Patent Application
20200115766
This U.S. patent application claims priority under 35 U.S.C. § 119 to Indian Application No. 201821030219, filed on Aug. 10, 2018. The entire contents of the aforementioned application are incorporated herein by reference.
The embodiments herein generally relates to the field of taxonomic profiling of microbial communities. More particularly, but not specifically, the invention provides a system and method for improving amplicon sequencing based taxonomic profiling or resolution of microbial communities.
Sequencing of 16S rRNA genes is a standard protocol for taxonomic characterization of bacterial species. Sanger sequencing has been conventionally used for obtaining “full-length” 16S rRNA gene sequences of individual bacterium. Next generation sequencing (NGS) technologies have enabled probing of microbial diversity in different environmental niches with unprecedented sequencing depth. Sequencing of such regions (encompassing one or more variable regions or V-regions) has been utilized in microbiome studies for obtaining taxonomic assignments for bacterial groups present in the studied environment. However, due to read-length limitations of popular NGS technologies, 16S amplicon sequencing based microbiome studies rely on targeting short stretches of the 16S rRNA gene encompassing a selection of variable (V) regions. In most cases such a short stretch constitutes a single V-region or a couple of V-regions placed adjacent to each other on the 16S rRNA gene. Given that different V-regions have different resolving ability with respect to various taxonomic groups, selecting the optimal V-region (or a combination thereof) remains one challenge.
Furthermore, NGS technologies although enable sequencing in ultrahigh-throughput mode, they are limited with respect to read-lengths. These technologies are currently capable of yielding short sequences (referred to as reads). Due to the mentioned limitations of read-length, 16S rRNA amplicon-based microbiome studies currently rely on sequencing short stretches within the span of the 16S rRNA gene. These short stretches encompass a selection of variable (V) regions. In most cases, the said short stretch constitutes a single V-region (˜150-250 base pairs in length) or a couple of V-regions placed adjacent to each other on the 16S rRNA gene. As compared to the taxonomic resolution obtained through analysis of ‘full-length’ 16S rRNA gene sequences (˜1200-1500 base pairs in length) generated using the classical Sanger sequencing technology, short reads provide limited taxonomic resolution because the limited information embedded in such short reads makes it computationally challenging to unambiguously compare and associate them with template/reference database sequences whose taxonomy is known with certainty.
Nevertheless, the relative lower cost of NGS and the throughput they achieve make them an attractive proposition. The challenge/problem is to reduce the difference/gap in taxonomic resolution obtained with conventional sequencing methods like Sanger (longer read lengths) and NGS technologies (shorter read lengths) without increasing the cost of sequencing employing NGS technologies.
Embodiments of the present disclosure present technological improvements as solutions to one or more of the above-mentioned technical problems recognized by the inventors in conventional systems. For example, in one embodiment, a system for improving amplicon based taxonomic resolution of microbial community. The system comprises a sample collection module, an input module, a DNA extraction module, a sequence, a first microbial abundance profile generation module, a second microbial abundance profile generation module, one or more hardware processors and a memory. The memory further comprises a computation table generation module and a combined microbial abundance profile generation module. The sample collection module collects a biological sample from environment. The input module obtains a first subsample and a second subsample from the biological sample. The DNA extraction module extracts microbial DNA from the first subsample and the second subsample. The sequencer sequences the extracted microbial DNA from the first subsample to get DNA sequence data, wherein the DNA sequence data comprises of a plurality of pairs of sequence fragments, and wherein each pair of the plurality of pairs of sequence fragments is generated through paired-end sequencing of an amplicon that comprises a first combination of informative regions within the amplicon, and wherein the said informative regions contain phylogenetically relevant information. The sequencer also sequences the extracted DNA from the second subsample to get DNA sequence data, wherein the DNA sequence data comprises of a plurality of pairs of sequence fragments, wherein each pair of the plurality of pairs of sequence fragments is generated through paired-end sequencing of the amplicon that comprises a second combination of informative regions within the amplicon, wherein the second combination of informative regions are different from the first combination of informative regions, and wherein the amplicon sequencing experiment targets a phylogenetic marker gene. The first microbial abundance profile generation module generated a microbial taxonomic abundance profile of the first sequenced subsample by employing a taxonomic classification method, wherein the taxonomic classification method utilizing phylogenetically relevant information corresponding to the first combination of informative regions, wherein the microbial taxonomic abundance profile comprises of abundance values corresponding to one or more pair of sequence fragments comprising the first combination of informative regions classified into a plurality of taxonomic groups. The second microbial abundance profile generation module generates a microbial taxonomic abundance profile of the second sequenced subsample by employing the taxonomic classification method, wherein the taxonomic classification method utilizing phylogenetically relevant information corresponding to the second combination of informative regions, wherein the microbial taxonomic abundance profile comprises of abundance values corresponding to one or more pair of sequence fragments comprising the second combination of informative regions classified into the plurality of taxonomic groups. The computation table generation module pre-computes taxonomic classification accuracies for all different possible combinations of informative regions for microbes belonging to the plurality of taxonomic groups, wherein the pre-computing is based on marker gene sequences of known taxonomic origin present in existing sequence databases, to generate a computation table. The combined microbial abundance profile generation module combines the microbial taxonomic abundance profiles of the first and the second sequenced subsample based on the computation table to generate a combined microbial taxonomic abundance profile, wherein the combined microbial taxonomic abundance profile has a refined abundance value and has improved taxonomic classification accuracy as compared to the microbial taxonomic abundance profiles obtained individually for the first and the second subsample, or as compared to a microbial taxonomic abundance profile obtained for the entire biological sample or any other subsample of a biological sample using amplicon sequencing targeting any of the combinations of informative regions in the phylogenetic marker gene.
In another aspect the embodiment here provides a method for improving accuracy of taxonomic profiling of a microbial community based on amplicon sequencing. Initially, a biological sample is collected from environment. A first subsample and a second subsample is then obtained from the biological sample. In the next step, microbial DNA is extracted from the first subsample and the second subsample. Later, The extracted microbial DNA from the first subsample is sequenced using a sequencer to get DNA sequence data, wherein the DNA sequence data comprises of a plurality of pairs of sequence fragments, and wherein each pair of the plurality of pairs of sequence fragments is generated through paired-end sequencing of an amplicon that comprises a first combination of informative regions within the amplicon, and wherein the said informative regions contain phylogenetically relevant information. Similarly, the extracted DNA from the second subsample is also sequenced using the sequencer to get DNA sequence data, wherein the DNA sequence data comprises of a plurality of pairs of sequence fragments, wherein each pair of the plurality of pairs of sequence fragments is generated through paired-end sequencing of the amplicon that comprises a second combination of informative regions within the amplicon, wherein the second combination of informative regions are different from the first combination of informative regions, and wherein the amplicon sequencing experiment targets a phylogenetic marker gene. In the next step, a microbial taxonomic abundance profile of the first sequenced subsample is generated by employing a taxonomic classification method, wherein the taxonomic classification method utilizing phylogenetically relevant information corresponding to the first combination of informative regions, wherein the microbial taxonomic abundance profile comprises of abundance values corresponding to one or more pair of sequence fragments comprising the first combination of informative regions classified into a plurality of taxonomic groups. Similarly, a microbial taxonomic abundance profile of the second sequenced subsample is generated by employing the taxonomic classification method, wherein the taxonomic classification method utilizing phylogenetically relevant information corresponding to the second combination of informative regions, wherein the microbial taxonomic abundance profile comprises of abundance values corresponding to one or more pair of sequence fragments comprising the second combination of informative regions classified into the plurality of taxonomic groups. In the next step, taxonomic classification accuracies are pre-computed for all different possible combinations of informative regions for microbes belonging to the plurality of taxonomic groups, wherein the pre-computing is based on marker gene sequences of known taxonomic origin present in existing sequence databases, to generate a computation table. And finally, the microbial taxonomic abundance profiles of the first and the second sequenced subsample is combined based on the computation table to generate a combined microbial taxonomic abundance profile, wherein the combined microbial taxonomic abundance profile has a refined abundance value and has improved taxonomic classification accuracy as compared to the microbial taxonomic abundance profiles obtained individually for the first and the second subsample, or as compared to a microbial taxonomic abundance profile obtained for the entire biological sample or any other subsample of a biological sample using amplicon sequencing targeting any of the combinations of informative regions in the phylogenetic marker gene.
