PEMA v2: addressing metabarcoding bioinformatics analysis challenges

Environmental DNA (eDNA) and metabarcoding have launched a new era in bio- and eco-assessment over the last years (Ruppert et al. 2019). The simultaneous identification, at the lowest taxonomic level possible, of a mixture of taxa from a great range of samples is now feasible; thus, the number of eDNA metabarcoding studies has increased radically (Deiner and 2017). While the experimental part of eDNA metabarcoding can be rather challenging depending on the special characteristics of the different studies, computational issues are considered to be its major bottlenecks. Among the latter, the bioinformatics analysis of metabarcoding data and especially the taxonomy assignment of the sequences are fundamental challenges. Many steps are required to obtain taxonomically assigned matrices from raw data. For most of these, a plethora of tools are available. However, each tool's execution parameters need to be tailored to reflect each experiment's idiosyncrasy; thus, tuning bioinformatics analysis has proved itself fundamental (Kamenova 2020). The computation capacity of high-performance computing systems (HPC) is frequently required for such analyses. On top of that, the non perfect completeness and correctness of the reference taxonomy databases is another important issue (Loos et al. 2020). Based on third-party tools, we have developed the Pipeline for Environmental Metabarcoding Analysis (PEMA), a HPC-centered, containerized assembly of key metabarcoding analysis tools. PEMA combines state-of-the art technologies and algorithms with an easy to get-set-use framework, allowing researchers to tune thoroughly each study thanks to roll-back checkpoints and on-demand partial pipeline execution features (Zafeiropoulos 2020). Once PEMA was released, there were two main pitfalls soon to be highlighted by users. PEMA supported 4 marker genes and was bounded by specific reference databases. In this new version of PEMA the analysis of any marker gene is now available since a new feature was added, allowing classifiers to train a user-provided reference database and use it for taxonomic assignment. Fig. 1 shows the taxonomy assignment related PEMA modules; all those out of the dashed box have been developed for this new PEMA release. As shown, the RDPClassifier has been trained with Midori reference 2 and has been added as an option, classifying not only metazoans but sequences from all taxonomic groups of Eukaryotes for the case of the COI marker gene. A PEMA documentation site is now also available. PEMA.v2 containers are available via the DockerHub and SingularityHub as well as through the Elixir Greece AAI Service. It has also been selected to be part of the LifeWatch ERIC Internal Joint Initiative for the analysis of ARMS data and soon will be available through the Tesseract VRE.

Download Full-text

PEMA: from the raw .fastq files of 16S rRNA and COI marker genes to the (M)OTU-table, a thorough metabarcoding analysis

10.1101/709113 ◽

2019 ◽

Author(s):

Haris Zafeiropoulos ◽

Ha Quoc Viet ◽

Katerina Vasileiadou ◽

Antonis Potirakis ◽

Christos Arvanitidis ◽

...

Keyword(s):

16S Rrna ◽

Programming Languages ◽

High Performance ◽

Marker Gene ◽

Environmental Dna ◽

Operational Taxonomic Unit ◽

Third Party ◽

Marker Genes ◽

New Era ◽

Efficient Performance

AbstractBackgroundEnvironmental DNA (eDNA) and metabarcoding, allow the identification of a mixture of individuals and launch a new era in bio- and eco-assessment. A number of steps are required to obtain taxonomically assigned (Molecular) Operational Taxonomic Unit ((M)OTU) tables from raw data. For most of these, a plethora of tools is available; each tool’s execution parameters need to be tailored to reflect each experiment’s idiosyncrasy. Adding to this complexity, for such analyses, the computation capacity of High Performance Computing (HPC) systems is frequently required.Software containerization technologies ease the sharing and running of software packages across operating systems; thus, they strongly facilitate pipeline development and usage. Likewise are programming languages specialized for big data pipelines, incorporating features like roll-back checkpoints and on-demand partial pipeline execution.FindingsPEMA is a containerized assembly of key metabarcoding analysis tools with a low effort in setting up, running and customizing to researchers’ needs. Based on third party tools, PEMA performs reads’ pre-processing, clustering to (M)OTUs and taxonomy assignment for 16S rRNA and COI marker gene data. Due to its simplified parameterisation and checkpoint support, PEMA allows users to explore alternative algorithms for specific steps of the pipeline without the need of a complete re-execution. PEMA was evaluated against previously published datasets and achieved comparable quality results.ConclusionsGiven its time-efficient performance and its quality results, it is suggested that PEMA can be used for accurate eDNA metabarcoding analysis, thus enhancing the applicability of next-generation biodiversity assessment studies.

