Figures
Abstract
Traditional Sanger sequencing as well as Next-Generation Sequencing have been used for the identification of disease causing mutations in human molecular research. The majority of currently available tools are developed for research and explorative purposes and often do not provide a complete, efficient, one-stop solution. As the focus of currently developed tools is mainly on NGS data analysis, no integrative solution for the analysis of Sanger data is provided and consequently a one-stop solution to analyze reads from both sequencing platforms is not available. We have therefore developed a new pipeline called MutAid to analyze and interpret raw sequencing data produced by Sanger or several NGS sequencing platforms. It performs format conversion, base calling, quality trimming, filtering, read mapping, variant calling, variant annotation and analysis of Sanger and NGS data under a single platform. It is capable of analyzing reads from multiple patients in a single run to create a list of potential disease causing base substitutions as well as insertions and deletions. MutAid has been developed for expert and non-expert users and supports four sequencing platforms including Sanger, Illumina, 454 and Ion Torrent. Furthermore, for NGS data analysis, five read mappers including BWA, TMAP, Bowtie, Bowtie2 and GSNAP and four variant callers including GATK-HaplotypeCaller, SAMTOOLS, Freebayes and VarScan2 pipelines are supported. MutAid is freely available at https://sourceforge.net/projects/mutaid.
Citation: Pandey RV, Pabinger S, Kriegner A, Weinhäusel A (2016) MutAid: Sanger and NGS Based Integrated Pipeline for Mutation Identification, Validation and Annotation in Human Molecular Genetics. PLoS ONE 11(2): e0147697. https://doi.org/10.1371/journal.pone.0147697
Editor: Junwen Wang, The University of Hong Kong, HONG KONG
Received: September 17, 2015; Accepted: January 7, 2016; Published: February 3, 2016
Copyright: © 2016 Pandey et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All data used and described in the manuscript are freely available from the MutAid source forge page (https://sourceforge.net/projects/mutaid/files/test_data.zip) and from NCBI SRA database (http://www.ncbi.nlm.nih.gov/sra/?term=SRP048874).
Funding: The authors have no support or funding to report.
Competing interests: The authors have declared that no competing interests exist.
Abbreviations: NGS, Next Generation Sequencing; WGS, Whole Genome Sequencing; WES, Whole Exome Sequencing; GTF, Gene Transfer Format; SFF, Standard Flowgram Format; dbSNP, Single Nucleotide Polymorphism Database; SNV, Single Nucleotide Variant; INDEL, Insertion-Deletion; VCF, Variant Call Format; BED, Browser Extensible Data; BAM, Binary Alignment Map; SAM, Sequence Alignment Map; QC, Quality Control; IGV, Integrative Genomics Viewer; NCBI, National Center for Biotechnology Information; SRA, Short Read Archive
Introduction
Next-Generation Sequencing (NGS) has become a powerful, efficient and cost-effective clinical tool for mutation screening and decoding a number of genetically heterogeneous diseases such as cancer [1]. During the past few years, several clinical studies have already been conducted by applying multi-gene panel tests [2–6], whole exome sequencing [7–9] and whole genome sequencing [10–14]. While NGS technologies have been used to identify variants in several patients in a cost and time effective manner, which typically yields hundreds to several thousands of variants per patients/samples, traditional Sanger sequencing has been used as a complementary method to confirm the NGS-detected variants before making clinical decisions [15].
Currently, the NGS and Sanger data analysis processes, including quality control, mapping, variant identification, validation and clinical annotation, are labor-intensive and cumbersome tasks, which hamper its usage in routine mutation screening. Therefore, a robust and integrated bioinformatics solution, which can be used to analyze several patients in parallel and produces an annotated list of variants from raw sequencing reads, is extremely useful. In addition, it should be possible to run the complete workflow with a single command, and NGS (Illumina, 454 and Ion torrent) as well as Sanger sequencing data analysis should be supported under a single platform.
At present, a few solutions are available for NGS data quality control, mapping, variant calling and variant effect prediction such as ngs_backbone [16], bcbio_nextgen [17], and SIMPLEX [18]. These are widely used in genomic research but do not provide a specific focus for the analysis of clinical samples. Moreover, none of these provide a complete pipeline for mutation screening by Sanger sequencing data analysis or extensive variant annotations. Bcbio_nextgen is a useful tool for short reads mapping, BAM file quality filtering and variant calling but does not provide quality control for raw reads, variant annotation or Sanger sequencing analysis. SIMPLEX is another pipeline for Whole Exome Sequencing (WES) data analysis of Illumina and ABI SOLiD reads, but does not support 454, Ion Torrent or Sanger sequencing data.
In addition to pipeline systems, workflow management tools, such as Galaxy [19], Knime [20] and Chipster [21], are capable of analyzing sequencing data, but do not provide workflows equipped for use in human molecular genetics research and genetic testing settings. Consequently, users do not have an integrated solution to identify variants with NGS data and validate them with Sanger sequencing data.