In another aspect the embodiment here provides one or more non-transitory machine readable information storage mediums comprising one or more instructions which when executed by one or more hardware processors cause improving accuracy of taxonomic profiling of a microbial community based on amplicon sequencing. Initially, a biological sample is collected from environment. A first subsample and a second subsample is then obtained from the biological sample. In the next step, microbial DNA is extracted from the first subsample and the second subsample. Later, The extracted microbial DNA from the first subsample is sequenced using a sequencer to get DNA sequence data, wherein the DNA sequence data comprises of a plurality of pairs of sequence fragments, and wherein each pair of the plurality of pairs of sequence fragments is generated through paired-end sequencing of an amplicon that comprises a first combination of informative regions within the amplicon, and wherein the said informative regions contain phylogenetically relevant information. Similarly, the extracted DNA from the second subsample is also sequenced using the sequencer to get DNA sequence data, wherein the DNA sequence data comprises of a plurality of pairs of sequence fragments, wherein each pair of the plurality of pairs of sequence fragments is generated through paired-end sequencing of the amplicon that comprises a second combination of informative regions within the amplicon, wherein the second combination of informative regions are different from the first combination of informative regions, and wherein the amplicon sequencing experiment targets a phylogenetic marker gene. In the next step, a microbial taxonomic abundance profile of the first sequenced subsample is generated by employing a taxonomic classification method, wherein the taxonomic classification method utilizing phylogenetically relevant information corresponding to the first combination of informative regions, wherein the microbial taxonomic abundance profile comprises of abundance values corresponding to one or more pair of sequence fragments comprising the first combination of informative regions classified into a plurality of taxonomic groups. Similarly, a microbial taxonomic abundance profile of the second sequenced subsample is generated by employing the taxonomic classification method, wherein the taxonomic classification method utilizing phylogenetically relevant information corresponding to the second combination of informative regions, wherein the microbial taxonomic abundance profile comprises of abundance values corresponding to one or more pair of sequence fragments comprising the second combination of informative regions classified into the plurality of taxonomic groups. In the next step, taxonomic classification accuracies are pre-computed for all different possible combinations of informative regions for microbes belonging to the plurality of taxonomic groups, wherein the pre-computing is based on marker gene sequences of known taxonomic origin present in existing sequence databases, to generate a computation table. And finally, the microbial taxonomic abundance profiles of the first and the second sequenced subsample is combined based on the computation table to generate a combined microbial taxonomic abundance profile, wherein the combined microbial taxonomic abundance profile has a refined abundance value and has improved taxonomic classification accuracy as compared to the microbial taxonomic abundance profiles obtained individually for the first and the second subsample, or as compared to a microbial taxonomic abundance profile obtained for the entire biological sample or any other subsample of a biological sample using amplicon sequencing targeting any of the combinations of informative regions in the phylogenetic marker gene.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.
The accompanying drawings, which are incorporated in and constitute a part of this disclosure, illustrate exemplary embodiments and, together with the description, serve to explain the disclosed principles.
Exemplary embodiments are described with reference to the accompanying drawings. In the figures, the left-most digit(s) of a reference number identifies the figure in which the reference number first appears. Wherever convenient, the same reference numbers are used throughout the drawings to refer to the same or like parts. While examples and features of disclosed principles are described herein, modifications, adaptations, and other implementations are possible without departing from the spirit and scope of the disclosed embodiments. It is intended that the following detailed description be considered as exemplary only, with the true scope and spirit being indicated by the following claims.
Referring now to the drawings, and more particularly to
According to an embodiment of the disclosure, a system 100 for improving accuracy of amplicon based taxonomic profiling of microbial community is shown in the block diagram of
In the present disclosure, the system 100 have been explained with the help of two experiments targeting two different combinations of variable regions. Though it should be appreciated that the system 100 can also be modified to involve more than two experiments to enable targeting even more combinations which might be relevant for the biological problem.
According to an embodiment of the disclosure, the system 100 is specifically using paired end sequencing. Paired-end sequencing protocols available with some of the NGS platforms allow sequencing of a stretch of DNA from both its ends. For example, Illumina HiSeq sequencing platforms can be used for paired-end sequencing to generate up to 2×250 bp reads. To this end, appropriate primers need to be designed against a desired stretch of the 16S rRNA gene, such that the targeted V-regions (either contiguously or non-contiguously placed) reside within this stretch, and are not far from either of its boundaries. Sequencing of the amplicon generated with these primers can then be performed with a paired-end sequencing protocol, whereby these (amplified) stretches of DNA are sequenced from both ends. Two reads sequenced from each such amplicon would cover the two targeted V-regions (one from each end). Since each of the sequenced reads from any given ‘pair’ targets a single V-region (situated at one of the ends of the amplicon), read-length limitations do not restrict capturing the entirety of the individual V-regions. Consequently, it becomes possible to sequence almost all possible pair wise combinations of V-regions, either arranged contiguously or non-contiguously. Paired-end sequencing protocols can therefore, in principle, be employed for sequencing various pair wise combinations of contiguous or non-contiguous V-regions in a single sequencing run.
According to an embodiment of the disclosure, the system 100 further comprises a sample collection module 102, an input module 104, a DNA extraction module 106, a sequencer 108, a first microbial abundance profile generation module 110, a second microbial abundance profile generation module 112, a memory 114 and one or more hardware processor 116 as shown in the block diagram of
According to an embodiment of the disclosure the sample collection module 102 is configured to collect a biological sample from the environment. The biological sample can be collected from various places such as gut, swab, saliva from human body or any other place outside the human body. The input module 104 is configured to obtaining a first subsample and a second subsample from the collected biological sample. In an example, the input module 104 could be same as the sample collection module 102. The sample collection module 102 and the input module 104 can include a variety of software and hardware interfaces, for example, a web interface, a graphical user interface, and the like and can facilitate multiple communications within a wide variety of networks N/W and protocol types, including wired networks, for example, LAN, cable, etc., and wireless networks, such as WLAN, cellular, or satellite.
According to an embodiment of the disclosure, the system 100 comprises the DNA extraction module 106 and the sequencer 108. The DNA extraction module 106 is configured to extract DNA fragments from the first subsample and the second subsample using laboratory standardized protocol. The sequencer 108 is configured to sequence the extracted DNA from the first subsample and the second subsample. The sequencing for the first subsample and the second subsample is performed separately and can be performed in any order.
The sequencing of the extracted microbial DNA from the first subsample is performed using a sequencer to get DNA sequence data. The DNA sequence data comprises of a plurality of pairs of sequence fragments. Each pair of the plurality of pairs of sequence fragments is generated through paired-end sequencing of an amplicon that comprises a first combination informative regions within the amplicon, wherein the informative regions contain phylogenetically relevant information. It should be appreciated that going forward in this disclosure, the informative region can also be referred as the variable regions (V-regions) specific to 16s rRNA.
The first combination of informative regions are having contiguously or non-contiguously located informative regions.
