, the process of linking and extension starts with selection of the highest ranking of the seed group as a seed candidate alignment for linking and extension. A second alignment of the seed group is then selected, and it is then determined whether the second alignment can be linked to the seed candidate alignment, based on whether the long-read coordinates of second alignment overlap with the long-read coordinates of the seed candidate alignment. Alternatively, if there is no overlap, linking can also occur when a termination point of the second alignment is within 500 bases of a termination point of the seed candidate alignment (i.e., when a gap between the seed candidate alignment and the second alignment is less than 500 bases wide). If the first of these conditions is met (overlap), it is then determined whether the overlap between the seed candidate alignment and the second alignment covers a threshold percentage percent or more of either alignment. In certain embodiments, the threshold overlap percentage is about 30% to 40%, about 40% to about 50%, about 60%, or about 60% to about 70%. If the overlap is higher than the overlap threshold, the second alignment is disqualified as a potential linker (extender). If, conversely, there is no overlap and a gap of less than 500 base gap, or if there is an overlap that is less than the overlap threshold, then the second alignment is linked to the seed candidate alignment as a combined fragment. The linking procedure is repeated, with the combined alignment(s) standing in the place of the original seed alignment for each repetition, until either the entire coordinate span of the long-read has been covered, or there are no further alignments in the seed group to link.
It is noted that in alternative embodiments, it is possible to start with seed candidates other than the highest ranked alignment. However, the highest ranked alignment will tend to be the longest fragment that aligns well with the reference genome, and therefore tends to provide a suitable initial fragment for the linking process.
An illustration of the linking and extension process is shown in
The linking step starts with a particular seed candidate alignment. However, the optimal linked extension may not be obtainable using the particular seed alignment selected. Therefore, the seed candidate selection (step 5) is repeated, followed by an attempt to link the other alignments in the seed group to each successively selected seed alignment (step 6). Put another way, the linking and extension, for any particular seed candidate alignment, occurs in a loop that is nested within a seed selection loop, so that iterative attempts to link and extend are made for all of the seed candidate alignments.
In some implementations, to focus more efficiently on the longest linked extensions, any linked alignment extensions that cover less than a threshold percentage of the entire long-read are discarded. In certain embodiments, the threshold percentage is about 50% to 60%, about 60% to about 70%, about 70% to about 80%, or about 80% to about 90%. The remaining linked alignment extensions are then scored. The score of the linked alignment extensions is calculated as the sum of the scores of all of the fragments that form the extensions. The linked alignment extension with the highest combined score is determined, and this extension and all extensions having a score within 10 percent of the highest score are retained for further analysis.
The final output of prior steps (1 to 8) comprises either one or more linked alignment extensions (i.e., combined alignments) that cover the threshold percentage of the long-read, or, if this criterion is not met, no valid output. In the latter case, if no combined alignment can be derived, the long-read analysis does not determine the presence of a structural variation. If a single linked alignment extension remains, the number of fragments within the linked extension is first determined. If it is determined that the linked extension contains only a single fragment (i.e., no other fragments are linked, and the single fragment cover the threshold percentage of the long-read itself), in this case again, the long-read analysis does not determine whether a structural variation is present in the long-read. If, on the other hand, it is determined that the linked extension contains two or more fragments, if each of the fragments are mapped to a single locus of the reference genome, then it is determined that the combined alignment is a “split-read” which may capture one or more structural variations. However, if the combined alignment contains more than one fragment mapped to multiple loci of the reference genome, then in this case as well, the long-read analysis stops and it is not determined whether structural variations are present in the long-read.
In addition, a further cross-check can be performed. If it turns out that there are two distinct combined alignments that have mutual overlap of another set percentage of their respective bases, the long-read is discarded as any structural variation information in this case is considered ambiguous. In certain embodiments, the set percentage for discarding is about 90% to about 92%, about 92% to about 94%, about 94% to about 96%, about 96% to about 98%, or about 98% to about 100%.
