VAT

From GersteinInfo

Revision as of 17:39, 28 March 2011

Introduction

The Variant Annotation Tool (VAT) consists of a set of modules to annotate genetic variants including SNPs, indels, and SVs. This software package also contains a program to aggregate these variants at the gene level. Subsequently, an image is generated for each gene to visualize the functional impact of these variants. This information can then be viewed and shared using a web-interface. In addition to annotation of the coding variants, this tool also integrates allele frequencies and genotype data providing population-specific information from published high quality variation databases such as 1000 Genomes Project.

Data formats

VCF

The Variant Call Format (VCF) is a tab-delimited text file format to represent a number of different genetic variants including single nucleotide polymorphisms (SNPs), small insertions and deletions (indels), and structural variants (SVs). This format was developed as part of the 1000 Genomes Project. A detailed summary of this file format can be found here. The annotation information is captured as part of the INFO field using the VA (Variant Annotation) tag. The string with the variant information has the following format:

AlleleNumber:GeneName:GeneId:Strand:Type:FractionOfTranscriptsAffected:{List of transcripts}TranscriptName:TranscriptId:TranscriptLength_RelativePositionOfVariant_RelativePositionOfAminoAcid_Substitution

The allele number refers to the numbering of the alleles. By definition, the reference allele has zero as the allele number, whereas the alternate alleles are numbered starting at one (some variants have more than one alternate alleles). The type refers to the type of variant. For SNPs, the types can take on the following values (generated by snpMapper): synonymous, nonsynonymous, prematureStop, removedStop, and spliceOverlap. For indels (generated by indelMapper), the types can take on the following values: spliceOverlap, startOverlap, endOverlap, insertionFS, insertionNFS, deletionFS, deletionNFS, where FS denotes 'frameshift' and NFS indicates 'non-frameshift'. The term spliceOverlap (for both SNPs and indels) refers to a genetic variant that overlaps with a splice site (either two nucleotides downstream of an exon or two nucleotides upstream of an exon).

Example 1: A SNP is introducing a premature stop codon. This variant affects one out of five transcripts for this gene.

chr1	23112837	.	A	T	.	PASS	AA=A;AC=7;AN=118;DP=168;SF=2;VA=1:EPHB2:ENSG00000133216:+:prematureStop:1/5:EPHB2-001:ENST00000400191:3165_3055_1019_K->*	

Example 2: A SNP leads to a non-synonymous substitution. This variant affects two out of four transcripts for this gene.

chr1	1110357	.	G	A	.	PASS	AA=G;AC=3;AN=118;DP=203;SF=2;VA=1:TTLL10:ENSG00000162571:+:nonsynonymous:2/4:TTLL10-001:ENST00000379288:1212_1187_396_R->H:TTLL10-202:ENST00000400931:1212_1187_396_R->H	

Example 3: A SNP causing a non-synonymous substitution in one transcript and a splice overlap in another transcript of the same gene.

chr9	35819390	rs2381409	C	T	.	PASS	AA=N;AC=157;AN=240;DP=49;SF=0,1;VA=1:TMEM8B:ENSG00000137103:+:nonsynonymous:1/7:TMEM8B-202:ENST00000360192:2109_166_56_P->S,1:TMEM8B:ENSG00000137103:+:spliceOverlap:1/7:TMEM8B-001:ENST00000450762:2106

Example 4: An indel with two alternate alleles. Each alternate allele leads to a non-frameshift deletion.

chr7	140118541	.	TACAACAACA	T,TACA	.	PASS	HP=1;VA=1:AC006344.1:ENSG00000236914:+:deletionNFS:1/1:AC006344.1-201:ENST00000434223:66_23_8_LQQQ->L,2:AC006344.1:ENSG00000236914:+:deletionNFS:1/1:AC006344.1-201:ENST00000434223:66_23_8_LQQ->L

Notice that multiple annotation entries are comma-separated. Multiple annotation entries arise when a variant causes different types of effects on different transcripts (Example 3) or if there are multiple alternate alleles (Example 4).

VAT also enables the grouping of samples. For examples, samples can be assigned to different sub-populations or they can be designated as cases or controls. This is done by modifying the header line using vcfModifyHeader. Specifically, the sample is prefixed by group identifier using the ':' character as a delimiter.