Similarly, the sequencing of the extracted DNA from the second subsample is performed using the sequencer to get DNA sequence data. The DNA sequenced data comprises of a plurality of pairs of sequence fragments, wherein each pair of the plurality of pairs of sequence fragments is generated through paired-end sequencing of the amplicon that comprises a second combination of informative regions within the amplicon. The second combination of informative regions are having contiguously or non-contiguously located informative regions. The first combination or the second combination, can both include one or more informative regions. The second combination of informative regions are different from the first combination of informative regions. The amplicon sequencing experiment targets a phylogenetic marker gene. Though it should be appreciated that there could be overlap between the informative regions of the first combination and the second combination but they can never be exactly same. Although, the first combinations and the second combination of informative regions are always expected to be different, one of the informative regions in both combinations may be shared by both the combinations.
According to an embodiment of the disclosure, the system 100 further comprises the first microbial abundance profile generation module 110 and the second microbial abundance profile generation module 112. The first microbial abundance profile generation module 110 is configured to generate the microbial taxonomic abundance profile of the first sequenced subsample by employing a taxonomic classification method, wherein the taxonomic classification method utilizing phylogenetically relevant information corresponding to the first combination of informative regions, wherein the microbial taxonomic abundance profile comprises of abundance values corresponding to one or more pair of sequence fragments comprising the first combination of informative regions classified into a plurality of taxonomic groups. Similarly, the second microbial abundance profile generation module 112 is configured to generate a microbial abundance profile of the second sequenced subsample by employing the taxonomic classification method, wherein the taxonomic classification method utilizing phylogenetically relevant information corresponding to the second combination of informative regions, wherein the microbial abundance profile comprises of abundance values corresponding to one or more pair of sequence fragments comprising the second combination of informative regions classified into the plurality of taxonomic groups.
According to an embodiment of the disclosure, the memory 114 further comprises the computation table generation module 118. The computation table generation module 118 is configured to generate a computation table. The computation table generation module pre-compute taxonomic classification accuracies for all different possible combinations of informative regions for microbes belonging to the plurality of taxonomic groups, wherein the pre-computing is based on marker gene sequences of known taxonomic origin present in existing sequence databases, to generate a computation table. The computation table generally comprises thousands of rows and various combination of variable regions in the column. A detailed methodology and rationale employed for computing the taxonomic classification accuracies is explained in the later part of the disclosure. Due to the space and the size constraint, only a part of tables are shown below.
(a) The individual V-regions (targeted in the experiments) in resolving each of the taxonomic groups under consideration is shown in TABLE 1.
(b) Pairs of (contiguously or non-contiguously located) V-regions within the 16S rRNA gene in resolving each of the taxonomic groups under consideration is shown in TABLE 2.
Abiotrophia
Acaricomes
Acetanaerobacterium
Zymophilus
According to an embodiment of the disclosure, the memory 114 further comprises the combined microbial abundance profile generation module 120. The combined microbial abundance profile generation module 120 is configured to combine the microbial taxonomic abundance profiles of the first and the second sequenced subsample based on the computation table to generate a combined microbial taxonomic abundance profile. The combined microbial taxonomic abundance profile has a refined abundance value and has improved taxonomic classification accuracy as compared to the microbial taxonomic abundance profiles obtained individually for the first and the second subsample, or as compared to a microbial taxonomic abundance profile obtained for the entire biological sample or any other subsample of a biological sample using amplicon sequencing targeting any of the combinations of informative regions in the phylogenetic marker gene.
In operation, a flowchart 200 illustrating a method for improving accuracy of amplicon based taxonomic profiling of microbial community is shown in
In the next step 212, the microbial taxonomic abundance profile of the first sequenced subsample is generated by employing a taxonomic classification method. The taxonomic classification method utilizes phylogenetically relevant information corresponding to the first combination of informative regions. The microbial taxonomic abundance profile comprises of abundance values corresponding to one or more pair of sequence fragments comprising the first combination of informative regions classified into a plurality of taxonomic groups. Similarly at step 214, the microbial abundance profile of the second sequenced subsample is generated by employing the taxonomic classification method. The taxonomic classification method utilizing phylogenetically relevant information corresponding to the second combination of informative regions. The microbial abundance profile comprises of abundance values corresponding to one or more pair of sequence fragments comprising the second combination of informative regions classified into the plurality of taxonomic groups.
At step 216, the taxonomic classification accuracies for all different possible combinations of informative regions for microbes belonging to the plurality of taxonomic groups are pre-computed. The pre-computing is based on marker gene sequences of known taxonomic origin present in existing sequence databases, to generate a computation table. At finally at step 218, the microbial abundance profiles of the first and the second sequenced subsample are combined based on the computation table to generate a combined microbial abundance profile. The combined microbial taxonomic abundance profile has a refined abundance value and has improved taxonomic classification accuracy as compared to the microbial taxonomic abundance profiles obtained individually for the first and the second subsample, or as compared to a microbial taxonomic abundance profile obtained for the entire biological sample or any other subsample of a biological sample using amplicon sequencing targeting any of the combinations of informative regions in the phylogenetic marker gene.
According to an embodiment of the disclosure, the system 100 can also be explained with the help experimental procedures and results. As mentioned earlier, the disclosure is using 16S rRNA as amplicon for the experimental procedures. Following are the steps involved in determining the combinatorial strategy for improving accuracy of amplicon based taxonomic profiling of microbial community as shown in the flowchart of
A microbial community (M) is initially considered for metagenomic profiling by two paired-end sequencing experiments (Ex and Ey). Each of these experiments can target 2 distinct V-regions (either arranged contiguously or non-contiguously on the 16S rRNA gene), using appropriate forward and reverse primers. In the current example, Ex targets the V-region combination Va+Vb, and Ey targets Vc+Vd. For example, combinations of V-regions selected in the two experiments could be V1+V4 and V2+V6 in one scenario. Based on the taxonomic resolution efficiencies of different (combinations of) V-regions, Ex and Ey will generate two different taxonomic abundance profiles Px and Py respectively, each of which constitutes of estimated abundance values (Ti) for different taxonomic groups (i)—
Px≡{T1x,T2x,T3x, . . . ,Tnx} Equation 1
Py≡{T1y,T2y,T3y, . . . ,Tny} Equation 2
Subsequently, for each of the taxonomic groups (Ti), a refined estimate of its abundance (Tixy) can be arrived at by combining the observed abundances Tix and Tiy, such that the refined abundance Tixy is relatively closer to the estimate obtained with the experiment (either of Ex or Ey) providing better classification accuracies for taxa ‘i’. Calculation of the refined estimate therefore takes into consideration the taxonomic classification accuracies of the combination of V-regions that had been used for the initial set of experiments Ex and Ey using the following equation:
Wherein Wix and Wiy are the relative accuracies in taxonomic classification for a particular taxonomic group ‘i’, obtained using the specific combination of V-regions chosen for experiments Ex and Ey respectively. These taxonomic classification accuracies can be calculated from the evaluation results obtained from the computation table generated in step 216 (methodology for computation of the values provided in these tables has been described in the later part of the disclosure), as a ratio of the correct assignments obtained for particular taxa using a specific combination of V-regions, and the total number of correct assignments obtained using the same V-region combination. For example, considering that the combination of Va+Vb was used in experiment Ex, Wix can be calculated as:
Similarly,
The denominator term representing “total correct assignments using Va+Vb” has been introduced to capture any additional specificity of the chosen Va+Vb region toward a particular taxon ‘i’ in context of the overall taxonomic classification performance of Va+Vb. Other simple ways of calculating the “relative accuracy in taxonomic classification” or weight (Wix), e.g., in a case wherein the denominator term is omitted, would also work fine when V-region combinations with decent classification accuracy are chosen.