Download Full-text

PEMA: a flexible Pipeline for Environmental DNA Metabarcoding Analysis of the 16S/18S ribosomal RNA, ITS, and COI marker genes

GigaScience ◽

10.1093/gigascience/giaa022 ◽

2020 ◽

Vol 9 (3) ◽

Cited By ~ 7

Author(s):

Haris Zafeiropoulos ◽

Ha Quoc Viet ◽

Katerina Vasileiadou ◽

Antonis Potirakis ◽

Christos Arvanitidis ◽

...

Keyword(s):

High Performance Computing ◽

Ribosomal Rna ◽

High Performance ◽

Marker Gene ◽

Environmental Dna ◽

Marker Genes ◽

Sequence Variant ◽

18S Ribosomal Rna ◽

Dna Metabarcoding ◽

Performance Computing

Abstract Background Environmental DNA and metabarcoding allow the identification of a mixture of species and launch a new era in bio- and eco-assessment. Many steps are required to obtain taxonomically assigned matrices from raw data. For most of these, a plethora of tools are available; each tool's execution parameters need to be tailored to reflect each experiment's idiosyncrasy. Adding to this complexity, the computation capacity of high-performance computing systems is frequently required for such analyses. To address the difficulties, bioinformatic pipelines need to combine state-of-the art technologies and algorithms with an easy to get-set-use framework, allowing researchers to tune each study. Software containerization technologies ease the sharing and running of software packages across operating systems; thus, they strongly facilitate pipeline development and usage. Likewise programming languages specialized for big data pipelines incorporate features like roll-back checkpoints and on-demand partial pipeline execution. Findings PEMA is a containerized assembly of key metabarcoding analysis tools that requires low effort in setting up, running, and customizing to researchers’ needs. Based on third-party tools, PEMA performs read pre-processing, (molecular) operational taxonomic unit clustering, amplicon sequence variant inference, and taxonomy assignment for 16S and 18S ribosomal RNA, as well as ITS and COI marker gene data. Owing to its simplified parameterization and checkpoint support, PEMA allows users to explore alternative algorithms for specific steps of the pipeline without the need of a complete re-execution. PEMA was evaluated against both mock communities and previously published datasets and achieved results of comparable quality. Conclusions A high-performance computing–based approach was used to develop PEMA; however, it can be used in personal computers as well. PEMA's time-efficient performance and good results will allow it to be used for accurate environmental DNA metabarcoding analysis, thus enhancing the applicability of next-generation biodiversity assessment studies.

Download Full-text

Bacteria are everywhere, even in your COI data: Τhe art of getting to know the unknown unknowns and shine light on the dark matter!

ARPHA Conference Abstracts ◽

10.3897/aca.4.e64966 ◽

2021 ◽

Vol 4 ◽

Author(s):

Haris Zafeiropoulos ◽

Laura Gargan ◽

Christina Pavloudi ◽

Evangelos Pafilis ◽

Jens Carlsson

Keyword(s):