Upon determining that a linked alignment extension is a split-read that may contain a structural variation, the analysis can proceed to classify the variation in terms of one of the known structural variation types. In performing classification, the coordinates of adjacent pairs of aligned fragments are compared. A first parameter (“sDiff”) is calculated which represents the distance between the adjacent fragments based on their reference genome coordinates. The sDiff value is greater than zero (>0) if there is a gap (measured in base pairs) between the adjacent segments, and is less than zero (<0) if the adjacent segments overlap. In addition, a second parameter (“qDiff”) is calculated which represents the distance between the adjacent fragments based on their long-read coordinates. Similarly, the qDiff value is greater than zero (>0) if there is a gap between the adjacent segments, and is less than zero (<0) if the adjacent segments overlap. Certain structural types can be determined directly based on the parameters sDiff and qDiff. For example, the following guidelines in the table below can be used for classifying insertions, deletions and indels. It is noted that the number of base pairs used as thresholds for sDiff and qDiff (e.g., 20 as indicated in the table below) can vary in different embodiments. For example, in an alternative embodiment, an insertion can be determined if sDiff <25 and qDiff >18. Likewise, in certain embodiments, the qDiff threshold is 15 to 17, 17 to 19, 19 to about 21, 21 to 23, or 23 to 25.
Other structural variations include (following i to iii from the table above): iv) inversions (INVs), which are classified by determining that the split-read fragments are from the same chromosome and have different orientation; v) translations (TLCs), which are classified by determining that the split-read fragments are from different chromosomes; vi) tandem duplications (TDs), which are classified by determining that the split-read fragments capture a duplication junction (TDJ) with the tail and head of a duplicated fragment; and vii) tandem duplication completes (TDCs) which are classified by determining that the split-read fragments capture a complete duplication fragment and sDiff ≤100. Further detail on assignments of split reads into seven classes of SVs is provided described elsewhere herein, see also
Referring again to
The method continues in
After step 524, the method continues in
The disclosed method is developed and executed using one more computers including terminals servers and storage devices. A computer generally includes one or more processors, computer-readable storage devices, and input/output devices. A processor may be any suitable processor such as the microprocessor. A computer-readable storage device can include any machine-readable medium or media on or in which is stored instructions (one or more software applications), data, or both. The instructions, when executed, can implement any or all of the functionality described herein. The data can be the genomic data as described herein. The term “computer-readable storage device” shall be taken to include, without limit, one or more disk drives, tape drives, memory devices (such as RAM, ROM, EPROM, etc.), optical storage devices, and/or any other non-transitory and tangible storage medium or media.
Input/output devices according to the invention may include a video display unit (e.g., a liquid crystal display (LCD) or a cathode ray tube (CRT) monitor), an alphanumeric input device (e.g., a keyboard), a cursor control device (e.g., a mouse or trackpad), a disk drive unit, a signal generation device (e.g., a speaker), a touchscreen, an accelerometer, a microphone, a cellular radio frequency antenna, and a network interface device, which can be, for example, a network interface card (NIC), Wi-Fi card, or cellular modem.
The disclosed methods utilize one or more development programming environments and databases. The databases may be accessible from the programming environment as online resources or locally stored. Any development environment, database, or language known in the art may be used to implement embodiments of the methods. Preferably, an object-oriented development language, database structure, or development environment is used. Exemplary languages, systems, and development environments include Perl, C++, Python, Ruby on Rails, JAVA, Groovy, Grails, Visual Basic .NET. In some embodiments, one or more applications have been developed in Perl (e.g., optionally using BioPerl). Object-oriented development in Perl is discussed in Tisdall, Mastering Perl for Bioinformatics, O'Reilly & Associates, Inc., Sebastopol, Calif. 2003.
To assess the utility of the long-read sequencing for detecting SVs, a long-read analysis was performed on the genome of the breast cancer cell line HCC118719, a model for triple-negative breast cancer (TNBC) whose genome harbors extensive structural variation that has been previously characterized at the molecular level by paired-end short-read whole genome sequencing. See Menghi, F. et al, “The tandem duplicator phenotype as a distinct genomic configuration in cancer”, Proc. Natl. Acad. Sci. USA 113, E2373-2382 (2016); Gazdar, A. F. et al. Characterization of paired tumor and non-tumor cell lines established from patients with breast cancer. Int. J. Cancer 78, 766-774 (1998). A comparison of the results of long-read sequencing analysis was made to this previously generated data set. Long-read sequencing libraries were prepared from fragmented high-molecular-weight genomic DNA and subjected to sequencing.
Long-Read Sequencing:
High molecular weight DNA was extracted from HCC1187 cells by MagAttract HMW DNA Kit (Qiagen, 67563) following the manufacturer's instructions. Briefly, 1×106 frozen cells were lysed with 220 μL of Buffer ATL, 20 μL Proteinase K and incubated overnight at 56° C. with 900 rpm. 4 μL RNase A was added to cleave RNA. 150 μL Buffer AL, 280 μL Buffer MB and 40 μL MagAttract Suspension G beads were then added to capture the HMW DNA. Next, the beads were cleaned up by 700 μL Buffer MW1, Buffer PE and NFW followed by elution with 150 μL Buffer AE.