It may be noted here, that in the experiment(s) using paired-end sequencing to capture two different V-regions from the 16S rRNA gene, the correspondence between the pairs of V-regions originating from the same 16S rRNA gene is retained. This allows joining the different V-regions together into a single DNA string (separated appropriately by ambiguous nucleotide characters) and providing the same as an input to taxonomic classification tools, such as the RDP classifier. However, for V-regions targeted in separate sequencing experiments, cross-experiment correspondence between the sequenced V-regions with respect to their origin 16S rRNA gene cannot be identified. This necessitates the indirect strategy of combining information obtained from different V-regions (or their combinations) for refining the taxonomic abundance estimates, as described above.
To avoid variations arising from experimental workflows and sample handling/preparations, it would be ideal to perform a single PCR step for amplicon generation, using different sets of primers appropriate for the chosen combinations of V-regions (Va+Vb, and Vc+Vd in the given example). However, it also needs to be mentioned here that the designed primers may have different affinities for the targeted regions on 16S rRNA genes originating from different taxonomic groups. This may again result in unequal proportions of 16S rRNA sequence fragments amplified by the different sets of primers, which would subsequently be reflected in the sequencing outcome. In such a scenario, the combination strategy needs to factor in this difference in proportions, while arriving at a refined taxonomic abundance estimate. Alternately, the experiment may target a combination of 3 V-regions (e.g. Va+Vb and Va+Vc or, Va+Vc and Vb+Vc), such that, either the forward primers or the reverse primers be common to the targeted combinations. This way, some equivalence in the proportions of fragments (targeting different taxonomic groups) can be maintained on account of the shared primer (for V-region) selected.
Further, it should be appreciated that if required the user can also obtain more than 2 subsamples, each targeting different V-regions. For example, there are 4 experiments—Ea, Eb, Ec, Ed . . . then the combinatorial formula can be written as—
Wherein the values for W and T can be calculated as mentioned earlier
Evaluation Results with Novel Combinatorial Strategy
Considering the fact that human gut is one of the most diverse and densely populated reservoir of microbes, the utility of the combinatorial strategy was assessed with a simulated metagenomic sample (namely GUT1—method of generating the same has been described in the later part of the disclosure) that was specifically generated for this purpose (along with 7 more simulated metagenomes pertaining to different human body sites). The taxonomic classification efficiency of the V-region combinations (at the species level) was assessed on the simulated metagenome GUT1. The V-region combinations V1+V4 and V1+V5 provided highest average classification accuracies for most of the host (human) associated environmental niches along with the simulated metagenome GUT1 as shown in
Results in Table 3 indicate that although the V1+V4 and V1+V5 regions can classify the reads with commendable accuracy, the abundance values provided for individual genera deviates from the actual (RDP) lineage by a certain extent. The combinatorial approach was observed to moderate these deviations to a significant extent, and relative abundance of individual genera ascertained by the combinatorial approach exhibited better coherence with the actual lineage. In quantitative terms, while the average deviations (from actual lineage) in relative taxonomic abundance predictions for V1+V4 and V1+V5 combination based approaches were 17.4% and 11.5% respectively, the combinatorial approach exhibited a significantly lower average deviation (6.9%) from the actual lineage. Similar improvements were also observed when this approach was tested on microbiomes pertaining to other host-associated/environmental. Given that the proposed combinatorial approach does not incur any significant additional sequencing cost and is a simple in silico extrapolation of the results obtained with standard pair-end sequencing, adoption of the same would be easy and would enable researchers to explore the taxonomic diversity of different environments with greater accuracy. While certain additional experimental costs for primers, multiplexing barcodes, additional PCR, and handling etc. are expected to be incurred to implement the proposed combinatorial strategy, the actual sequencing (reagents) cost, constituting the bulk of the total expenditure, remains the same. The additional pre-processing and handling efforts can at most be twice compared to the sample handling efforts needed for a single paired-end sequencing experiment. However, the potential benefits in terms of an improved taxonomic resolution are expected to outweigh any inhibitions arising due to the additional, but trivial, pre-processing and handling efforts.
Faecalibacteriumprausnitzii
Bacteroides faecis
Prevotellaamnii
Prevotellani
rescens
Me
amonashyperme
ale
Bacteroides pyo
enes
Bacteroides fine
oldii
Alistipesputredinis
Roseburia hominis
Bacteroides nordii
Bacteroides e
erthii
Bacteroides helco
enes
Bacteroides caccae
Bacteroides massiliensis
Bacteroides coprocola
Bacteroides salyersiae
Bacteroides stercoris
Bacteroides uniformis
Bacteroides acidifaciens
Proteiniphilumacetati
enes
Bacteroides cellulosilyticus
Bacteroides intestinalis
Roseburiafaecis
Roseburia intestinalis
Parasutterellasecunda
Roseburiainulinivorans
Phascolarctobacteriumsuccinat
Parabacteroides distasonis
Parabacteroides merdae
Parasutterellaexcrementihomin
Dorealon
icatena
Phascolarctobacterium faecium
Blautiaproducta
Escherichia/Shi
ella fer
usonii
Escherichia/Shi
ella albertii
Escherichia/Shi
ella flexneri
Escherichia/Shi
ella
Dialisterinvisus
Me
asphaeraelsdenii
Blautia
lucerasea
Blautiahydro
enotrophica
Blautiaschinkii
Mitsuokella
alaludinii
Collinsellaaerofaciens
Bifidobacterium lon
um
Bifidobacterium animalis
Ruminococcusflavefaciens
Blautiahansenii
Me
asphaera sp. NMBHI-10
Klebsiella pneumoniae
indicates data missing or illegible when filed
Similarly, Table 4, Table 5, Table 6, Table 7, Table 8, Table 9, Table 10 and Table 11 show the results of other simulated microbiome of GUT2, Sputum (oral), sub-gingival (oral), skin, soil, aquatic, vagina and nematode gut respectively. The same has also been shown in
Bacteroides faecis (T)
Alistipes putredinis (T)
Faecalibacterium prausnitzii
Bacteroides pyo
enes (T)
Bacteroides fine
oldii (T)
Parabacteroides merdae (T)
Parabacteroides distasonis (
Oscillibacter valerici
enes (T
Bacteroides acidifaciens (T)
Bacteroides salyersiae (T)
Bacteroides coprocola (T)
Bacteroides massiliensis (T)
Bacteroides intestinalis (T)
Bacteroides uniformis (T)
Bacteroides stercoris (T)
Bacteroides cellulosilyticus (