Interval

The Interval format consists of eight tab-delimited columns and is used to represent genomic intervals such as genes. This format is closely associated with the intervalFind module, which is part of BIOS. This module efficiently finds intervals that overlap with a query interval. The underlying algorithm is based on containment sublists: Alekseyenko, A.V., Lee, C.J. "Nested Containment List (NCList): A new algorithm for accelerating interval query of genome alignment and interval databases" Bioinformatics 2007;23:1386-1393 [1].

1.   Name of the interval
2.   Chromosome 
3.   Strand
4.   Interval start (with respect to the "+")
5.   Interval end (with respect to the "+")
6.   Number of sub-intervals
7.   Sub-interval starts (with respect to the "+", comma-delimited)
8.   Sub-interval end (with respect to the "+", comma-delimited)   

Note: For the purpose of VAT, the name field in the Interval file must contain four pieces of information delimited by the '|' symbol (geneId|transcriptId|geneName|transcriptName). Using gencode2interval program ensures proper formatting:

Example file:

ENSG00000008513|ENST00000319914|ST3GAL1|ST3GAL1-201	chr8	-	134472009	134488267	6	134472009,134474117,134475656,134477020,134478136,134487961	134472180,134474237,134475702,134477200,134478333,134488267
ENSG00000008513|ENST00000395320|ST3GAL1|ST3GAL1-202	chr8	-	134472009	134488267	6	134472009,134474117,134475656,134477020,134478136,134487961	134472180,134474237,134475702,134477200,134478333,134488267
ENSG00000008513|ENST00000399640|ST3GAL1|ST3GAL1-203	chr8	-	134472009	134488267	6	134472009,134474117,134475656,134477020,134478136,134487961	134472180,134474237,134475702,134477200,134478333,134488267
ENSG00000008516|ENST00000325800|MMP25|MMP25-201	chr16	+	3097544	3105947	4	3097544,3100009,3100254,3105830	3097548,3100145,3100546,3105947
ENSG00000008516|ENST00000336577|MMP25|MMP25-202	chr16	+	3096918	3109096	10	3096918,3097415,3100009,3100254,3107033,3107310,3107531,3108181,3108412,3108827	3097017,3097548,3100145,3100547,3107210,3107395,3107614,3108334,3108670,3109096

In this example, each interval (line) represents a transcript, while the sub-intervals denote exons. The geneId is utilized to determine if multiple transcripts belong to the same gene model.

Note: the coordinates in the Interval format are zero-based and the end coordinate is not included.

List of programs

VAT Core Modules

snpMapper

snpMapper is a program to annotate a set of SNPs in VCF format. The program determines the effect of a SNP on the coding potential (synonymous, nonsynonymous, prematureStop, removedStop, spliceOverlap) of each transcript of a gene.

Usage:

snpMapper <annotation.interval> <annotation.fa>

• Inputs: Takes a VCF input from STDIN
• Outputs: Outputs annotated SNPs in VCF format. The annotation information is captured as part of the INFO field. For details refer to the VCF format specification.
• Required arguments
  • annotation.interval - Annotation file representing the genomic coordinates of the gene models in Interval format. Each line in this file represents a transcript. This file is typically generated using the gencode2interval program.
  • annotation.fa - File with the transcript sequences in FASTA format for each entry specified in annotation.interval. This file is typically generated by the interval2sequences program using the 'exonic' mode.
• Optional arguments
  • None

indelMapper

indelMapper is a program to annotate a set of indels in VCF format. The program determines the effect of an indel on the coding potential (frameshift insertion, non-frameshift insertion, frameshift deletion, non-frameshift deletion, spliceOverlap, startOverlap, endOverlap) of each transcript of a gene.

Usage:

indelMapper <annotation.interval> <annotation.fa>

• Inputs: Takes a VCF input from STDIN
• Outputs: Outputs annotated indels in VCF format. The annotation information is captured as part of the INFO field. For details refer to the VCF format specification.
• Required arguments
  • annotation.interval - Annotation file representing the genomic coordinates of the gene models in Interval format. Each line in this file represents a transcript. This file is typically generated using the gencode2interval program.
  • annotation.fa - File with the transcript sequences in FASTA format for each entry specified in annotation.interval. This file is typically generated by the interval2sequences program using the 'exonic' mode.
• Optional arguments
  • None

genericMapper

genericMapper is a program to annotate an number of different variants in VCF format. The program checks whether a variant overlaps with entries in the specified annotation set (it does not determine the effect on the coding potential).