Long-read sequencing libraries were prepared according to the target size and the sequencing kits supplied by Oxford Nanopore Technologies (ONT). Human molecular weight genomic DNA was fragmented by one of miniTUBE Blue (Covaris, 520065, for 3 Kb), miniTUBE Red (Covaris, 520066, for 5 Kb) or g-TUBE (Covaris, 520079, for 8 Kb and 12 Kb). For libraries targeted at 12 Kb, size selection was performed for sheared fragments larger than 10 Kb using a 0.75% agarose cassette (Sage Science, BLF7510) by the Blue Pippin™ DNA Size Selection System. For libraries of less than 10 Kb, AMPure XP beads (Beckman Coulter, A63881) were used for clean-up. Next, libraries were prepared according to recommendations of ONT.
Briefly, NEBNext FFPE RepairMix (NEB, M6630) was added to repair nicks in the DNA. Then end-repair and dA-tailing were performed using NEBNext Ultra II End-Repair/dA-tailing Module (NEB, E7546). For 2D libraries (WTD01-WTD13), a ligation reaction was prepared as follows: 38 μL water (DNA), 10 μL Adapter Mix (AMX), 2 μL Hairpin Adapter (HPA) and 50 μL Blunt/TA Ligase Master Mix (New England Biolabs, M0367). Ligation was performed at room temperature for 15 minutes. A 1 μL Hairpin Tether (HPT) was added to the reaction and incubated at room temperature for 15 minutes. Then 50 μL MyOne C1 beads (Thermo Fisher, 65001) beads were prewashed twice with 100 μL Bead Binding Buffer (BBB). The MyOne C1 beads resuspended in 100 μL BBB were added to the ligated DNA reaction and incubated on a rotator at room temperature for 15 minutes. The beads were washed twice with 150 μL BBB and eluted in 25 μL Elution Buffer (ELB). For 1D libraries (WTD14-WTD15), the ligation reaction was prepared as follows: 30 μL water (DNA), 20 μL Adapter Mix (AMX) and 50 μL Blunt/TA Ligase Master Mix (New England Biolabs, M0367). Ligation was also performed at room temperature for 15 minutes. The AMPure XP beads were resuspended at room temperature by vortex. 40 μL of beads were added into the ligation product. The beads were washed twice with 140 μL Aldapter Bead Binding (ABB) buffer and eluted in 25 μL Elution Buffer (ELB). The eluted product was the adaptor-ligated library as the Pre-sequencing Mix (PSM) used in nanopore sequencing.
The libraries were sequenced on MinION Mk1b device using R9 and R9.4 flow cells following the standard 48 h run scripts. Real-time basecalling was performed on EPI2ME cloud platform (ONT). Read sequences were extracted from base-called fastf5 files by Poretools (version 0.5.1) to generate a fastq file. All 2D reads from WTD01-13 (2D ligation libraries) and all 1D reads from WTD14 and WTD15 (1D ligation libraries) were used for subsequent analysis.
In one set of test experiments, the above-described long-read computational analysis pipeline was applied to 796,029 2D reads to detect and classify SVs in the HCC1187 genome (
In addition, DNA templates were prepared as described in 12 Kb nanopore sequencing libraries. A PacBio genomic DNA library was prepared (12 Kb) according to the manufacturer's instructions (Greater Than 10 Kb Template Preparation Protocol). The library was sequenced on PacBio RS II instrument. PacBio sequencing data was processed by the PacBio SMRT Portal pipeline of Read of Insert with the parameters “min full pass=0” and “min predicted accuracy=75”.
Pipeline for SV Detection:
The long-reads were processed according to the long-read analysis method disclosed above. Firstly, the nanopore long-reads were mapped to human genome (hg19) using the LAST aligner (last755) with the parameters “−r1−q1−a0−b2” set to 1 (reward), −1 (penalty), 0 (gap cost (open)) and 0 (gap extension cost) for high sensitivity. See Stephens, P. J. et al. “Complex landscapes of somatic rearrangement in human breast cancer genomes”, Nature 462, 1005-1010 (2009). The LAST aligner reported all high scoring pair (HSP) fragments for each nanopore read. Next, fragments with EG2>1.74743e−12 or % Identity <55 were discarded. Nanopore reads with no aligned fragments post-filtering were considered unmapped and reads with only a single segment were reported as uniquely mapped. For read with multiple fragments, the best fragment (lowest EG2 and highest score) was used to determine the EG2 threshold (EG2best) and % identity (% identitybest). Fragments which EG2<1e49 and the absolute difference between alignment % identity and % identitybest <=10% were retained.