Bacteroides e
erthii (T)
Bacteroides caccae (T)
Proteiniphilum acetat
enes (
Bacteroides helco
enes (T)
Bacteroides nordii (T
Ruminococcus flavefaciens (
Ruminococcus albus (T)
Roseburia hominis (T)
Odoribacter laneus (T)
Roseburia intestinalis (T)
Parasutterella secunda (T)
Phascolarctobacterium succi
Roseburia faecis (T)
Dialister invisus (T)
Phascolarctobacterium faeciu
Prevotella amnii (T)
Prevotella ni
rescens (T)
Roseburia inulinivorans (T)
Flavonifractor plautii (T)
Blautia producta (T)
Coprococcus catus (T)
Parasutterella excrementiho
Dialister pneumosintes (T)
Dorea lon
icatena (T)
Ruminococcus bromii (T)
Blautia hydro
enotrophica (T
Coprococcus eutactus (T)
Blautia
lucerasea (T)
Blautia schinkii (T)
Blautia wexlerae (T)
Butyriccoccus pullicaecorum
Butyricmonas syner
istica (
Blautia hansenii (T)
Ruminococcus faecis (T)
indicates data missing or illegible when filed
Prevotella amnii (T)
Prevotella ni
rescens (T)
Streptococcus salivarius (T)
Streptococcus suis (T)
Granulicatella adiacens (T)
Fusobacterium nucleatum (T
Streptococcus a
alactiae (T)
Staphylococcus aureus (T)
Streptococcus pyo
enes (T)
Neisseria menin
itidis
Streptococcus
allolyticus (T
Rothia dentocariosa (T)
Streptococcus dys
alactiae (
Veillonella parvula (T)
Propionibacterium acnes (T)
Rothia aeria (T)
Veillonella tobetsuensis (T)
Streptococcus pneumoniae (
Rothia mucila
inosa (T)
Haemophilus ae
uytius (T)
Gemella san
uinis (T)
Gemella haemolysans (T)
Actinomyces neuii (T)
Gemella ber
eri (T)
Gemella morbillorum (T)
Dolosi
ranulum pi
rum (T)
Veillonella criceti (T)
Streptococcus e
ui (T)
Streptococcus infantarius (T)
Pelomonas saccharophila (T
Rothia endophytica
Staphylococcus warneri (T)
Acinetobacter baumannii (T)
Veillonella atypica (T)
Rothia amarae (T)
Veillonella denticariosi (T)
Veillonella ratti (T)
Anoxybacillus rupiensis (T)
Actinomyces coleocanis (T)
Staphylococcus cohnii (T)
Streptococcus constellatus (
Peptostreptococcus russellii
Solobacterium moorei (T)
Staphylococcus hominis (T)
Parvimonas micra (T)
Stenotrophomonas rhizophila
Streptococcus iniae (T)
Gemella palaticanis (T)
Peptostreptococcus anaerobi
Streptococcus pseudoporcinu
indicates data missing or illegible when filed
Fusobacterium nucleatum (T
Prevotella ni
rescens (T)
Prevotella amnii (T)
Parvimonas micra (T)
Streptococcus salivarius (T)
Streptococcus suis (T)
Streptococcus a
alactiae (T)
Streptococcus pyo
enes (T)
Capnocytopha
a canimorsus
Granulicatella adiacens (T)
Porphyromonas crevioricanis
Campylobacter lari (T)
Treponema maltophilum (T)
Acinetobacter baumannii (T)
Streptococcus
allolyticus (T
Fusobacterium necrophorum
Streptococcus dys
alactiae (
Neisseria menin
itidis
Leptotrichia buccalis (T)
Porphyromonas somerae (T)
Enhydrobacter aerosaccus (
A
re
atibacter aphrophilus
Actinomyces neuii (T)
Filifactor villosus (T)
Fusobacterium varium (T)
Reyranella massiliensis
Streptococcus pneumoniae (
Treponema lecithinolyticum (
Veillonella parvula (T)
Streptococcus e
ui (T)
Treponema amylovorum (T)
Veillonella tobetsuensis (T)
Fusobacterium mortiferum (T
Treponema socranskii (T)
Streptococcus infantarius (T)
Porphyromonas asaccharolyti
Leptotrichia wadei (T)
Porphyromonas endodontalis
Acinetobacter calcoaceticus
Leptotrichia
oodfellowii (T)
Veillonella criceti (T)
Actinomyces coleocanis (T)
Porphyromonas
ulae (T)
Sphin
obacterium spiritivoru
Catonella morbi (T)
Porphyromonas cansulci (T)
Rothia aeria (T)
Leptotrichia hofstadii (T)
Acinetobacter lwoffii (T)
Capnocytopha
a cynode
mi
indicates data missing or illegible when filed
Staphylococcus aureus (T)
Propionibacterium acnes (T)
Staphylococcus warneri (T)
Methylobacterium populi (T)
Cupriavidus taiwanensis (T)
Schle
elella thermodepolyme
Staphylococcus cohnii (T)
Staphylococcus hominis (T)
Cupriavidus basilensis (T)
Uruburuella suis (T)
Corynebacterium diphtheriae
Corynebacterium
lutamicum
Cupriavidus respiraculi (T)
Staphylococcus sciuri (T)
Micrococcus yunnanensis (T
Methylobacterium koma
atae
Streptococcus salivarius (T)
Methylobacterium
oesin
en
Dermacoccus nishinomiyaen
Corynebacterium bovis (T)
Staphylococcus e
uorum (T)
Streptococcus suis (T)
Methylobacterium hispanicu
Fine
oldia ma
na (T)
Schle
elella a
uatica (T)
Ralstonia syzy
ii
Cupriavidus pauculus (T)
Staphylococcus capitis (T)
Geobacillus stearothermophil
Wautersia numazuensis (T)
Methylobacterium mesophilic
Staphylococcus pasteuri (T)
Cupriavidus campinensis (T)
Staphylococcus carnosus (T)
Cupriavidus alkaliphilus (T)
Methylobacterium rhodesianu
Methylobacterium marchantia
Lactobacillus plantarum (T)
Corynebacterium ulcerans (T
Propionibacterium acidipropio
Propionibacterium acidifacien
Staphylococcus succinus (T)
Propionibacterium freudenrei
Geobacillus thermodenitrifica
Cupriavidus sp. ASC-64
Geobacillus thermoleovorans
Methylobacterium brachiatum
Stenotrophomonas rhizophila
Reyranella massiliensis
Streptococcus pyo
enes (T)
indicates data missing or illegible when filed
Bradyrhizobium pachyrhizi (
Rhodomicrobium vannielii
Gemmata obscuri
lobus (T)
Gemmatimonas aurantiaca (
Ktedonobacter racemifer (T)
Bradyrhizobium diazoefficien
Bradyrhizobium
aponicum (
Bradyrhizobium liaonin
ense
A
uisphaera
iovannonii (T)
Gaiella occulta (T)
Mycobacterium leprae
Bradyrhizobium canariense (
Phenylobacterium muchan
p
Bradyrhizobium sp. OO99
Bradyrhizobium rifense
Burkholderia fun
orum (T)
Phenylobacterium composti (
Bradyrhizobium sp. LMTR 2
Pedomicrobium ferru
ineum
Pedomicrobium australicum
Pedomicrobium man
anicum
Massilia aurea (T)
Thermoleophilum album (T)
Domibacillus robi
inosus (T)
Acidisoma tundrae (T)
Domibacillus sp. NIO-1016
Acidisoma sibiricum (T)
Dyella
aponica (T)
Opitutus terrae (T)
Bradyrhizobium iriomotense
Tumebacillus
insen
isoli (T
Burkholderia phenoliruptrix (
Burkholderia unamae (T)
Burkholderia phytofirmans (T
Pedomicrobium americanum
Burkholderia bannensis
Bradyrhizobium denitrificans
Rhodopila
lobiformis (T)
Sinomonas atrocyanea (T)
Burkholderia tuberum (T)
Burkholderia mimosarum (T)
Microvir
a sp. BR3299
Vampirovibrio chlorellavorus
Burkholderia sediminicola (T
Le
ionella pneumophila (T)
Burkholderia udeis
Chromobacterium vaccinii (T
Se
etibacter koreensis (T)
Phenylobacterium falsum (T)
Phenylobacterium immobile (
indicates data missing or illegible when filed
Polynucleobacter necessariu
Polynucleobacter cosmopolit
Mycobacterium leprae
Luteolibacter al
ae (T)
Rhodoferax saidenbachensis
Polynucleobacter acidiphobu
Acidovorax delafieldii (T)
Rhodoferax antarcticus (T)
Acidovorax temperans (T)
Methylophilus methylotrophu
Rhodoferax fermentans (T)
Opitutus terrae (T)
Luteolibacter pohnpeiensis (
Haliscomenobacter hydrossis
Acidovorax cattleyae (T)
Mycobacterium iranicum (T)
Mycobacterium novocastrens
Mycobacterium marinum (T)
Methylomonas methanica (T
Mycobacterium tuberculosis
Microbacterium paraoxydans
Al
oripha
us namhaensis
Polynucleobacter rarus (T)
Mycobacterium cookii (T)
Acidovorax caeni (T)
Mycobacterium arupense (T)
Flavobacterium de
erlachei (
Methylomonas koyamae (T)
Acidovorax avenae (T)
Methylophilus leisin