Dark Matter ◽

Phylogenetic Tree ◽

Sequence Data ◽

Marker Gene ◽

Environmental Dna ◽

Tree Of Life ◽

Taxonomic Assignment ◽

Species Identity ◽

Consensus Sequences ◽

Coi Sequences

Environmental DNA (eDNA) metabarcoding has been commonly used in recent years (Jeunen et al. 2019) for the identification of the species composition of environmental samples. By making use of genetic markers anchored in conserved gene regions, universally present acrooss the species of large taxonomy groups, eDNA metabarcoding exploits both extra- and intra-cellular DNA fragments for biodiversity assessment. However, there is not a truly “universal” marker gene that is capable of amplifying all species across different taxa (Kress et al. 2015). The mitochondrial cytochrome C oxidase subunit I gene (COI) has many of the desirable properties of a “universal" marker and has been widely used for assessing species identity in Eukaryotes, especially metazoans (Andjar et al. 2018). However, a great number of COI Operational Taxonomic Units (OTUs) or/and Amplicon Sequence Variants (ASVs) retrieved from such studies do not match reference sequences and are often referred to as “dark matter” (Deagle et al. 2014). The aim of this study was to discover the origins and identities of these COI dark matter sequences. We built a reference phylogenetic tree that included as many COI-sequence-related information across the tree of life as possible. An overview of the steps followed is presented in Fig. 1a. Briefly, the Midori reference 2 database was used to retrieve eukaryotes sequences (183,330 species). In addition, the API of the BOLD database was used as source for the corresponding Bacteria (559 genera) and Archaea (41 genera) sequences. Consensus sequences at the family level were constructed from each of these three initial COI datasets. The COI-oriented reference phylogenetic tree of life was then built by using 1,240 consensus sequences with more than 80% of those coming from eukaryotic taxa. Phylogeny-based taxonomic assignment was then used to place query sequences. The a) total number of sequences, b) sequences assigned to Eukaryotes and c) unassigned subsets of OTUs, from marine and freshwater samples, retrieved during in-house metabarcoding experiments, were placed in the reference tree (Fig. 1b). It is clear that a large proportion of sequences targeting the COI region of Eukaryotes actually represents bacterial branches in the phylogenetic tree (Fig. 1b). We conclude that COI metabarcoding studies targeting Eukaryotes may come with a great bias derived from amplification and sequencing of bacterial taxa, depending on the primer pair used. However, for the time being, publicly available bacterial COI sequences are far too few to represent the bacterial variability; thus, a reliable taxonomic identification of them is not possible. We suggest that bacterial COI sequences should be included in the reference databases used for the taxonomy assignment of OTUs/ASVs in COI-based eukaryote metabarcoding studies to allow for bacterial sequences that were amplified to be excluded enabling researchers to exclude non-target sequences. Further, the approach presented here allows researchers to better understand the unknown unknowns and shed light on the dark matter of their metabarcoding sequence data.

Download Full-text

RESCRIPt: Reproducible sequence taxonomy reference database management

PLoS Computational Biology ◽

10.1371/journal.pcbi.1009581 ◽

2021 ◽

Vol 17 (11) ◽

pp. e1009581

Author(s):

Michael S. Robeson ◽

Devon R. O’Rourke ◽

Benjamin D. Kaehler ◽

Michal Ziemski ◽

Matthew R. Dillon ◽

...

Keyword(s):

Nucleotide Sequence ◽

Marker Gene ◽

Environmental Dna ◽

Reference Sequence ◽

Genome Comparison ◽

Reference Database ◽

Reference Databases ◽

Ncbi Refseq ◽

Quality Filtering ◽

Metagenome Sequencing

Nucleotide sequence and taxonomy reference databases are critical resources for widespread applications including marker-gene and metagenome sequencing for microbiome analysis, diet metabarcoding, and environmental DNA (eDNA) surveys. Reproducibly generating, managing, using, and evaluating nucleotide sequence and taxonomy reference databases creates a significant bottleneck for researchers aiming to generate custom sequence databases. Furthermore, database composition drastically influences results, and lack of standardization limits cross-study comparisons. To address these challenges, we developed RESCRIPt, a Python 3 software package and QIIME 2 plugin for reproducible generation and management of reference sequence taxonomy databases, including dedicated functions that streamline creating databases from popular sources, and functions for evaluating, comparing, and interactively exploring qualitative and quantitative characteristics across reference databases. To highlight the breadth and capabilities of RESCRIPt, we provide several examples for working with popular databases for microbiome profiling (SILVA, Greengenes, NCBI-RefSeq, GTDB), eDNA and diet metabarcoding surveys (BOLD, GenBank), as well as for genome comparison. We show that bigger is not always better, and reference databases with standardized taxonomies and those that focus on type strains have quantitative advantages, though may not be appropriate for all use cases. Most databases appear to benefit from some curation (quality filtering), though sequence clustering appears detrimental to database quality. Finally, we demonstrate the breadth and extensibility of RESCRIPt for reproducible workflows with a comparison of global hepatitis genomes. RESCRIPt provides tools to democratize the process of reference database acquisition and management, enabling researchers to reproducibly and transparently create reference materials for diverse research applications. RESCRIPt is released under a permissive BSD-3 license at https://github.com/bokulich-lab/RESCRIPt.