To speed up processing, fragments were grouped into representative subgroups, namely ‘exact’, ‘subset’, ‘similar’. Fragments in subgroup ‘exact’ share the exact read start and end coordinates although their genomic coordinates differ. Subgroup ‘subset’ contains fragments having read coordinates that are a sub-span of the representative fragment for the subgroup. Members of subgroup ‘similar’ have 95% reciprocal overlap and % identity difference <=5 with respect to the group's representative fragment. A greedy seed-and-extend strategy was used to stitch (link) fragments together and produce the linked alignment extensions that span as much of the nanopore read as possible. The fragments for which EG2=EG2best were designated as the seeds. If there were >=5 seeds, the top 200 of them were used as potential extenders. Otherwise, potential extenders could consist of the fragments group(s) where EG2=0 minimally, with the successive inclusion of the 4 fragment group collections binned by 0<EG2<=1e−300, 1e−300<EG2<=1e−200, 1e−200<EG2<=1e−100, and 1e−100<EG2<=1e−50, with the total number of extenders capped at 100.
The seed and extension process began with selection of a seed candidate and attempts were made to extend the seed candidate with the potential extenders to maximize the coverage of the entire nanopore read. The criteria for an extender to be selected was that it extend the current read coverage by at least 200 bases and that the overlap between the extender and current combined fragments account for <50% of each length. This process then continued until the entire nanopore long read was covered or there were no remaining extender candidates. The linked alignments (the permutation of 5′ extensions, seed and 3′ extensions) were then scored as the sum of their component fragment score. Combined alignments that did not cover more than 70% of the read were discarded. All combined alignments with score within 90% of the best score for a nanopore read were reported.
In cases in which there were no alignment extensions, the long-read read was classified as “without candidate”. If multiple combined alignments remained, it was classified as a multi-candidates read. When only a single combined alignment remained, it was classified as either a single high scoring pair (Single Combined, Single Fragment(SCSF)), or multiple high scoring pairs (Single Combined, Multiple Fragments (SCMF)). SCMF reads were used to capture potential structural variations. Structural variation determination was performed on SCMF where each high scoring pair fragment mapped to a single locus (SCMF-single locus (SCMFSL)), also known as a split read. SCMFs with fragments mapping to multi-loci (SCMF-multiple loci (SCMFML)) were ignored. Finally, SV determination was performed based on the order and the distance between aligned fragments of the SCMF, using sDiff and qDiff as described above.
Methods of the invention included the assignment of split reads into seven classes of SVs: inversion (INV), or segments aligned to the same chromosome but in different orientations; translocation (TLC), or segments aligned to different chromosomes; tandem duplication (TD), which included segments containing a complete duplicated region (TCD) as well as those only spanning a duplication junction (TDJ); simple insertion (INS) or deletion (DEL), or segments corresponding to the same chromosomal region in the same orientation but either flanking a sequence that does not match the reference genome or lacking an intervening sequence observed in the reference genome, respectively; and INDELs, segments indicating both INS and DEL for the same split read.
A correction method was also employed to reduce the impact of the high nanopore error rate in homopolymer sequences on SV detection. Specifically, homopolymers in all four nucleotide contexts (An, Tn, Cn and Gn) were significantly underrepresented relative to expectation.
Comparison Between Long- and Short-Read Data:
In a comparison between long-read data and short-read data, the nanopore reads (long-reads) that aligned uniquely (with possible multiple fragments) to the human genome (hg19) were extracted for genome coverage computation with BedTools (v2.25.01). Similarly, HCC1187 Illumina paired-end sequencing data (SRA accession SRX969058) was mapped to human genome with BWA-MEM (v0.7.12). The mapped reads were sampled at 2.5×, 10×, 30× and 60×. Genome coverage was computed on reads with a mapping quality of 60. For the comparison of long-read with short-read SV calls, SVs from long reads and short reads were deemed overlapping if (1) their genomic spans overlapped and (2) the ration of the larger SV length to the smaller SV length did not exceed 3. Density plots were generated from the specific SV spans.
NA12878 Nanopore Data SV Comparison.