To address above limitations in data analysis and interpretation for mutation identification, validation and annotation, we present here a new bioinformatics solution, MutAid, a powerful, integrated and easy to use pipeline for Sanger and NGS data analysis. It covers gene panel, exome as well as whole genome sequencing widely used in human molecular genetics. The tool supports one-stop analysis of various sequencing data from raw reads to an annotated variant list. Moreover, aligned reads in BAM/SAM format generated from any sequencing data can be used as direct inputs in MutAid to identify and confirm mutations. MutAid provides a robust and powerful pipeline to analyze NGS sequencing data by supporting five short read mappers including BWA [22], Bowtie [23], Bowtie2 [24], GSNAP [25] and TMAP [26] and four variant callers including GATK-HaplotypeCaller [27], SAMTOOLS [28], Freebayes [29] and VarScan2 [30]. The parallel execution of these mappers and variant callers enable users to quickly select the consensus variants (called by more than one variant caller), which are more reliable than variants only identified by a single tool. For each variant MutAid predicts the variant effect, such as codon change, amino acid change, and frameshift. Furthermore, it cross-references each variant with more than 30 clinically relevant public databases, which contain already reported SNPs and INDELs from previous experimental studies along with associated genomic, proteomic, and clinical information. MutAid can be used to analyze, elucidate and interpret mutational variants from data generated by targeted re-sequencing, gene-panel sequencing, exome, and whole genome sequencing.
Implementation
The MutAid pipeline uses Python 2.7.9 and implements multiprocessing capabilities to efficiently analyze diagnostic sequencing data produced from Sanger sequencing, Illumina, 454 and Ion Torrent sequencing. It has been designed to specifically consider the Sanger and NGS sequencing chemistries, data format and data quality control requirements. MutAid provides a complete solution for diagnostic sequencing by the seamless integration of traditional Sanger sequencing and high throughput NGS under a single platform. The tool is provided as a fully configured Virtual Machine (VM), which can be used on Windows, Linux and Mac OSX systems with a free Virtual Box installation (https://www.virtualbox.org/). The complete data analysis workflow of MutAid can be started with a simple command and can be customized with many useful parameters. Optionally, the pipeline can be run step-by-step in a modular approach, which allows users to examine the quality of the sequencing data and fine-tune the parameters for specific requirements.
MutAid pipeline
The MutAid pipeline (Fig 1) consists of six sequential steps: 1) quality control and filtering, 2) mapping reads to reference genome, 3) variant detection, 4) variant effect prediction, 5) variant annotation and 6) creation of a variant summary table.
The MutAid pipeline can be run with a single command. Sanger sequencing data analysis has one start point and the flow of analysis runs from top to the bottom, illustrated by a black arrow. NGS has three starting points: 1) raw reads (red color), 2) high quality FASTQ file (blue color)—in this case first step is skipped and 3) mapped reads in BAM or SAM file format (green color)—in this case step 1 and 2 are skipped. * Step is optional.
As shown in Fig 1, Sanger data analysis has only one start point and always starts with raw read files (AB1, SCF). NGS data analysis has three optional start points in the MutAid pipeline: 1) starts with raw read files (FASTQ, SFF, FASTA-QUAL), 2) starts with FASTQ file with high quality reads, and 3) mapped reads in BAM/SAM file.
Step 1. Quality control and filtering.
The MutAid quality control algorithm is designed to take care of platform specific quality issues such as different sequencing error rates or variant calling in low quality regions. This step first converts reads from the original format of the sequencer to the FASTQ format with PHRED quality encoding.
The Sanger sequencing read quality trimming and base recalling are done by TraceTuner [31]. Adapters and Primer sequences are trimmed by AlienTrimmer [32], which is a flexible and sensitive sequence trimmer with mismatch tolerance. Finally reads are trimmed based on the specified minimum average base quality. Reads are discarded from further analysis if they do not meet the given minimum base quality or minimum/maximum read length criteria.
After quality trimming and filtering, a detailed QC report is generated using the widely used FASTQC [33] tool. The report contains various useful plots for each FASTQ file to compare quality statistics before and after quality control.
Step 2. Mapping reads to reference genome.
MutAid maps the sequencing reads to the reference genome followed by post- filtering for minimum mapping quality, proper pair filtering (only for paired end reads), and PCR duplicate removal. By default the long Sanger sequencing and 454 sequencing reads are mapped using the BWA-SW algorithm, which is specially designed for longer reads up to several KB length. For Ion Torrent data, MutAid uses the TMAP mapper, which is tailored to the analysis of Ion Torrent reads. For Illumina data, MutAid uses BWA as the default mapper. MutAid provides three additional mappers including Bowtie, Bowtie2, and GSNAP. Consequently, the user has the option to map a dataset with multiple mappers and compare the resulting variants or select the consensus variants obtained by multiple mappers. The BAM file generated during mapping can be stored or used for re-analysis or co-analysis in other experiments or studies.
Step 3. Variant detection.
MutAid detects SNVs, insertions and deletions from BAM files using SAMTOOLS for Sanger data and GATK for NGS data as default variant callers. In addition to GATK, MutAid supports three more variant callers for NGS including Freebayes, SAMTOOLS and VarScan2.
After mapping, NGS data is first processed by tools from the GATK bundle including IndelRealigner, BaseRecalibrator and HaplotypeCaller for INDEL realignment, quality recalibration, and variant calling, respectively [34]. Further, the variants are filtered based on read coverage and a minor allele frequency threshold defined in the “MutAidOption”file. The resulting realigned BAM files and the VCF files for each patient are generated for further downstream analysis and can be used to visualize the sequences and/or variants in IGV [35].
Step 4. Variant effect prediction.
Using the variant coordinates, the identified variants are assigned to genomic features (e.g., RefSeq Gene, exon, CDS, intron). In addition, MutAid performs variant effect prediction of SNVs and INDELs. For codon and amino acid changes, MutAid extract the CDS coordinates and reading frame information from gene annotation files. Based on the variant position and reading frame the reference allele is replaced with the variant allele to obtain the variant codon. Once the reference codon and the variant codon have been obtained, the corresponding amino acid from the codon translation table is determined. Both codons and amino acids are reported in the final variant summary table. For INDELs, MutAid defines the frameshift mutation in coding regions by incorporating the inserted and deleted base pair into the actual coding sequence and calculates the new reading frame. Finally, MutAid reports the new reading frame along with the original reading frame in the output table.