eri (T)
Rhodomicrobium vannielii
Fluviicola taffensis
Comamonas testosteroni (T)
Bei
erinckia indica (T)
Al
oripha
us antarcticus (T)
Acidovorax radicis (T)
Methylocystis rosea (T)
Methylomonas scandinavica
Methylophilus flavus (T)
Stenotrophomonas rhizophila
Methylocystis hirsuta (T)
Comamonas
ian
duensis (T
Al
oripha
us halophilus (T)
Al
oripha
us lutimaris (T)
Verrucomicrobium spinosum
Acidovorax kon
aci (T)
A
uisphaera
iovannonii (T)
Belnapia moabensis (T)
Caulobacter henricii (T)
Prosthecobacter vanneerveni
indicates data missing or illegible when filed
Lactobacillus plantarum (T)
Lactobacillus paracasei (T)
Lactobacillus fermentum
Lactobacillus delbrueckii (T)
Prevotella amnii (T)
Prevotella ni
rescens (T)
Sneathia san
uine
ens (T)
Atopobium rimae (T)
Lactobacillus reuteri (T)
Lactobacillus diolivorans (T)
Lactobacillus farra
inis (T)
Lactobacillus sakei (T)
Lactobacillus amylovorus (T)
Lactobacillus kimchii (T)
Lactobacillus
asseri (T)
Atopobium minutum (T)
Lactobacillus kefiri (T)
Lactobacillus futsaii
Lactobacillus kefiranofaciens
Lactobacillus farciminis
Fine
oldia ma
na (T)
Lactobacillus buchneri (T)
Parvimonas micra (T)
Lactobacillus mucosae (T)
Lactobacillus animalis (T)
Lactobacillus parabuchneri (
Lactobacillus florum (T)
Lactobacillus kunkeei (T)
Dialister invisus (T)
Streptococcus salivarius (T)
Lactobacillus coryniformis (T
Aerococcus viridans
Lactobacillus vaccinostercus
Lactobacillus in
luviei (T)
Anaerococcus murdochii (T)
Lactobacillus helveticus (T)
Anaerococcus va
inalis (T)
Streptococcus suis (T)
Lactobacillus paracollinoides
Dialister pneumosintes (T)
Lactobacillus va
inalis (T)
Lactobacillus oeni (T)
Mobiluncus curtisii (T)
Lactobacillus crustorum (T)
Lactobacillus rossiae (T)
Ureaplasma urealyticum (T)
Lactobacillus harbinensis (T)
Lactobacillus acetotolerans (
Streptococcus a
alactiae (T)
Lactobacillus sunkii (T)
indicates data missing or illegible when filed
Acinetobacter baumannii (T)
Cellvibrio vul
aris (T)
Cellvibrio
aponicus (T)
Cellvibrio fibrivorans (T)
Reyranella massiliensis
Pseudoalteromonas tetraodo
Enhydrobacter aerosaccus (
Acinetobacter calcoaceticus
Cellvibrio mixtus (T)
Cellvibrio sp. E50
Marinomonas primoryensis (
Acinetobacter lwoffii (T)
Escherichia/Shi
ella flexneri
Escherichia/Shi
ella fer
uson
Escherichia/Shi
ella dysenter
Staphylococcus aureus (T)
Escherichia/Shi
ella albertii (
Alkanindi
es illinoisensis (T)
Pseudomonas aeru
inosa (
Arcobacter butzleri (T)
Cellvibrio fulvus (T)
Propionibacterium acnes (T)
Cellvibrio
andavensis (T)
Cellvibrio ostraviensis (T)
Marinomonas arctica (T)
Oceanospirillum maris (T)
Peredibacter starrii (T)
Pseudoalteromonas arctica (
Delftia lacustris (T)
Oleispira antarctica (T)
Oceanospirillum bei
erinckii (
Acinetobacter
unii (T)
Listonella an
uillarum (T)
Brevundimonas nae
an
sane
Pseudoalteromonas shioyasa
Oceanospirillum linum
Vibrio cholerae (T)
Brevundimonas diminuta (T)
Persicirhabdus sediminis (T)
Leucobacter chromiiresistens
Acinetobacter radioresistens
Brevundimonas terrae (T)
Vibrio rotiferianus (T)
Acinetobacter
uillouiae (T)
Leucobacter tardus (T)
Brevundimonas bullata (T)
Leucobacter koma
atae (T)
Microbacteriaceae bacterium
Acinetobacter nectaris (T)
Brevundimonas intermedia (
indicates data missing or illegible when filed
Method for Generation of the Computation Table
Rationale and methodology employed for pre-computing/pre-estimating the accuracies of
As a one-time procedure, the following steps were performed for pre-computing the discriminating capability i.e. the accuracy of different V-regions (or combinations of various possible pairs of the same) with respect to different taxonomic lineages. The pre-generated set of accuracy values are required for solving Equation 3.
Rationale and Procedure
Full length bacterial 16S rRNA gene sequences (along with their annotated lineages) present in the RDP database (release 11.3) were retrieved. The RDP hierarchy browser was used for this purpose with the following filters—Strain=‘Both’; Source=‘Isolates’; Size ‘>=1200’; Taxonomy=‘NCBI’; Quality=‘Good’, which resulted in a downloaded set of 232,163 sequences. Further, sequences not containing any of the nine V-regions (V1-V9) were filtered out from the set of sequences, leaving a total of 84,711 16S rRNA sequences belonging to 11,810 species. Subsequently, both full-length as well as different portions of the 16S rRNA gene sequences were extracted in silico to represent outcomes of amplicon sequencing experiments, and were provided as input to the Wang classifier (algorithm used in RDP classifier), for taxonomic classification. The current version of RDP classifier 16S training set was used as the reference database for these taxonomic assignment steps, and the taxonomic hierarchy information of the reference sequences were appropriately used while training the Wang classifier in order to enable obtaining taxonomic classifications resolved up to species level. Only a subset (57,632 sequences) of the originally downloaded full-length 16S rRNA gene sequences, which could be classified at species level with >=80% bootstrap confidence threshold, was later used as a pool for randomly drawing sequences during creation of mock/simulated metagenomic datasets (as described later in this section).
For evaluating the discriminating ability of individual V-regions and their combinations, the regions of interest were parsed out from corresponding full-length 16S rRNA gene sequences using an in house modified version of the V-xtractor program, and submitted as query sequences to the Wang classifier, after appropriate pre-processing. First, the effectiveness of individual V-regions in resolving between different taxonomic groups was evaluated. For this purpose, different V-regions from all the 16S rRNA gene sequences, downloaded from the RDP database, were extracted. Subsequently, each of these individual V-regions were subjected to taxonomic classification with the Wang classifier, and the resultant assignments at the genus level were checked for accuracy and specificity against the taxonomic attributes provided by RDP for the corresponding full-length sequences.
The taxonomic classification accuracies of different V-regions in resolving different taxonomic groups are depicted in
Except for V1, V5 and V9, all other V-regions were observed to have certain utility in taxonomic classification, even when targeted individually. It was also evident from the plot that some V-regions provide comparatively higher taxonomic classification accuracies of classification for specific taxonomic groups. For example, the V4 region has the highest accuracy while classifying sequences pertaining to the phylum Bacteroidetes (75.9%), whereas the V2 region classifies best with respect to the phylum Firmicutes (68.2%). A detailed list of taxonomic classification accuracies in taxonomic classification obtained with different V-regions at genus level is also calculated and collated in a table (not provided in the disclosure due to large size).