A12878 nanopore reads were downloaded. NanoSV was used to call SVs (set N) with the parameter “−c 2” on LAST alignment. LAST alignments were generated on the basis of previously established last-train scoring parameters. Sniffles was used to call SVs (set S) with the parameters “−n−1 −s 2” on NGMLR alignments. Methods of the invention were used to call SVs (set P) on LAST alignment generated with the parameters “−C2 −K2 −r1 −q3 −a2 −b1.” All alignments were done against the human reference genome hg19. The insertions and deletions identified by the PacBio long-read data from the same genome were used as the reference call set (set R). Only SVs with length >30 bases were used for comparison. To determine the overlap between SV set X and Y, the number of SV calls x that overlapped SV call y, and the number of SV calls y that overlapped SV call x were counted. SV calls x and y were deemed overlapping if their genomic spans overlapped and the ratio of the larger SV length to the smaller SV length did not exceed 3. To determine the sensitivity of deletion calling, the overlap calculations were repeated with the required minimum read support enumerated from 2 to 20.
Phased Adjacent SVs from Multi-Breakpoint Long Reads.
Additionally, pairs of adjacent SVs determined to be present were counted in all long-reads having multiple breakpoints under the assumption that the expected count would follow the distribution of independently drawing two SVs from the population of the SVs from all the multi-breakpoint long reads. The log-likelihood was then computed.
SV Span-Size Distribution.
To explore the genomic features of the SVs found, different methods were used to ascertain the distributions of their span size. For DEL, INS and INDEL, the total numbers of SVs from each genomic bin (bin size=20 bp) were calculated. For INDELs, sDiff and qDiff were used as deletion and insertion spans, respectively. For TDC, TDJ and INV, density plots were generated from their span distributions to show their large size variations.
Genomic Distribution of SVs and Breakpoints.
Breakpoint density was computed from the numbers of breakpoints per Mb across the genome. The density of translocation pairs was calculated using the translocation breakpoint distribution in pair per Mb across the genome. A Circos plot was performed using the breakpoint densities, spans of TDC, TDSR (<20 Mb), INV and TLC pairs (counts >3).
Association Between Gene Coding and Breakpoint Density.
Gene density was computed as the fraction of bases overlapping with annotated gene regions (exons and introns; GenCode V24) in each Mb bin across the genome. Violin plots of the gene density from the top 10% breakpoint dense regions (breakpoint density >40) and the bottom 10% least dense regions (breakpoint density <9) were generated. A Mann-Whitney test was performed.
Breakpoint Landscape Analysis.
Each breakpoint was stepwise assigned to different class of genomic features based on the GENCODE v24 gene model.
Repeat Analysis.
To determine whether the inserted DNA fragments from INS and INDEL as well as TD regions contained repetitive sequences, and if so, which class of repeats, inserted or duplicated DNA sequences were extracted from their corresponding nanopore long-reads and allocated to different repeat classes by aligning them to public annotated repeat sequences using RepeatMasker-open-4-0-6 (http://www.repeatmasker.org/). A violin plot was generated with the percentages of the SV fragments annotated to repeats (hg19). The relative ratios of the most predominant repeat class, all other repeats and no repeat from each of the five SV types (TDC, deletion regions in INDEL, insertion regions in INDEL, DEL and INS) were produced in 20 bp span size intervals.
Distribution of Genomic Micro-Insertions.
Unaligned DNA sequences found between each breakpoint junction (nanopore long-reads having alignments with qDiff >20) were extracted from their corresponding nanopore reads from INDEL, TD, TLC and INV. The size distribution of the unaligned sequences was plotted for these four classes.
ICP Analysis.
“ICP” was defined as the sum of a region's inter-chromosomal contact frequencies divided by its total contact frequencies. The TCC (Tethered Conformation Capture) interaction matrix was downloaded from NCBI SRA accession SRX030110 (see Branco, M. R. & Pombo, A., “Intermingling of chromosome territories in interphase suggests role in translocations and transcription-dependent associations”, PLoS Biol. 4, e138 (2006). The ICP was then computer on the entire genome. See Howarth, K. D. et al., “Array painting reveals a high frequency of balanced translocations in breast cancer cell lines that break in cancer-relevant genes.” Oncogene 27, 3345-3359 (2008). The counts of breakpoints were partitioned into four groups from low to high. The correlation was plotted from ICPs found in different partitions.
Multidimensional Scaling of Gene Expression Analysis.