Download Full-text

Aquatic biofilms can act as natural environmental DNA samplers

ARPHA Conference Abstracts ◽

10.3897/aca.4.e64812 ◽

2021 ◽

Vol 4 ◽

Author(s):

Sinziana Rivera ◽

Valentin Vasselon ◽

Frederic Rimet ◽

Agnès Bouchez

Keyword(s):

Water Samples ◽

Ecological Status ◽

Fish Communities ◽

Morphological Characteristics ◽

Environmental Dna ◽

Reference Database ◽

Macroinvertebrate Communities ◽

Taxonomic Assignment ◽

Novel Approach ◽

Single Matrix

Diatoms, macroinvertebrates and fish communities are widely used for the assessment of the ecological status of rivers and lakes. Metabarcoding studies of these communities are usually performed from “bulk” samples in the case of diatoms and macroinvertebrates; and from water samples in the case of fish. Recent studies, suggest that aquatic biofilms can physically act as environmental catchers of environmental DNA (eDNA) (e.g. Mariani et al. 2019). Thus, we propose an alternative metabarcoding approach to study macroinvertebrates and fishes directly from this matrix. The capacity of aquatic biofilms to catch macroinvertebrate eDNA was tested from a previous study in Mayotte Island were both biofilm samples and macroinvertebrate morphological inventories were available at same river sites (Rivera et al. 2021). First, macroinvertebrate specimens were identified based on their morphological characteristics. Second, DNA was extracted from biofilms, and macroinvertebrate communities were targeted using a standard COI barcode. The resulting morphological and molecular inventories were compared. Our results showed that both methods provided comparable structures and diversities for macroinvertebrate communities when using unassigned OTUs suggesting that macroinvertebrate DNA is present in biofilms and representative of the communities. However, after taxonomic assignment of OTUs, diversity and richness were no longer correlated. Indeed, many constraints were observed as the need for: a) more specific primers to avoid co-amplification of untargeted taxa inhabiting biofilms, b) primers targeting shorter barcodes to sequence more easily degraded eDNA that may be captured in biofilms, and c) a reference database well adapted to our tropical study sites. Finally, even if the results of this first study were encouraging, we wanted to test the biofilm approach on organisms that do not inhabit this environmental matrix in order to be able to distinguish between intra or extra-cellular DNA. Based on these observations, a second study looking for a fish eDNA signal in aquatic biofilms was performed. Environmental biofilm and water samples were collected in parallel at littoral sites at Lake Geneva. DNA was extracted from these samples, and fish communities were targeted using a standard 12S barcode. The molecular inventories derived from the biofilm and the water samples were compared. Both methods provide comparable floristic lists, providing a novel approach for ecological studies related to fish phenology using eDNA in biofilms. Our results open the door to the study of diatoms, macroinvertebrates and fish communities through metabarcoding from a single matrix reducing sampling efforts and costs.

Download Full-text

RESCRIPt: Reproducible sequence taxonomy reference database management for the masses

10.1101/2020.10.05.326504 ◽

2020 ◽

Cited By ~ 1

Author(s):

Michael S. Robeson ◽

Devon R. O’Rourke ◽

Benjamin D. Kaehler ◽

Michal Ziemski ◽

Matthew R. Dillon ◽

...

Keyword(s):

Nucleotide Sequence ◽

Marker Gene ◽

Environmental Dna ◽

Reference Sequence ◽

Genome Comparison ◽

Reference Database ◽

Reference Databases ◽

Quality Filtering ◽

Metagenome Sequencing ◽

The Masses

AbstractBackgroundNucleotide sequence and taxonomy reference databases are critical resources for widespread applications including marker-gene and metagenome sequencing for microbiome analysis, diet metabarcoding, and environmental DNA (eDNA) surveys. Reproducibly generating, managing, using, and evaluating nucleotide sequence and taxonomy reference databases creates a significant bottleneck for researchers aiming to generate custom sequence databases. Furthermore, database composition drastically influences results, and lack of standardizations limits cross-study comparisons. To address these challenges, we developed RESCRIPt, a software package for reproducible generation and management of reference sequence taxonomy databases, including dedicated functions that streamline creating databases from popular sources, and functions for evaluating, comparing, and interactively exploring qualitative and quantitative characteristics across reference databases.ResultsTo highlight the breadth and capabilities of RESCRIPt, we provide several examples for working with popular databases for microbiome profiling (SILVA, Greengenes, NCBI-RefSeq, GTDB), eDNA, and diet metabarcoding surveys (BOLD, GenBank), as well as for genome comparison. We show that bigger is not always better, and reference databases with standardized taxonomies and those that focus on type strains have quantitative advantages, though may not be appropriate for all use cases. Most databases appear to benefit from some curation (quality filtering), though sequence clustering appears detrimental to database quality. Finally, we demonstrate the breadth and extensibility of RESCRIPt for reproducible workflows with a comparison of global hepatitis genomes.ConclusionsRESCRIPt provides tools to democratize the process of reference database acquisition and management, enabling researchers to reproducibly and transparently create reference materials for diverse research applications. RESCRIPt is released under a permissive BSD-3 license at https://github.com/bokulich-lab/RESCRIPt.