Usage:

genericMapper <annotation.interval> <nameFeature>

• Inputs: Takes a VCF input from STDIN
• Outputs: Outputs the annotated variants in VCF format. The annotation information is captured as part of the INFO field; only includes the alleleNumber and the name of the overlapping interval entry.
• Required arguments
  • annotation.interval - Annotation file representing the genomic coordinates of the gene models in Interval format. This can be a generic Interval.
  • nameFeature - Specifies the type of the annotation feature (for example promotor regions). The name of the feature is included as part of the annotation information (in the INFO field) in the resulting VCF file.
• Optional arguments
  • None

vcfSummary

vcfSummary is a program to aggregate annotated variants across genes and samples.

Usage:

vcfSummary <file.vcf.gz> <annotation.interval>

• Inputs: None
• Outputs: Generates two output files. The first file, named file.geneSummary.txt, contains the number of variants categorized by type for each gene. A second file, named file.sampleSummary.txt, summarizes number of variants categorized by type for each sample.
• Required arguments
  • file.vcf.gz - VCF file with annotated variants (this can be a mixture of indels and SNPs). This file must be compressed using bgzip and indexed using the tabix program.
  • annotation.interval - Annotation file representing the genomic coordinates of the gene models in Interval format. Each line in this file represents a transcript. This file is typically generated using the gencode2interval program.
• Optional arguments
  • None

vcf2images

vcf2images is a program to generate an image for each gene to visualize effect of the annotated variants.

Usage:

vcf2images <file.vcf.gz> <annotation.interval> <outputDir>

• Inputs: None
• Outputs: Generates an image in PNG format for each gene that has at least one annotated variant.
• Required arguments
  • file.vcf.gz - VCF file with annotated variants (this can be a mixture of indels and SNPs). This file must be compressed using bgzip and indexed using the tabix program.
  • annotation.interval - Annotation file representing the genomic coordinates of the gene models in Interval format. Each line in this file represents a transcript. This file is typically generated using the gencode2interval program.
  • outputDir - The output directory where the images are stored
• Optional arguments
  • None

vcfSubsetByGene

vcfSubsetByGene is a program to subset a VCF file with annotated variants by gene.

Usage:

vcfSubsetByGene <file.vcf.gz> <annotation.interval> <outputDir>

• Inputs: None
• Outputs: Generates a VCF file for each gene that has at least one annotated variant.
• Required arguments
  • file.vcf.gz - VCF file with annotated variants (this can be a mixture of indels and SNPs). This file must be compressed using bgzip and indexed using the tabix program.
  • annotation.interval - Annotation file representing the genomic coordinates of the gene models in Interval format. Each line in this file represents a transcript. This file is typically generated using the gencode2interval program.
  • outputDir - The output directory where VCF files are stored
• Optional arguments
  • None

vcfModifyHeader

vcfModifyHeader is a program to modify the header line (part of the meta-lines) in a VCF file. Specifically, it assigns each sample to a group or population (these assignments are used by other programs including vcfSummary).

Usage:

vcfModifyHeader <oldHeader.vcf> <groups.txt>

• Inputs: None
• Outputs: Generates a VCF header file.
• Required arguments
  • oldHeader.vcf - The meta lines of a VCF file. It can be obtained by using the following command:

grep '#' file.vcf > file.header.vcf

• • groups.txt - This tab-delimited file that assigns each sample present in the VCF to a group/population. Here is a small sample file:

HG00629	CHS
HG00634	CHS
HG00635	CHS
HG00637	PUR
HG00638	PUR
HG00640	PUR
NA06984	CEU
NA06985	CEU
NA06986	CEU
NA06989	CEU
NA06994	CEU

• Optional arguments
  • None

Auxiliary programs

gencode2interval

gencode2interval converts a GENCODE annotation file (in GTF format) to the Interval format.