Tests were also conducted to determine and analyze the multidimensional scaling of gene expression. Structural variations with spans range from 1 Kb to 1 Mb were selected to compare their impacts on gene expressions between Triple Negative Breast Cancer (TNBC) and non-TNBC samples. There are 1,260 coding and 711 non-coding genes in the selected 537 DEL, 2,383 INDEL and 188 TDJ events (SV-genes). Their expression in 113 TNBC and 851 non-TNBC samples was retrieved from gene expression data of the breast carcinoma (BRCA) of the cancer genome atlas (TCGA) based on a previous study. See Tjong, H. et al., “Population-based 3D genome structure analysis reveals driving forces in spatial genome organization.”, Proc. Natl. Acad. Sci. USA 113, E1663-1672 (2016). Two data sets, the control set and the permutation set, were used to evaluate the significance of grouping analysis. The control data was selected from the non-SV genes of equivalent number that were expressed at similar level as the SV-genes. For the permutation data, the expressions of SV-genes were permutated sample-wise to ensure that each gene had the same expression profile with a distinct sample order. Multi-dimensional Scaling (MDS) was used to analyze the expression of SV-genes in the BRCA dataset and visualize the sample relationship.
Validation of SV Candidates.
To validate SV candidates, 3-46 structural variation events from each SV classification were selected.
In detail, for INVs, TLCs, and TDJs, the validation was done by PCR across the breakpoint junctions, and candidate SVs in these classes were selected mainly on the basis of (1) the presence of sequence-specific PCR primers on each side of the breakpoint and (2) amplicon sizes ranging from 0.3 kb to 1 kb. For INS, DEL, and TDC validation, entire SV regions were amplified. SV candidates that allowed for PCR primers designed in the non-repeat regions surrounding the SV junctions and enabled amplicon sizes in the range of 0.3-1 kb (DEL), 0.3-5 kb (INS), and 0.2-2.1 kb (TDC) were chosen. To minimize the confounding effect of micro-insertions in PCR primer design and amplicon size confirmation, only SV candidates that contained <50-bp micro-inserted sequences were selected for PCR validation. The presence of micro-inserted sequences was validated separately by PCR followed by sequencing to obtain nucleotide-resolution sequence information.
Checking Short-Read-Called SVs Against PCR-Validated SVs.
As an additional check against short-read structural variation identification and PCR-validated SVs, HCC1187 Illumina paired-end sequencing data (SRA accession SRX969058) was mapped to human genome (hg19) with BWA-MEM (v0.7.12). The mapped reads were also sampled at 2.5×, 10×, 30× and 60×. LUMPY (v0.2.13) was used to call SVs using both non-redundant split-read and discordant paired-end reads with the minimum weight for a call (−mw) as 2, 3, 5, 10, and 16 for the subsampled sets 2.5×, 10×, 30×, and 60× and the whole dataset (102×) respectively. The LUMPY generated vcf files were loaded in an IGV browser to visual check the PCR validated SVs locus for the same SV type called by LUMPY.
Sequencing of the single strand templates of the DNA fragments was performed to generate 1D data sets, while sequencing of both the template and complement strands, enabled through their covalent bridging by a hairpin adaptor, was performed to generate 2D data sets (
Compared to paired-end short reads at equivalent sequencing depth, long read data extended further into repetitive sequence regions of the genome and covered more of the genome (82% vs. 77%).
When compared with the read length distribution from the established PacBio SMRT long-read sequencing platform, using identical preparation of 12 Kb DNA templates, nanopore exhibited less bias in read length, as indicated in
The SV determination method described above was applied to 796,029 2-dimensional long-reads to detect and classify SVs in the HCC1187 genome.
As an example, a specific nanopore read of 32.5 Kb included three consecutive segments of 12, 9.5 and 11 Kb that each aligned to a different chromosome (chromosome 7, 20 and 2) with high identity (>97%).
In contrast to short read SV detection tools such as LUMPY, which assumes no more than two uniquely mapped segments and one breakpoint per split read (see Frith, M. C., Hamada, M. & Horton, P., “Parameters for accurate genome alignment”, BMC Bioinformatics 11, 80 (2010)), the long-read SV determination method disclosed herein incorporates an algorithmic rationale to interpret long reads spanning multiple breakpoints that can encompass the entire structural variation. Therefore, the long-read analysis method has sufficient sensitivity to capture small SVs, particularly short tandem duplications. For example, a 0.9-Kb tandem duplication on chr.10p12.31 was misclassified as translocation by LUMPY but uncovered by the long-read analysis methods disclosed herein from a 12.9 Kb nanopore read.