To obtain the Gene annotation GTF file, reference genome FASTA file and other annotation files, MutAid uses an additional tool (prepref), which downloads all genes and SNP annotations from the UCSC genome browser [36] and prepares them for MutAid analysis.
MutAid calculates the following summary statistics for each mutation: 1) mean and median base quality around each variant in a 10bp flanking region, 2) total sequencing coverage and coverage of A, T, C, G separately before and after quality filtering, 3) p-value before and after multiple testing correction (FDR < = 0.05) calculated using two-tailed fisher’s exact test [37] for each variant in a 2x2 contingency table. A detailed description of Fisher’s exact test is given in S2 Appendix. These summary statistics can be used for sequencing data interpretation and selection of potential candidate mutations.
Step 5. Variant cross-referencing.
MutAid links each resulting variant with more than 30 relevant databases including gene resources, clinical resources, pathways resources, protein resources, GWAS resources, genome browsers and other databases. The reference information files have been used from the UCSC genome browser and NCBI [38].
For variants, which fall in annotated genes, MutAid first assigns RefSeq gene, transcript, protein IDs as well as Entrez gene ID and a gene symbol. In addition, it constructs for each annotation a link to the databases given in Table 1. For intergenic variants, MutAid constructs links to several genome browsers, such as UCSC and Ensemble, using the variant coordinates as well as to the dbSNP database and the NCBI variation viewer.
MutAid constructs direct links to more than 30 publically available databases for each variant in the output summary table. These links are created based on coordinates, and the Entrez gene ID. The table lists, which links will be created for known and novel variants, in exonic and intergenic regions.
As described in Table 1, cross-referencing by coordinate works for all three types of variants (known, novel and intergenic variants) and cross-referencing by dbSNP identifier (SNP ID) works for known and intergenic variants (except for PolyPhen-2).
Step 6. Producing variant summary output.
MutAid creates a final output table that contains one line for each variant along with experimental information, patient information, identified mutations and their genomic effects as well as clinical annotations. In addition, it creates links in the output table to the QC report, the mapped BAM file, the variant VCF file and the resulting FASTQ file for each identified mutation. The produced BAM files and VCF files can be visualized in IGV, or directly plugged into other downstream analysis tools and stored or used in other related projects.
MutAid input
MutAid uses a uniform input and output data model for both Sanger and NGS sequencing data analysis. It can analyze hundreds of sample/patient data in parallel and produces a single variant output file. MutAid requires at least three input files:
1. Target file.
A file containing experimental and sequencing file information for each patient (S1 Table).
2. Adapter-Primer file.
This input file is optional and is required if primer and adapter sequences need to be trimmed. The tab-separated text file with four columns (S2 Table) is used in the first step (quality control and trimming) of the analysis procedure.
3. MutAidOptions file.
This file contains all options for various parts of the pipeline and the path to the installation directory of third party tools including mappers and variant callers. A default MutAidOptions file for Sanger and NGS data analysis is provided separately (S1 and S2 Texts).
4. Sequencing reads.
MutAid supports Sanger sequencing reads in AB1 and SCF file format, Illumina reads in FASTQ format, 454 reads in SFF as well as FASTQ-QUAL format and Ion Torrent reads in SFF and FASTQ format. In addition to raw read files, MutAid also supports calling and annotating variants from alignment (BAM/SAM) files. These sequencing data files can be specified in Target file (S1 Table).
MutAid output
MutAid produces output files in a standardized format for both Sager and NGS, which facilitates post analysis output handling, further downstream analysis and comparison. MutAid returns five output files 1) a variant summary table, 2) a QC report, 3) a VCF file per patient, 4) an indexed BAM file per patient and 5) a Sanger encoded FASTQ file per patient. All these output files will be created in the output directory, which is specified in “MutAidOption” file (S1 and S2 Texts).
The variant summary table is the main output, which contains one line for each resulting variant (SNV or INDEL) and links it with a wide range of information as described in Table 2. The other four outputs QC report, FASTQ file, BAM file and VCF files are produced for each sample/patient separately and are linked from each variant in the variant summary table. The detailed QC report is generated in HTML format and includes various useful plots to check the sequencing data. Two QC reports for each patient/sample before and after quality control are generated, which allows direct comparison of quality control result.
MutAid produces a final variant summary with one line per variant including experimental information, patient information, variant information, variant effects and database cross-references.
Availability, installation and usage
MutAid is an open-source pipeline and freely available for all researchers and clinicians. It is available as a fully configured Virtual Machine accessible at https://sourceforge.net/p/mutaid/wiki/Virtual_Machine/. Moreover, MutAid source code along with pre-compiled required third party tools is available from https://sourceforge.net/projects/mutaid/files/MutAid_v1.0-linux.zip for Linux and from https://sourceforge.net/projects/mutaid/files/MutAid_v1.0-macos.zip for MACOSX operating system. An extensive user manual is available at https://sourceforge.net/projects/mutaid/files/Manual.pdf. Test data sets for Sanger, Illumina, 454 and Ion Torrent are available at https://sourceforge.net/projects/mutaid/files/test_data.zip.
Results and Discussion
MutAid is an open-source bioinformatics solution, which enables users to analyze both NGS and Sanger sequencing data in human genetic testing. It enables users to rapidly analyze hundreds of patients in parallel by executing just a single command. This pipeline is equipped with various custom parameters and provides proper guidelines for the analysis. All components of MutAid and their inputs are summarized in Fig 1 and all key features are described below.