Usage:

gencode2interval

• Inputs: Takes a GENCODE annotation file in GTF format from STDIN
• Outputs: Outputs the GENCODE annotation file in Interval format to STDOUT
• Required arguments
  • None
• Optional arguments
  • None

Note: To obtain the coding sequences of the elements with gene_type protein_coding and transcript_type protein_coding the following command should be used:

awk '/\t(HAVANA|ENSEMBL)\tCDS\t/ {print}' gencode.v3c.annotation.GRCh37.gtf | grep 'gene_type "protein_coding"' | grep 'transcript_type "protein_coding"' > gencode.v3c.annotation.GRCh37.cds.gtpc.ttpc.gtf
gencode2interval < encode.v3c.annotation.GRCh37.cds.gtpc.ttpc.gtf > encode.v3c.annotation.GRCh37.cds.gtpc.ttpc.interval

interval2sequences

Module to retrieve genomic/exonic sequences for an annotation set in Interval format.

Usage:

interval2sequences <file.2bit> <file.annotation> <exonic|genomic>

• Inputs: None
• Outputs: Reports the extracted sequences in FASTA format
• Required arguments
  • file.2bit - genome reference sequence in 2bit format
  • file.annotation - annotation set in Interval format (each line represents one transcript)
  • < exonic | genomic > - exonic means that only the exonic regions are extracted, while genomic indicates that the intronic sequences are extracted as well
• Optional arguments
  • None

External programs

bgzip/tabix

Tabix is generic tool that indexes position-sorted files in tab-delimited formats to facilitate fast retrieval. This tool was developed by Heng Li. For more information consult the tabix documentation page.

VCF tools

VCF tools consists of a suite of very useful modules to manipulate VCF files. For more information consult the documentation page.

Example workflow

This workflow shows how the 1000 Genomes Project, Phase I, chr22, SNP calls data set was processed.

Prerequisites

Download the GENCODE annotation set (version 3c, hg19):

$ wget ftp://ftp.sanger.ac.uk/pub/gencode/release_3c/gencode.v3c.annotation.GRCh37.gtf.gz

Download the human genome (hg19) in 2bit format. This is used by interval2sequences to extract the genomic sequences for the entries specified in the annotation set:

$ wget http://hgdownload.cse.ucsc.edu/goldenPath/hg19/bigZips/hg19.2bit

Download the SNP files in VCF format and a third file that assigns each sample to a population:

$ wget ftp://ftp-trace.ncbi.nih.gov/1000genomes/ftp/release/20100804/ALL.2of4intersection.20100804.genotypes.vcf.gz
$ wget ftp://ftp-trace.ncbi.nih.gov/1000genomes/ftp/release/20100804/ALL.2of4intersection.20100804.genotypes.vcf.gz.tbi
$ wget ftp://ftp-trace.ncbi.nih.gov/1000genomes/ftp/release/20100804/20100804.ALL.panel

Extract variants on chromosome 22:

$ tabix -h ALL.2of4intersection.20100804.genotypes.vcf.gz 22 | bgzip -c > ALL.2of4intersection.20100804.chr22.genotypes.vcf.gz

Preprocessing of the annotation file

Decompress the annotation file:

$ gunzip gencode.v3c.annotation.GRCh37.gtf.gz

Extract the coding sequence (CDS) elements where the both the gene_type and transcript_type are protein_coding:

$ awk '/\t(HAVANA|ENSEMBL)\tCDS\t/ {print}' gencode.v3c.annotation.GRCh37.gtf | grep 'gene_type "protein_coding"' | grep 'transcript_type "protein_coding"' > gencode.v3c.annotation.GRCh37.cds.gtpc.ttpc.gtf

Convert the GENCODE GTF file into Interval format:

$ gencode2interval < gencode.v3c.annotation.GRCh37.cds.gtpc.ttpc.gtf > gencode.v3c.annotation.GRCh37.cds.gtpc.ttpc.interval

Retrieve the genomic sequences for the transcripts specified in the annotation file.