Validation tests were performed to confirm the accuracy of the long read analysis and structural variation determination method disclosed herein. Structural variations were validated via PCR across either identified breakpoint junctions (TLCs, INVs and TDJs) or full-length rearranged regions (INSs, DELs and TDCs). Overall, over 200 SV events were validated based on the presence of PCR-amplified products of the expected sizes. All (100%-40/40) of the predicted SVs that were supported by more than one long read were validated, while 136 out of 173 (79%) SVs that were supported by only a single long read were validated. Data is shown in
Multiple PCR fragments amplified from both normal and rearranged haplotypes were also observed, which suggested the presence of heterozygosity in SV-containing loci, see
Using the 176 validated SVs as a reference data set, a specificity comparison was made between long read and short read analyses. As shown, short-read sequencing analysis by LUMPY could only accurately detect a subset of SV classes (TLC, TDJ and DEL) given sufficient coverage (>30×), but drastically underperformed in classifying short-span TDs (TDC) and inversions, and was unable to unambiguously define insertion events.
To further evaluate the sensitivity of long-read analysis method, tests were performed to determine whether the long-read analysis could detect and classify the SVs identified by a previous short-read data analysis. See Gazdar, A. F. et al., “Characterization of paired tumor and non-tumor cell lines established from patients with breast cancer”, Int. J. Cancer 78, 766-774 (1998). Excluding the 27 SV regions not covered by the long read data, all the remaining 69 SVs classified by the short-read analysis method were correctly classified by the long-read analysis method as well as shown in the table of
The structural variations identified in the HCC1187 genome exhibit a broad span distribution, ranging from 20 bp up to 100 Mb.
Split read linked alignment extensions aligned across SV junctions enable the long-read analysis methods disclosed herein to characterize breakpoint junctions in their entirety at nucleotide-resolution. A total of 66,660 breakpoint junctions from different SV types were examined, and the presence of additional non-aligned DNA sequences inserted within the breakpoints was observed. While these non-aligned DNA insertions were most frequently detected within the junctions of DELs (57%) and TDs (26%), they were also found in a subset of TLCs (14%) and INVs (17%) as shown in
Applying the BLAST tool to these insertions against the NCBI non-redundant nucleotide database reveals that 90% of the inserted DNA pieces are completely novel with no homology to any known sequences, 5% overlap with known repetitive elements (mostly SINE/Alu), and the remaining 5% are derived from un-assembled human sequences, including newly emerged retrotransposed pseudogenes.
In summary, the long-read SV determination method disclosed herein uncovers large number of variants in repetitive sequence regions and precisely characterizes breakpoint sequences.
Furthermore, long-read analysis shows that the distribution of the large numbers of SVs in the HCC1187 genome reflects the highly jumbled nature of this genome.
These results suggest that high transcription activity could impact genome fragility. Interestingly, two of the hyper-density breakpoint loci (2q21.3-2q22.1; 65 breakpoints/Mb and 4q35; 139 breakpoints/Mb; highlighted in
In addition, the high frequency of SVs associated with the interconnected chromatins of high gene density further implicates linkage between genome structure and transcriptional regulation. The distribution of different SV associated-breakpoints among intergenic, coding sequence (CDS), promoter (2.5 Kb upstream of TSS), 5′ and 3′ un-translated regions (UTRs), and introns shown in
Interestingly, repeat-rich versus repeat-poor TDs exhibited contrasting distribution patterns as indicated in
SVs occurring in promoters/regulatory elements can selectively lead to oncogene activation or tumor suppressor gene inactivation. Such effects on gene expression are likely to be cancer pathway or origin specific. To test this, the genes affected by the SVs were determined and the expressions from 113 TNBC and 851 non-TNBC tissues of the breast carcinoma w(BRCA) dataset within the cancer genome atlas (TCGA) were examined. See Chung, I. F. et al., “DriverDBv2: a database for human cancer driver gene research”, Nucleic Acids Res. 44, D975-979 (2016). In contrast to the control genes and with permutation, the expression of 1,260 coding genes disrupted by the major SV classes (INDEL and TD) effectively distinguished TNBC from non-TNBC types. This is illustrated by the multidimensional scaling (MDS) plot shown in
As described above, the long-read structural variation method disclosed herein enables detection of a full spectrum of structural variants in a highly complex cancer genome model. Experimental results demonstrate that long read sequencing offers superior performance to high depth, short-read analysis in specificity for SV detection, even at a modest depth. The specificity of the long read-to-genome alignment uncovers short span SVs residing in different repetitive elements, which are highly abundant but largely understudied by existing approaches. Repetitive elements are known to provide architectural framework in genome organization and as a major source of variation inducing human diseases. See Shapiro, J. A. & von Sternberg, R, “Why repetitive DNA is essential to genome function”, Biol. Rev. Camb. Philos. Soc. 80, 227-250 (2005); Padeken, J., Zeller, P. & Gasser, S. M., “Repeat DNA in genome organization and stability”, Curr. Opin. Genet. Dev. 31, 12-19 (2015). However, the lack of knowledge of SVs localized to these repetitive regions severely limits the analysis of important structural mutations in cancer, especially those SVs in cancers with intact homologous recombination mechanisms.