An integrated pipeline
Many clinics and hospitals across the globe have been widely using traditional Sanger sequencing for genetic testing, which is still the gold standard even after the advent of NGS technologies. However, it is time consuming and expensive when multiple genes are sequenced. In contrast, NGS platforms are producing billions of short reads at an unprecedented speed in a cost effective manner. Therefore, a single data analysis solution that combines analysis of traditional Sanger sequencing and NGS data could tremendously improve genetic testing in clinical research and diagnostics. With MutAid users can identify the candidate mutations by using NGS sequencing in a time and cost effective manner and validate them with Sanger sequencing. MutAid can be used by expert and non-expert users to get a list of variants from raw sequencing data in a reads-to-variant manner without manual interaction.
Sanger and NGS integration
MutAid provides an integrated solution for the analysis of Sanger and NGS (Illumina, 454 and Ion Torrent) data analysis under a single platform. It takes a target file as input, which contains sequencing files for all patients/samples along with experimental information for processing. As shown in S1 Fig, Sanger, Illumina, 454 and Ion Torrent have exactly identical target file formats. Thus, the uniform input format enables user to easily prepare a tab delimited input file for different platforms. For both Sanger and NGS, MutAid produces variants in standard output formats, such as VCF and BAM, for each patient/sample, which can be easily visualized. Finally, MutAid creates identical variant summary output tables (Table 2) for both Sanger and NGS data analysis. Hence, identical input, output and visualization models facilitate the direct comparison of variants and confirmation by visual and manual inspection.
Variant calling validation study
MutAid has been designed to support the parallel execution of analysis runs using several different mapper-variant caller combinations. The results of the different runs can be used to eliminate false positives variants calls from high throughput NGS data analysis. User can use all combinations of mappers and variant callers in a single run and MutAid produces a final comprehensive variant summary output table. In the final summary output table, MutAid reports how many mapper(s) and variant caller(s) have confirmed the resulting variant. In order to demonstrate the capabilities of MutAid, we have designed and executed a variant calling validation study involving three mappers and four variant callers. First we examined the effect of using the same variant caller method but three different mapping tools. In the second part of the study, we used the same mapping result as a base for variant calling with four different variant calling tools. This was executed with four different mapping results.
Effect of mapping result on variant calling.
First, we compared the called variants after mapping with three mappers; BWA, Bowtie2 and GSNAP. As shown in Figs 2 and 3 the output of the variant callers is highly consistent when using the result of different mapping programs as base for variant calling. For SNV and INDEL identification 76 to 93.29% total variants overlap between at least two mappers. The availability of multiple mappers allows users to evaluate different mapping programs based on their input data and choose the best mapper for further analysis runs.
GATK shows 93.29% overlap between at least two mappers whereas Varscan2 shows least overlap among all four variant callers with 78%. SAMTOOLS and Freebayes show 92.23% and 88%, respectively, agreement with at least two mappers.
GATK shows 90.78% overlap between at least two mappers and SAMTOOLS shows least overlap among all four variant callers with 74.34%. Varscan2 and Freebayes show 76.56% and 83.70%, respectively, agreement with at least two mappers.
Effect of variant callers on variant calling.
In the second part of our validation study we compared the results of different variant calling tools starting with the same mapping result. MutAid supports four variant callers for NGS data analysis. Users can call variants with all four callers simultaneously with a single MutAid command. Furthermore, users can select the variants identified by all four variant callers to get high confidence results. By default SAMTOOLS is used for variant calling of Sanger sequencing data and GATK-HaplotypeCaller for Illumina, 454 and Ion Torrent sequencing.
In our validation study, we called variants (SNVs and INDELs) with four variant callers from three mapping files produced by BWA, Bowtie2 and GSNAP and compared the resulting variants. As shown in Fig 4, 75% - 84% SNVs were identified by at least two variant callers. Similarly, more than 78% INDELS have been identified by at least two variant callers (Fig 5).
Result shows that 75% - 84% SNVs are common with at least two out of four variant callers. With all 3 mappers Varscan2 identified novel SNVs from 16% - 24%.
Consistent with SNV results more than 78% INDELS are identified by at least two variant callers.
Flexibility and scalability
MutAid provides full flexibility to adjust the input parameters for specific requirements. All input files and parameters can be set using the MutAidOption file (S1 and S2 Texts) to run the whole pipeline without user interaction. MutAid provides three start points for NGS data analysis: 1) with original sequencing reads in SFF, FASTA-QUAL or FASTQ format, 2) reads in FASTQ format with PHRED quality encoding after quality control and filtering (starts with mapping), and 3) with mapped reads in BAM/SAM format (starts with variant calling) (Fig 1). MutAid is designed to handle sequencing data ranging from single-gene to whole genome sequencing of several samples/patients. The tool can be used to analyze several patient data in parallel. Furthermore, it employs a multiprocessing concept to use all assigned CPUs to process data efficiently in parallel.
Quality control report
Employing the correct NGS quality control steps is very crucial to increase the sensitivity and specificity of identified variants. User can check and compare the read quality plots generated by MutAid. This will help to identify the best parameter sets to get high quality reads for mapping and variant calling.
Data visualization
Genomic data analysis often requires additional confirmation of the candidate mutation by visual inspection. MutAid provides a direct link for each mutation to visualize it in the UCSC genome browser, the Ensembl genome browser [39] and the GWAS Central browser [40] on their respective web pages. For local visualization in IGV, MutAid produces two output files for each patient 1) a VCF file and 2) a BAM file for both Sanger and NGS. Thus, the user can view and confirm potential pathogenic mutations by visualizing the BAM files in IGV. As shown in Fig 6, a single nucleotide substitution (T>C) in human breast cancer 2 (BRCA2: NM_000059) has been identified with Illumina whole genome sequencing data (top panel) and confirmed by Sanger sequencing reads (middle panel).