Given these observations, it would seem logical for a microbiome study design to sequence two (or more) V-regions from a 16S rRNA gene fragment which have complementary abilities with respect to classification of different taxonomic groups. Furthermore, the choice of the combination of V-regions could also be guided by the environment from where the metagenomic sample is being collected, given that diverse environments may be differentially enriched with different taxonomic groups. A preferred combination of V-regions cannot always be expected to be situated in a contiguous stretch on the 16S rRNA gene. Given the read length limitations of NGS technologies, targeting an amplicon constituting the preferred regions becomes difficult in reality.
The length distributions of V-regions and C-regions (constant/conserved regions flanking the V-regions) across different bacterial taxonomic groups are provided in
Paired-end sequencing protocols available with some of the NGS platforms allow sequencing of a stretch of DNA from both its ends. For example, Illumina HiSeq sequencing platforms can be used for paired-end sequencing to generate up to 2×250 bp reads. The current work proposes, and evaluates in silico, the utilization of paired-end sequencing protocols for sequencing various pair wise combinations of non-contiguous V-regions in a single sequencing run. To this end, appropriate primers need to be designed against a desired stretch of the 16S rRNA gene, such that the targeted V-regions (either contiguously or non-contiguously placed) reside within this stretch, and are not far from either of its boundaries. Sequencing of the amplicons generated with these primers can then be performed with a paired-end sequencing protocol, whereby these (amplified) stretches of DNA are sequenced from both ends. Two reads sequenced from each such amplicon would cover the two targeted V-regions (one from each end). Since each of the sequenced reads from any given ‘pair’ targets a single V-region (situated at one of the ends of the amplicon), read-length limitations do not restrict capturing the entirety of the individual V-regions. Consequently, it becomes possible to sequence almost all possible pair wise combinations of V-regions, either arranged contiguously or non-contiguously. The results pertaining to the in silico evaluation of the effectiveness of different combinations of V-regions (see Methods), in providing accurate taxonomic classifications (at the species level) for sequences listed in the RDP database, is depicted in
Taxonomic classification accuracies provided by several combinations of non-contiguously placed V-region pairs, namely V1+V3 (77.7%), V1+V4 (77.4%), V1+V8 (76.6%), V2+V5 (73.6%), etc., were sufficiently high, and exceeded the taxonomic classification accuracies provided by combination of adjacently placed V-regions by a fair margin. It was also significant to note that many of the individual V-regions, which had very low taxonomic discriminating ability of their own (
It may be noted in this context, that reads generated during amplicon sequencing may often encompass flanking ‘constant’ regions in addition to the targeted V-region(s), depending on choice of primers and the maximum read-length attainable by the sequencing technology. Consequently, our evaluation exercise, pertaining to combination of V-regions, aimed at mimicking 250 bp×2 paired-end sequencing, wherein the extracted regions (representing sequenced reads) also encompass such flanking regions. To achieve this, regions from the full length 16S rRNA genes were extracted in such a way that either of the 250 bp reads (constituting a read-pair) contained one of the target V-regions, flanked in both directions by equivalent portions (lengths) of the surrounding ‘constant’ regions. HMMs corresponding to constant regions surrounding the V-regions, as provided by the V-xtractor program were used for this purpose. In case two adjacent V-regions were targeted, there was a significant chance of finding an overlap between two reads constituting a pair. This overlap was utilized to join the pair of reads together into a single sequence before submitting the same as a query to the Wang classifier. In contrast, on sequencing two distantly separated non-contiguous V-regions, no overlap between the read pairs could be expected. Accordingly, the pair of reads in this case were concatenated using a string of eight consecutive ‘N’s, while preserving their orientation, prior to processing with Wang classifier. Given that Wang classifier (or RDP classifier) utilizes 8-mer nucleotide frequencies during taxonomic assignment, joining two non-overlapping sequenced fragments with 8 ambiguous nucleotides (N) ensures avoiding generation of spurious 8-mers consisting nucleotides from non-adjacent regions of the gene. Taxonomic assignments generated by the Wang classifier at a predetermined taxonomic level with a confidence threshold score of >=80% were used for all downstream comparative analyses.
The utility of all possible pair wise combinations of V regions, either arranged contiguously or non-contiguously, were investigated in silico in terms of accuracy of taxonomic classifications provided by each such combination. As mentioned earlier, sequence fragments mimicking outcomes of 250 bp×2 paired-end sequencing, which target different contiguous/non-contiguous combinations of V-regions, were derived from the downloaded 16S rRNA gene sequences. These fragments were subsequently subjected to taxonomic classification with the Wang classifier and the assignments obtained at species level were checked for accuracy and specificity against the pre-annotated taxonomic attributes of their source (full-length) 16S rRNA genes.
To assess the utility of the proposed non-contiguous combination of V-regions on a microbiome dataset, while avoiding any bias arising out of the proportion of sequences pertaining to different bacterial groups currently catalogued in reference databases like RDP, taxonomic classification exercises were further performed with mock and simulated metagenomic datasets.
Each of the mock microbiome datasets were constructed using 10,000 randomly selected 16S rRNA gene sequences from one of the five randomized 16S gene pools. Each of these gene pools consisted of sequences downloaded from the RDP database, wherein the proportion of sequences selected from different organisms were also randomized (see Methods). The results, in terms of classification accuracy at the species level, are depicted in Table 12. It was interesting to note that 18 out of the 20 combinations of V-regions, which could provide classification accuracy >=60% on average, constituted of non-contiguous V-regions. The best performing combination of adjacent V-regions was V2-V3, which on average provided 69.1% classification accuracy. In comparison, the combination of the non-contiguously placed V-regions V1+V4 demonstrated a high average classification accuracy of 77.2%.
The specific combinations of V-regions, which provided comparatively higher accuracies of taxonomic classification with the RDP database sequences, were made subject to this further evaluation wherein 5 mock 16S metagenomic gene pools were created from randomly selected sets of 50 organisms (genus) listed in RDP database (Table 13).