The long read breakpoint analysis disclosed herein suggests the linkage between the propensity of inter-chromosomal connectivity and frequency of genomic lesions. Nuclear regions with high transcriptional activity are shown to have extensive inter-chromatin contacts and intermingling between chromatin territories can influence translocation efficiency. See Kahlor et al., Branco et al, cit. supra. Therefore, the accessibility and conformation in active chromatin domains may provide the structural basis of genome fragility. The preferential enrichment of breakpoints in the regulatory repertoire of the genome further implicates a potential mechanism in which genome rearrangement can reconfigure the transcriptional program in the cancer transformation process. Expanding the high-resolution of SV analysis throughout tumorigenesis using long read analysis can further our understanding of the homeostasis between genome architecture, variation and carcinogenesis.
Long-read sequencing possesses many unique features that can improve upon the current state of SV detection. Yield from long read sequencing techniques such as nanopore sequencing has dramatically increased over the past year with higher sequencing speed and pore stability. Furthermore, multiple strategies have been devised in parallel to improve base identification accuracy, including a new base identifier called “Scrappie”. See Jain, M. et al., “Nanopore sequencing and assembly of a human genome with ultra-long reads”, preprint at bioRxiv https://doi.org/10.1101/128835 (2017). With the continuous progress made on throughout, accuracy and analytic tools, genome-wide haplotype specific SV detection can be attempted to interrogate the diversity and complexity of cancer genome variation. Given the superior alignment specificity, higher resolution and broader utility of single-molecule long-read data, we anticipate that there will be soon a paradigm shift with respect to sequencing approaches for genome structural analysis; this shift will likely reveal new insights in the dynamics of SVs and the mechanisms of their generation during tumorigenesis.
All publications, patents, and patent applications mentioned in this specification are herein incorporated by reference. To the extent publications and patents or patent applications incorporated by reference contradict the disclosure contained in the specification, the specification is intended to supersede and/or take precedence over any such contradictory material.
It is to be understood that any functional details disclosed herein are not to be interpreted as limiting the systems and methods, but rather are provided as a representative embodiment and/or arrangement for teaching one skilled in the art one or more ways to implement the methods.
It is to be further understood that like numerals in the drawings represent like elements through the several figures, and that not all components and/or steps described and illustrated with reference to the figures are required for all embodiments or arrangements
The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used herein, the singular forms “a”, “an” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms “comprises” and/or “comprising”, when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
Also, the phraseology and terminology used herein is for the purpose of description and should not be regarded as limiting. The use of “including,” “comprising,” or “having,” “containing,” “involving,” and variations thereof herein, is meant to encompass the items listed thereafter and equivalents thereof as well as additional items.
While the invention has been described with reference to exemplary embodiments, it will be understood by those skilled in the art that various changes may be made and equivalents may be substituted for elements thereof without departing from the scope of the invention. In addition, many modifications will be appreciated by those skilled in the art to adapt a particular instrument, situation or material to the teachings of the invention without departing from the essential scope thereof. Therefore, it is intended that the invention not be limited to the particular embodiment disclosed as the best mode contemplated for carrying out this invention, but that the invention will include all embodiments falling within the scope of the appended claims.
This application claims the benefit under 35 U.S.C. § 119(e) of U.S. Provisional application Ser. No. 62/558,053 filed Sep. 13, 2017 and U.S. Provisional application 62/676,003 filed May 24, 2018, the disclosure of each of which is incorporated by reference herein in its entirety.
| Number | Date | Country | |
|---|---|---|---|
| 62676003 | May 2018 | US | |
| 62558053 | Sep 2017 | US |