MutAid produces BAM files for NGS and Sanger, which can be loaded into IGV to view and confirm the identified variants. In blue color we can see that SNV (T>C) has been identified by NGS (top panel) and confirmed by Sanger sequencing (middle panel).
Hotspot mutation annotation
MutAid offers a unique feature to co-analyze and annotate the hotspot mutations along with sequencing data. If a variant overlaps with a known hotspot mutation, it is flagged accordingly in the output table. The user can provide these known hotspot mutations in BED format file (S3 Table).
Interpretation of sequencing data
To facilitate sequencing data interpretation and selection of potential candidate mutations, MutAid provides the following relevant summary statistics and information for each mutation (columns are described in Table 2):
- Mean and median base quality around each variant in a 10bp flanking window
- Total sequencing coverage and coverage of A, T, C, G separately before and after quality filtering
- Codon and amino acid change for protein coding genes
- Genomic feature assignment (exonic, intronic, UTR, coding region)
- The statistical significance of allele frequency differences is determined with Fisher's exact test. The resulting p-values before and after FDR correction are reported for each variant. The fisher’s exact test is performed with a 2x2 contingency table by using the allele frequencies (ref and alt) of the variant versus the mean allele frequencies of all remaining variants in a sample/patient (S2 Appendix).
- Cross-references and annotations with more than 30 clinically relevant public databases
Based on this readily available information for each identified variant, users will be able to narrow down and confirm potential candidate mutations.
Interpretation of novel variants
Each NGS test may yield several variants that are novel, which poses a great challenge in clinical research, particularly when resources for functional characterization on an individual patient basis are limited [41–42]. MutAid offers useful information for each mutation, which can be readily used to interpret the novel variants. First, MutAid provides summary and quality statistics about each variant, such as base quality, coverage, mapping quality, codon and amino acid change, and frameshift information. Based on this information, users can identify false calls and only variants with high confidence can be selected for further analysis. In addition, MutAid provides a direct link to many genome browsers such as UCSC, Ensembl, Decipher, GWAS Central and NCBI variation viewer for visualization (Table 1), where plenty of genomic and clinical information is available. For instance, users can now consult the UCSC genome browser and analyze the conservation level of the novel variant position (Fig 7). MutAid supports the combined analysis of family members by using the Family ID input field. If a novel mutation is also found in a control sample it may be ignored from further analysis.
MutAid constructs a direct link to the UCSC genome browser for all variants including novel variants. On top, reference nucleotides are displayed and in the bottom panel (highlighted with green color) the conservation track of several species is displayed. To confirm novel variants, conservation analysis can be performed for each mutation position. A novel mutation might be ignored if a position has poor conservation among the species (pointed by red color arrow). A novel mutation may be further analyzed if the position is highly conserved (pointed by blue color arrow).
Variant annotation
MutAid provides previously described genetic, gene function and pathway information, which could help to select the clinically relevant variants. As shown in Table 1, MutAid cross-references each variant with more than 30 relevant publically available databases. The resources are described below:
Gene resources.
In the final output table each variant is cross-referenced to Entrez Gene [43], HomoloGene [44], GEO Profiles [45], UniGene [46] and Pubmed [47]. These resources will be extremely useful for known and novel variant alike. For unknown variants, Pubmed will be particularly useful to search literature and online databases.
Clinical resources.
MutAid cross-references each variant with the Single Nucleotide Polymorphism Database [48], ClinVar [49], dbVar [50], NCBI variation viewer [51] Cosmic database [52], Genetic Testing Registry [53], Online Mendelian Inheritance in Man (OMIM) [54], HUGO Gene Nomenclature Committee (HGNC) [55], PolyPhen-2 [56] and Decipher database [57].
Pathways resources.
MutAid links each variant with publically available pathway databases, which provide information about the function of a gene. The linked databases are Kyoto Encyclopedia of Genes and Genomes (KEGG) [58] and Reactome [59]. Before starting the data analysis with MutAid, user need to prepare reference information files (genome FASTA sequence, gene annotation, and variant information) by using the “prepref” tool, which is available within MutAid pipeline. prepref download RefSeq gene annotation and linked database cross-reference ID of various databases from the UCSC Table browser (http://genome.ucsc.edu/cgi-bin/hgTables). Therefore, after variant identification, the coordinates are assigned to a RefSeq Gene ID, then to KEGG pathway ID (hsa00232) and UniProt Protein ID (P11245). Based on KEGG pathway ID, MutAid creates a direct link to KEGG pathway page and to the Reactome page.
Protein resources.
MutAid also cross-references protein coding genes with protein databases such as UniProt [60], RefSeq Protein [61] and QuickGO [62].
MutAid validation
To evaluate the MutAid pipeline for mutation screening with NGS data, we used publicly available data (SRP048874) produced from Human Whole Exome Sequencing (WES) of a CEPH/UTAH female individual (HapMap ID: NA12878) (~420 million read pairs, 2x100bp). We downloaded seven sequencing runs from NCBI SRA (SRA ID: SRP048874). After pooling the reads from 7 libraries, data were mapped to the GRCh37/hg19 human reference sequence using the BWA program (0.7.9a) with default parameters. Next, we extracted all reads, which were mapped on chromosome 13 for subsequent analysis.