Acetobacterium
Achromobacter
Acidiphilium
Acidithiobacillus
Acidovorax
Acinetobacter
Actinobacillus
Actinomadura
Aggregatibacter
Agromyces
Alcaligenes
Alcanivorax
Alicyclobacillus
Alkalibacterium
Alteromonas
Arcobacter
Arthrobacter
Asaia
Azoarcus
Azospirillum
Bacillus
Bifidobacterium
Borrelia
Bosea
Brachybacterium
Bradyrhizobium
Brevibacillus
Brevundimonas
Brucella
Buchnera
Burkholderia
Butyrivibrio
Campylobacter
Carnobacterium
Caulobacter
Cellulomonas
Chromobacterium
Chromohalobacter
Chryseobacterium
Citrobacter
Colwellia
Comamonas
Corallococcus
Corynebacterium
Cronobacter
Curtobacterium
Deinococcus
Delftia
Desulfosporosinus
Desulfotomaculum
Edwardsiella
Enterococcus
Erythrobacter
Eubacterium
Exiguobacterium
Flavobacterium
Francisella
Fusobacterium
Gallibacterium
Geobacillus
Glaciecola
Gluconobacter
Haemophilus
Halobacillus
Halomonas
Helicobacter
Herbaspirillum
Hydrogenophaga
Idiomarina
Kitasatospora
Klebsiella
Kocuria
Komagataeibacter
Lactobacillus
Lactococcus
Legionella
Leifsonia
Leptospira
Leucobacter
Leuconostoc
Listeria
Loktanella
Lysinibacillus
Lysobacter
Marinobacter
Marinobacterium
Marinomonas
Massilia
Methylobacterium
Microbispora
Micromonospora
Moraxella
Moritella
Mycoplasma
Neisseria
Nitrosomonas
Nocardioides
Novosphingobium
Oceanobacillus
Paenibacillus
Pandoraea
Pantoea
Paracoccus
Pectobacterium
Pediococcus
Photobacterium
Photorhabdus
Phyllobacterium
Planococcus
Polaribacter
Polynucleobacter
Proteus
Pseudomonas
Pseudoxanthomonas
Psychrobacter
Rahnella
Ralstonia
Rhizobium
Rhodopirellula
Rickettsia
Ruegeria
Ruminococcus
Salmonella
Selenomonas
Serratia
Shewanella
Sorangium
Sphingobium
Spiroplasma
Sporolactobacillus
Staphylococcus
Stenotrophomonas
Streptococcus
Streptomyces
Streptosporangium
Taylorella
Thalassospira
Thermoanaerobacter
Thermoanaerobacterium
Thermus
Thiomonas
Trueperella
Vibrio
Virgibacillus
Weissella
Xanthomonas
Xenorhabdus
Xylella
To obtain reads for building the mock metagenomic datasets corresponding to these pools, each time 10,000 16S rRNA genes were drawn randomly from a gene pool, such that the proportion of 16S rRNA genes drawn from any of the organisms are also randomized. 5 such datasets (with 10,000 reads each) corresponding to each of the 5 gene pools (a total of 25 mock datasets) were constructed for comparative evaluation. Different contiguous as well as non-contiguous combinations of V-regions were subsequently extracted from each of the 16S rRNA genes belonging to these mock datasets, and subjected to taxonomic analysis using Wang classifier, following the classification methodology described above. Taxonomic abundance values (obtained using different combinations of V-regions) were averaged over 5 mock datasets pertaining to the same gene pool. The averaged abundance values for each of the mock gene pools were compared against each other and the pre-annotated taxonomic attributes of their source (full-length) 16S rRNA genes, to assess the utility of the chosen combinations of V-regions. The results, in terms of classification accuracy at the species level, are depicted in Table 12. It was interesting to note that 18 out of the 20 combinations of V-regions, which could provide classification accuracy >=60% on average, constituted of non-contiguous V-regions. The best performing combination of adjacent V-regions was V2-V3, which on average provided 69.1% classification accuracy. In comparison, the combination of the non-contiguously placed V-regions V1+V4 demonstrated a high average classification accuracy of 77.2%.
Nine more simulated microbiomes mimicking different environmental and host associated niches—namely, gut, skin, vaginal, sub-gingival (oral), sputum (oral), nematode gut, soil, and aquatic were also generated. Taxonomic abundance estimates for eight of these environmental microbiomes were derived from datasets used in an earlier in silico study evaluating functional potential of diverse metagenomes Taxonomic abundance estimates for the aquatic microbiome was derived from a recent study. To populate these simulated microbiomes, sequences from RDP database were randomly drawn, while making sure that the proportions of 16S rRNA genes drawn from different genera were roughly similar to the proportions observed earlier for these environments. The taxonomic classification efficiency of the V-region combinations (at the species level) was also assessed on this set of simulated microbiome. The efficiency of the proposed non-contiguous combination of V-regions was further tested on nine additional simulated metagenomes mimicking different environmental and host associated niches as shown in Table 14. The data in the Table 14 have been collected from various sources in the art.
Results pertaining to the human associated simulated metagenomes, namely, gut, skin, vaginal, subgingival (oral) and sputum (oral) are depicted in
The combination of V1+V4 regions provided the maximum accuracy of classification for skin (60.2%) and one of the gut (86.0%) metagenomes (GUT2), whereas metagenomes pertaining to vaginal and sub-gingival niches were best resolved by the combination V1+V9 (with accuracies of 83.3% and 78.6% respectively). Optimal classification of sputum metagenomic samples (72.1%) could be obtained by another non-contiguous combination, viz. V1+V5 regions, which could also provide relatively more accurate classification for the GUT1 metagenome (82.5%). These results further reiterate the need of choosing an optimal combination of V-regions, preferably non-contiguous, for a specific sampled environment.
It was also noted that the high variability in taxonomic classification accuracies of individual V-region combinations while classifying samples pertaining to different environments. For example, while the combination V2+V4 could classify one of the gut microbiomes (Gut2) with 85.93% accuracy, the classification results were not as high when the same combination was used to classify the aquatic microbiome (69.2%). On the other hand, the combination V2+V7 was observed to provide decent classification for the simulated aquatic microbiome (72.8%), while performing not so well for the simulated gut microbiome datasets (65.8% for Gut1 and 70.9% for Gut2). These results further reiterate the need of choosing an optimal combination of V-regions, preferably non-contiguous, for a specific sampled environment.
It may be noted here that the paired-end reads generated for in silico evaluation of the utility of different combinations of V-regions were based on HMMs pertaining to the flanking constant regions, as provided by the V-xtractor program. Actual primer design may not always allow generation of reads identical to the in silico experiment, and results from a sequencing experiment may slightly vary from the validation results presented. A comparison of the paired-end reads generated in the in silico experiments with respect to those which may be obtained by using different sets of primers currently available for 16S rRNA amplicon sequencing as shown in
It may be mentioned here that assessment of primer specificity on all sequences from RDP database (a total of 232,163 sequences having length >=1,200 bp) revealed that the combinations/pairs (either contiguous or non-contiguous) involving the V1-region could potentially amplify a lower fraction of sequences compared to other combinations. Apparently, the fraction of sequences that can be amplified by the said combinations is limited by the specificity/universality of the primer for V1-region. The presence of many incomplete/truncated SSU rRNA sequences in RDP database, which might be missing the V1 primer binding sites may also contribute to this observation. The overall results, however, do not indicate any significant deviations in the specificity (fraction of bacterial sequences amplified) of primer pairs targeting non-contiguous V-regions, when compared to the primers targeting contiguously placed V-regions.
The illustrated steps are set out to explain the exemplary embodiments shown, and it should be anticipated that ongoing technological development will change the manner in which particular functions are performed. These examples are presented herein for purposes of illustration, and not limitation. Further, the boundaries of the functional building blocks have been arbitrarily defined herein for the convenience of the description. Alternative boundaries can be defined so long as the specified functions and relationships thereof are appropriately performed. Alternatives (including equivalents, extensions, variations, deviations, etc., of those described herein) will be apparent to persons skilled in the relevant art(s) based on the teachings contained herein. Such alternatives fall within the scope and spirit of the disclosed embodiments. Also, the words “comprising,” “having,” “containing,” and “including,” and other similar forms are intended to be equivalent in meaning and be open ended in that an item or items following any one of these words is not meant to be an exhaustive listing of such item or items, or meant to be limited to only the listed item or items. It must also be noted that as used herein and in the appended claims, the singular forms “a,” “an,” and “the” include plural references unless the context clearly dictates otherwise.
Furthermore, one or more computer-readable storage media may be utilized in implementing embodiments consistent with the present disclosure. A computer-readable storage medium refers to any type of physical memory on which information or data readable by a processor may be stored. Thus, a computer-readable storage medium may store instructions for execution by one or more processors, including instructions for causing the processor(s) to perform steps or stages consistent with the embodiments described herein. The term “computer-readable medium” should be understood to include tangible items and exclude carrier waves and transient signals, i.e., be non-transitory. Examples include random access memory (RAM), read-only memory (ROM), volatile memory, nonvolatile memory, hard drives, CD ROMs, DVDs, flash drives, disks, and any other known physical storage media.
It is intended that the disclosure and examples be considered as exemplary only, with a true scope and spirit of disclosed embodiments being indicated by the following claims.
| Number | Date | Country | Kind |
|---|---|---|---|
| 201821030219 | Aug 2018 | IN | national |