Download Full-text

Are multiplexed metabarcoding panels comparable to individual marker gene library preparations?

ARPHA Conference Abstracts ◽

10.3897/aca.4.e64895 ◽

2021 ◽

Vol 4 ◽

Author(s):

Niamh Eastwood ◽

Luisa Orsini

Keyword(s):

Environmental Gradients ◽

Marker Gene ◽

Environmental Dna ◽

18S Rrna Gene ◽

Marker Genes ◽

Rrna Gene ◽

Rbcl Gene ◽

Single Marker ◽

Multiple Regions ◽

Phylogenetic Resolution

Analysis of multiple marker genes using metabarcoding of environmental DNA (eDNA) can offer information greater than that from sequencing single marker genes, such as responses from across the phylogenetic tree to environmental gradients (Cordier et al. 2019). Furthermore, multiple regions of the same gene can be sequenced to improve phylogenetic resolution (Fuks et al. 2018). However, separate amplification reactions and library preparation steps for each marker can be costly and time consuming. Here, we have designed and optimised a multiplex panel of four marker genes (two regions of 18S rRNA gene, one region of the 16S rRNA gene and one region of the rbcL gene). By combining steps into a single reaction, the labwork required is decreased, reducing cost and time. This multiplex is compared with a widely available commercial microbial (bacterial and fungal) screening panel and individual library preparations of each marker gene.

Download Full-text

Assessing accuracy and completeness of GenBank for eDNA metabarcoding: towards a reliable marine fish reference database

ARPHA Conference Abstracts ◽

10.3897/aca.4.e64671 ◽

2021 ◽

Vol 4 ◽

Author(s):

Cristina Claver ◽

Oriol Canals ◽

Naiara Rodriguez-Ezpeleta

Keyword(s):

Ribosomal Rna ◽

Gap Analysis ◽

Environmental Dna ◽

Reference Database ◽

Fish Diversity ◽

Cyt B ◽

Taxonomic Assignment ◽

Reference Databases ◽

Taxonomic Groups ◽

High Level

Environmental DNA (eDNA) metabarcoding, the process of sequencing DNA collected from the environment for producing biodiversity inventories, is increasingly being applied to assess fish diversity and distribution in marine environments. Yet, the successful application of this technique deeply relies on accurate and complete reference databases used for taxonomic assignment. The most used markers for fish eDNA metabarcoding studies are the cytochrome C oxidase subunit 1 (COI), 16S ribosomal RNA (16S), the 12S ribosomal RNA (12S) and cytochrome b (cyt b) genes, whose sequences are usually retrieved from GenBank, the largest DNA sequence database that represents a worldwide public resource for genetic studies. Thus, the completeness and accuracy of GenBank is critical to derive reliable estimations from fish eDNA metabarcoding data. Here, we have i) compiled the checklist of European marine fishes, ii) performed a gap analysis of the four genes and, within COI and 12S, also of the most used barcodes for fish, and iii) developed a workflow to detect potentially incorrect records in GenBank. We found that from the 1965 species in the checklist (1761 Actinopterygii, 189 Elasmobranchii, 9 Holocephali, 4 Petromyzonti and 2 Myxini), about 70% have sequences for COI, whereas less have sequences for 12S, 16S and cyt b (45-55%). Among the species for which COI ad 12S sequences are available, about 60% and 40% have sequences covering the most used barcodes respectively. The analysis of pairwise distances between sequences revealed pairs belonging to the same species with significantly low similarity and pairs belonging to different high level taxonomic groups (class, order) with significantly large similarity. In light of this further confirmation of presence of a substantial number of incorrect records in GenBank, we propose a method for identifying and removing spurious sequences to create reliable and accurate reference databases for eDNA metabarcoding.