We ran our MutAid pipeline with ~8 million read pairs. We performed quality control and trimming with minimum read length of 50 base pairs and minimum base quality of 20 and subsequently mapped reads (with passed quality filters) to the chromosome 13 with BWA version 0.7.9a, Bowtie2 version 2.2.3 and GSNAP version gmap-2014-12-16 using their default mapping parameters. We further called variants using MutAid with four variant callers: GATK-HaplotypeCaller version 3.0, SAMTOOLS version 0.1.19, and Freebayes version v0.9.14-14-gb00b735. Next, we filtered variants with minimum read coverage of 20, a minimum variant allele coverage of 4 and variant allele frequency 0.10. The SNVs and INDELs identified by each combination of mappers to variant callers are given in Tables 3 and 4.
SNVs were called using four variant callers (for each mapping result) with a minimum read coverage of 20, minimum variant allele coverage of 4 and a base quality of at least 20. The percentage given in brackets is the fraction of SNVs having an entry in the Single Nucleotide Polymorphism Database (dbSNP) version 137.
INDELs were called using four variant callers (for each mapping result) using the same settings as for SNV calling. The percentage given in brackets is the fraction of INDELs having an entry in dbSNP version 137.
1. Cross-validation of called SNVs with dbSNP database.
In order to evaluate our results, we compared the SNVs and INDELs called by MutAid with the Single Nucleotide Polymorphism database (dbSNP v137) [43] to identify the fraction of already known variants reported in dbSNP. Our results show that 55.65% - 98.10% of all identified variants by MutAid with different variant callers were found in dbSNP with a known function (Table 3).
2. Cross-validation of called INDELs with dbSNP.
We have also compared the INDELs identified by MutAid with dbSNP and we found that 30.19% - 45.92% INDELs were found in dbSNP. Consistent with SNVs, VarScan2 predicted most de-novo INDELs (~69%), which could not be found in dbSNP (Table 4).
Comparison with existing tools
MutAid provides a one-stop solution for simultaneous genetic testing of hundreds of patients by using Sanger and NGS sequencing data. MutAid can be run on any Unix system using a single command and has many user-friendly features, which enable non-expert users to identify and validate the disease causing mutations in a time effective manner with great accuracy and reliability. A comparison of the most important features of MutAid to other available tools is given in Table 5. All similar solutions are available for genomic research with no focus on clinical settings. Moreover none of them supports Ion Torrent and Sanger data analysis for disease causing variant detection and confirmation.
Conclusions
MutAid is a new robust, user-friendly, and integrated bioinformatics pipeline to analyze NGS and Sanger sequencing data with a single command. The pipeline takes raw reads as inputs and outputs a list of annotated variants with great accuracy for hundreds of patients. It is a useful tool to analyze up to thousands of Sanger sequencing trace files automatically in batch mode and produces a comprehensive single output table of identified substitution, insertion and deletion.
In clinical research and diagnostics, NGS and traditional Sanger sequencing are being used as complementary techniques for mutation screening. Consequently, an integrated bioinformatics solution, which offers NGS and Sanger data analysis under a single platform, provides great potential. We believe that MutAid will be very useful in human genetic testing and diagnostic mutation screening when research and diagnostics are coalescent.
Despite the different nature of Sanger and NGS input files, MutAid provides a common input and output interface to facilitate streamlined use and evaluation of detected variants. With the integration of four variant callers, MutAid provides a powerful approach to select the consensus variants among all variant callers, thus lowering the false positive rate.
The source code along with its documentation is freely available under the AGPL license and can be obtained from https://sourceforge.net/projects/mutaid.
Supporting Information
S1 Table. Target file (mandatory input file) to run MutAid.
https://doi.org/10.1371/journal.pone.0147697.s001
(TXT)
S2 Table. Adapter-Primer file (optional input file) to perform adapters and PCR primers trimming.
https://doi.org/10.1371/journal.pone.0147697.s002
(TXT)
S3 Table. Hotspot mutation file in BED format to co-analyze and annotate with MutAid pipeline.
https://doi.org/10.1371/journal.pone.0147697.s003
(TXT)
S1 Text. MutAidOptions_Sanger file to specify all input files and parameters for Sanger sequencing data analysis.
https://doi.org/10.1371/journal.pone.0147697.s004
(TXT)
S2 Text. MutAidOptions_NGS file to specify all input files and parameters for NGS data analysis.
https://doi.org/10.1371/journal.pone.0147697.s005
(TXT)
S1 Appendix. MutAid user manual.
An extensive guide for user to perform Sanger and NGS data analysis with MutAid
https://doi.org/10.1371/journal.pone.0147697.s006
(PDF)
S1 Fig. Shows the uniform Target file implementation for Sanger and NGS in MutAid.
https://doi.org/10.1371/journal.pone.0147697.s008
(TIF)
Acknowledgments
We are grateful to R. Kallmeyer, R. Kreuzhuber and I. Visne for helpful discussion and for testing the pipeline.
Author Contributions
Conceived and designed the experiments: AK RVP AW. Performed the experiments: RVP. Analyzed the data: RVP. Contributed reagents/materials/analysis tools: RVP AK SP. Wrote the paper: RVP. Contributed in manuscript revision and provided helpful suggestion throughout the manuscript preparation: SP.
References
- 1. Ardeshirdavani A, Souche E, Dehaspe L, Van Houdt J, Vermeesch JR, Moreau Y NGS-Logistics: federated analysis of NGS sequence variants across multiple locations. Genome Med. 2014 Sep 17;6(9):71. pmid:25328540
- 2. Hagemann IS, Devarakonda S, Lockwood CM, Spencer DH, Guebert K, Bredemeyer AJ, et al. Clinical next-generation sequencing in patients with non-small cell lung cancer. Cancer. 2014.