$ interval2sequences hg19.2bit gencode.v3c.annotation.GRCh37.cds.gtpc.ttpc.interval exonic > gencode.v3c.annotation.GRCh37.cds.gtpc.ttpc.fa

Annotation of the SNPs

Annotate the variants using snpMapper

$ zcat ALL.2of4intersection.20100804.chr22.genotypes.vcf.gz | snpMapper gencode.v3c.annotation.GRCh37.cds.gtpc.ttpc.interval gencode.v3c.annotation.GRCh37.cds.gtpc.ttpc.fa > ALL.2of4intersection.20100804.chr22.genotypes.annotated.vcf

Modification the VCF header line

Modify the VCF header line to assign individual samples to populations (groups). This is done by using the following syntax: group:sample (i.e. CEU:NA0705).

First get the old meta-data lines:

$ grep "#" ALL.2of4intersection.20100804.chr22.genotypes.annotated.vcf > ALL.2of4intersection.20100804.chr22.genotypes.annotated.oldHeader.vcf 

Store the annotated variants in a separate file:

$ grep "#" -v ALL.2of4intersection.20100804.chr22.genotypes.annotated.vcf > ALL.2of4intersection.20100804.chr22.genotypes.annotated.variants.vcf

Create the new meta-data lines:

$ vcfModifyHeader ALL.2of4intersection.20100804.chr22.genotypes.annotated.oldHeader.vcf 20100804.ALL.panel > ALL.2of4intersection.20100804.chr22.genotypes.annotated.newHeader.vcf 

Merge the new meta-data lines with the annotated variants and create a new file called ALL.2of4intersection.20100804.chr22.vcf:

$ cat ALL.2of4intersection.20100804.chr22.genotypes.annotated.newHeader.vcf ALL.2of4intersection.20100804.chr22.genotypes.annotated.variants.vcf > ALL.2of4intersection.20100804.chr22.vcf

Compress the newly created VCF file with the annotated variants:

$ bgzip ALL.2of4intersection.20100804.chr22.vcf

Index the newly created VCF file with the annotated variants:

$ tabix -p vcf ALL.2of4intersection.20100804.chr22.vcf.gz

Generation of summaries and images

Generate gene and sample summaries for the annotated variants

$ vcfSummary ALL.2of4intersection.20100804.chr22.vcf.gz gencode.v3c.annotation.GRCh37.cds.gtpc.ttpc.interval

Resulting files: ALL.2of4intersection.20100804.chr22.geneSummary.txt and ALL.2of4intersection.20100804.chr22.sampleSummary.txt

Make a new directory to store the images and VCF files for each gene.

$ mkdir ALL.2of4intersection.20100804.chr22

Generate an image for each gene with at least one annotated variant.

$ vcf2images ALL.2of4intersection.20100804.chr22.vcf.gz gencode.v3c.annotation.GRCh37.cds.gtpc.ttpc.interval ./ALL.2of4intersection.20100804.chr22

Subset the VCF file with the annotated variants by gene.

$ vcfSubsetByGene ALL.2of4intersection.20100804.chr22.vcf.gz gencode.v3c.annotation.GRCh37.cds.gtpc.ttpc.interval ./ALL.2of4intersection.20100804.chr22

Setting up the web server

Make a TAR ball of the relevant files:

• Directory with the images and the VCF files for each gene (ALL.2of4intersection.20100804.chr22)
• File with the gene summary (ALL.2of4intersection.20100804.chr22.geneSummary.txt)
• File with the sample summary (ALL.2of4intersection.20100804.chr22.sampleSummary.txt)
• Compressed VCF file with the annotated variants (ALL.2of4intersection.20100804.chr22.vcf.gz)
• Index file of the annotated variants (ALL.2of4intersection.20100804.chr22.vcf.gz.tbi)

$ tar -cvf ALL.2of4intersection.20100804.chr22.tar \
   ALL.2of4intersection.20100804.chr22 \
   ALL.2of4intersection.20100804.chr22.geneSummary.txt \
   ALL.2of4intersection.20100804.chr22.sampleSummary.txt \
   ALL.2of4intersection.20100804.chr22.vcf.gz \
   ALL.2of4intersection.20100804.chr22.vcf.gz.tbi

Copy the relevant files to the web server

$ scp ALL.2of4intersection.20100804.chr22.tar lh372@gw.gersteinlab.org:/nfs/web/lh372/public_html/VAT

Unpack the TAR on the web server

$ tar -xvf ALL.2of4intersection.20100804.chr22.tar 

View the results: Link to web server