Download Full-text

ensembleTax: an R package for determinations of ensemble taxonomic assignments of phylogenetically-informative marker gene sequences

PeerJ ◽

10.7717/peerj.11865 ◽

2021 ◽

Vol 9 ◽

pp. e11865

Author(s):

Dylan Catlett ◽

Kevin Son ◽

Connie Liang

Keyword(s):

Marker Gene ◽

Ensemble Methods ◽

R Package ◽

Reference Database ◽

Gene Sequences ◽

Taxonomic Assignment ◽

Informative Marker ◽

Data Set ◽

Reference Databases ◽

Taxonomic Assignments

Background High-throughput sequencing of phylogenetically informative marker genes is a widely used method to assess the diversity and composition of microbial communities. Taxonomic assignment of sampled marker gene sequences (referred to as amplicon sequence variants, or ASVs) imparts ecological significance to these genetic data. To assign taxonomy to an ASV, a taxonomic assignment algorithm compares the ASV to a collection of reference sequences (a reference database) with known taxonomic affiliations. However, many taxonomic assignment algorithms and reference databases are available, and the optimal algorithm and database for a particular scientific question is often unclear. Here, we present the ensembleTax R package, which provides an efficient framework for integrating taxonomic assignments predicted with any number of taxonomic assignment algorithms and reference databases to determine ensemble taxonomic assignments for ASVs. Methods The ensembleTax R package relies on two core algorithms: taxmapper and assign.ensembleTax. The taxmapper algorithm maps taxonomic assignments derived from one reference database onto the taxonomic nomenclature (a set of taxonomic naming and ranking conventions) of another reference database. The assign.ensembleTax algorithm computes ensemble taxonomic assignments for each ASV in a data set based on any number of taxonomic assignments determined with independent methods. Various parameters allow analysts to prioritize obtaining either more ASVs with more predicted clade names or more robust clade name predictions supported by multiple independent methods in ensemble taxonomic assignments. Results The ensembleTax R package is used to compute two sets of ensemble taxonomic assignments for a collection of protistan ASVs sampled from the coastal ocean. Comparisons of taxonomic assignments predicted by individual methods with those predicted by ensemble methods show that conservative implementations of the ensembleTax package minimize disagreements between taxonomic assignments predicted by individual and ensemble methods, but result in ASVs with fewer ranks assigned taxonomy. Less conservative implementations of the ensembleTax package result in an increased fraction of ASVs classified at all taxonomic ranks, but increase the number of ASVs for which ensemble assignments disagree with those predicted by individual methods. Discussion We discuss how implementation of the ensembleTax R package may be optimized to address specific scientific objectives based on the results of the application of the ensembleTax package to marine protist communities. While further work is required to evaluate the accuracy of ensemble taxonomic assignments relative to taxonomic assignments predicted by individual methods, we also discuss scenarios where ensemble methods are expected to improve the accuracy of taxonomy prediction for ASVs.

Download Full-text

Multi-view feature selection for identifying gene markers: a diversified biological data driven approach

BMC Bioinformatics ◽

10.1186/s12859-020-03810-0 ◽

2020 ◽

Vol 21 (S18) ◽

Author(s):

Sudipta Acharya ◽

Laizhong Cui ◽

Yi Pan

Keyword(s):

Gene Expression ◽

Feature Selection ◽

Gene Selection ◽

Marker Gene ◽

Biological Data ◽

Protein Interaction Data ◽

Marker Genes ◽

Data Sets ◽

Gene Markers ◽

Multi Objective

Abstract Background In recent years, to investigate challenging bioinformatics problems, the utilization of multiple genomic and proteomic sources has become immensely popular among researchers. One such issue is feature or gene selection and identifying relevant and non-redundant marker genes from high dimensional gene expression data sets. In that context, designing an efficient feature selection algorithm exploiting knowledge from multiple potential biological resources may be an effective way to understand the spectrum of cancer or other diseases with applications in specific epidemiology for a particular population. Results In the current article, we design the feature selection and marker gene detection as a multi-view multi-objective clustering problem. Regarding that, we propose an Unsupervised Multi-View Multi-Objective clustering-based gene selection approach called UMVMO-select. Three important resources of biological data (gene ontology, protein interaction data, protein sequence) along with gene expression values are collectively utilized to design two different views. UMVMO-select aims to reduce gene space without/minimally compromising the sample classification efficiency and determines relevant and non-redundant gene markers from three cancer gene expression benchmark data sets. Conclusion A thorough comparative analysis has been performed with five clustering and nine existing feature selection methods with respect to several internal and external validity metrics. Obtained results reveal the supremacy of the proposed method. Reported results are also validated through a proper biological significance test and heatmap plotting.

Download Full-text