- 3. Rehm HL Disease-targeted sequencing: a cornerstone in the clinic. Nat Rev Genet. 2013;14(4):295–300. pmid:23478348
- 4. Shanks ME, Downes SM, Copley RR, Lise S, Broxholme J, Hudspith KA, et al. Next-generation sequencing (NGS) as a diagnostic tool for retinal degeneration reveals a much higher detection rate in early-onset disease. Eur J Hum Genet. 2013;21(3):274–80. pmid:22968130
- 5. Renkema KY, Stokman MF, Giles RH, Knoers NV. Next-generation sequencing for research and diagnostics in kidney disease. Nat Rev Nephrol. 2014;10(8):433–44. pmid:24914583
- 6. Vandrovcova J, Thomas ER, Atanur SS, Norsworthy PJ, Neuwirth C, Tan Y, et al. The use of next-generation sequencing in clinical diagnosis of familial hypercholesterolemia. Genet Med. 2013;15(12):948–57. pmid:23680767
- 7. Hosen MJ, Van Nieuwerburgh F, Steyaert W, Deforce D, Martin L, Leftheriotis G, et al. Efficiency of Exome Sequencing for the Molecular Diagnosis of Pseudoxanthoma Elasticum. J Invest Dermatol. 2014.
- 8. Bamshad MJ, Ng SB, Bigham AW, Tabor HK, Emond MJ, Nickerson DA, et al. Exome sequencing as a tool for Mendelian disease gene discovery. Nat Rev Genet. 2011;12(11):745–55. pmid:21946919
- 9. Blue Cross and Blue Shield Association. Special report: exome sequencing for clinical diagnosis of patients with suspected genetic disorders. Technol Eval Cent Assess Program Exec Summ. 2013;28(3):1–4. pmid:24066368
- 10. Baird RD, Caldas C. Genetic heterogeneity in breast cancer: the road to personalized medicine? BMC Med. 2013;11:151. pmid:23800221
- 11. Ellis MJ, Ding L, Shen D, Luo J, Suman VJ, Wallis JW, et al. Whole-genome analysis informs breast cancer response to aromatase inhibition. Nature. 2012;486(7403):353–60. pmid:22722193
- 12. Ulahannan D, Kovac MB, Mulholland PJ, Cazier JB, Tomlinson I. Technical and implementation issues in using next-generation sequencing of cancers in clinical practice. Br J Cancer. 2013;109(4):827–35. pmid:23887607
- 13. Van Dijk EL, Auger H, Jaszczyszyn Y, Thermes C. Ten years of next-generation sequencing technology. Trends Genet. 2014;30(9):418–26. pmid:25108476
- 14. Quesada V, Ramsay AJ, Rodríguez D, Puente XS, Campo E, López-Otín C. The genomic landscape of chronic lymphocytic leukemia: clinical implications. BMC Med. 2013;11:124. pmid:23656622
- 15. Johnston JJ, Rubinstein WS, Facio FM, Ng D, Singh LN, Teer JK, et al. Secondary variants in individuals undergoing exome sequencing: screening of 572 individuals identifies high-penetrance mutations in cancer-susceptibility genes. Am J Hum Genet. 2012 Jul 13;91(1):97–108. pmid:22703879
- 16. Blanca JM, Pascual L, Ziarsolo P, Nuez F, Cañizares J. ngs_backbone: a pipeline for read cleaning, mapping and SNP calling using next generation sequence. BMC Genomics. 2011;12:285. pmid:21635747
- 17.
bcbio-nextgen [https://bcbio-nextgen.readthedocs.org]
- 18. Fischer M, Snajder R, Pabinger S, Dander A, Schossig A, Zschocke J, et al. SIMPLEX: cloud-enabled pipeline for the comprehensive analysis of exome sequencing data. PLoS One. 2012;7(8):e41948. pmid:22870267
- 19. Goecks J, Nekrutenko A, Taylor J, Galaxy Team. Galaxy: a comprehensive approach for supporting accessible, reproducible, and transparent computational research in the life sciences. Genome Biol. 2010;11(8):R86. pmid:20738864
- 20.
Knime. Accessed: 2016-18-01. Available from: https://www.knime.org/
- 21.
Chipster. Accessed: 2016-18-01. Available from: http://chipster.github.io/chipster/]
- 22. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 2009, 25:1754–1760. pmid:19451168
- 23. Langmead B, Trapnell C, Pop M, Salzberg SL. Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol. 2009b;10(3):R25.
- 24. Langmead B, Salzberg SL: Fast gapped-read alignment with Bowtie 2. Nat Methods 2012;9(4):357–359. pmid:22388286
- 25. Wu TD, Nacu S. Fast and SNP-tolerant detection of complex variants and splicing in short reads. Bioinformatics 2010;26:873–881. pmid:20147302
- 26.
TMAP. Accessed: 2016-18-01. Available from: https://github.com/iontorrent/TMAP.
- 27. McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A, et al. The Genome Analysis Toolkit: A MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res. 2010;20(9):1297–1303. pmid:20644199
- 28. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics. 2009 Aug 15;25(16):2078–2079. pmid:19505943
- 29.
Freebayes. Accessed: 2016-18-01. Available from: https://github.com/ekg/freebayes.
- 30. Koboldt DC, Zhang Q, Larson DE, Shen D, McLellan MD, Lin L, et al. VarScan 2: somatic mutation and copy number alteration discovery in cancer by exome sequencing. Genome Res. 2012;22(3):568–576. pmid:22300766
- 31.
TraceTuner. Accessed: 2016-18-01. Available from: http://sourceforge.net/projects/tracetuner/.
- 32. Criscuolo A, Brisse S. AlienTrimmer: a tool to quickly and accurately trim off multiple short contaminant sequences from high-throughput sequencing reads. Genomics. 2013;102(5–6):500–6. pmid:23912058
- 33.
FastQC A quality control tool for high throughput sequence data. Babraham Bioinformatics Web site. Accessed: 2016-18-01. Available from: http://www.bioinformatics.babraham.ac.uk/projects/fastqc/.
- 34. Van der Auwera GA, Carneiro MO, Hartl C, Poplin R, Del Angel G, Levy-Moonshine A et al. From FastQ data to high confidence variant calls: the Genome Analysis Toolkit best practices pipeline. Curr Protoc Bioinformatics. 2013 Oct 15;11(1110):11. pmid:25431634
- 35. Thorvaldsdottir H, Robinson JT, Mesirov JP. Integrative Genomics Viewer (IGV): high-performance genomics data visualization and exploration. Brief Bioinform 2013;14:178–192. pmid:22517427
- 36. Kent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, Zahler AM, et al. UCSC Genome Browser: The human genome browser at UCSC. Genome Res. 2002;12(6):996–1006.
- 37. Fisher RA. On the interpretation of χ2from contingency tables, and the calculation of P. J. R. Stat. Soc. 1922;85:87–94.
- 38.
National Center for Biotechnology Information (NCBI). Accessed: 2016-18-01. Available from: http://www.ncbi.nlm.nih.gov/.
- 39. Kersey PJ, Lawson D, Birney E, Derwent PS, Haimel M, Herrero J, et al. Ensembl Genomes: extending Ensembl across the taxonomic space. Nucleic Acids Res. 2010;38(Database issue):D563–9. pmid:19884133
- 40. Beck T, Hastings RK, Gollapudi S, Free RC, Brookes AJ. GWAS Central: a comprehensive resource for the comparison and interrogation of genome-wide association studies. Eur J Hum Genet. 2014;22(7):949–52. pmid:24301061
- 41. Rehm HL. Disease-targeted sequencing: a cornerstone in the clinic. Nat Rev Genet. 2013 14(4):295–300. pmid:23478348
- 42. Renkema KY, Stokman MF, Giles RH, Knoers NV. Next-generation sequencing for research and diagnostics in kidney disease. Nat Rev Nephrol. 2014 Aug;10(8):433–44. pmid:24914583
- 43.
Entrez Gene. Accessed: 2016-18-01. Available from: http://www.ncbi.nlm.nih.gov/gene.
- 44.
HomoloGene. Accessed: 2016-18-01. Available from: http://www.ncbi.nlm.nih.gov/homologene.
- 45.
GEO Profiles. Accessed: 2016-18-01. Available from: http://www.ncbi.nlm.nih.gov/geoprofiles.
- 46.
UniGene. Accessed: 2016-18-01. Available from: http://www.ncbi.nlm.nih.gov/unigene.
- 47.
Pubmed. Accessed: 2016-18-01. Available from: http://www.ncbi.nlm.nih.gov/pmc.
- 48. Sherry ST, Ward MH, Kholodov M, Baker J, Phan L, Smigielski EM, et al. dbSNP: the NCBI database of genetic variation. Nucleic Acids Res. 2001;29(1):308–11. pmid:11125122
- 49.
ClinVar. Accessed: 2016-18-01. Available from: http://www.ncbi.nlm.nih.gov/clinvar/.
- 50.
dbVar. Accessed: 2016-18-01. Available from: http://www.ncbi.nlm.nih.gov/dbvar.
- 51.
NCBI Variation Viewer. Accessed: 2016-18-01. Available from: http://www.ncbi.nlm.nih.gov/variation/view/.
- 52.
COSMIC. Accessed: 2016-18-01. Available from: http://cancer.sanger.ac.uk/cosmic/gene/.
- 53.
Genetic Testing Registry (GTR). Accessed: 2016-18-01. Available from: http://www.ncbi.nlm.nih.gov/gtr/genes.
- 54.
OMIM. Accessed: 2016-18-01. Available from: http://omim.org.
- 55.
HGNC. Accessed: 2016-18-01. Available from: http://www.genenames.org/data.
- 56.
PolyPhen-2. Accessed: 2016-18-01. Available from: http://genetics.bwh.harvard.edu/cgi-bin/pph2.
- 57.
Decipher. Accessed: 2016-18-01. Available from: https://decipher.sanger.ac.uk.
- 58.
KEGG. Accessed: 2016-18-01. Available from: http://www.kegg.jp.
- 59.
Reactome. Accessed: 2016-18-01. Available from: http://www.reactome.org.
- 60.
UniProt. Accessed: 2016-18-01. Available from: http://www.uniprot.org/uniprot.
- 61.
RefSeq Protein. Accessed: 2016-18-01. Available from: http://www.ncbi.nlm.nih.gov/protein.
- 62.
QuickGO. Accessed: 2016-18-01. Available from: http://www.ebi.ac.uk/QuickGO.
- 63.
WikiGenes. Accessed: 2016-18-01. Available from: http://www.wikigenes.org/e/gene.
- 64.
GeneTes. Accessed: 2016-18-01. Available from: https://www.genetests.org/genes.
- 65.
BioGPS. Accessed: 2016-18-01. Available from: http://biogps.org.
- 66.
GENATLAS. Accessed: 2016-18-01. Available from: http://genatlas.medecine.univ-paris5.fr.
- 67.
GeneCards. Accessed: 2016-18-01. Available from: http://www.genecards.org.
- 68.
GOPubmed. Accessed: 2016-18-01. Available from: http://www.gopubmed.org/web/gopubmed.
- 69.
H_InvDB. Accessed: 2016-18-01. Available from: http://biodb.jp.