Skip to content

Latest commit

 

History

History
172 lines (144 loc) · 6.3 KB

resource_bundle.md

File metadata and controls

172 lines (144 loc) · 6.3 KB
Raw

GATK Resource Bundle

Introduction

The GATK resource bundle is a collection of standard files for working with human resequencing data with the GATK.

Bundle Access

The bundles are available on the GATK public FTP server.

  • If you're using an FTP client (e.g. FileZilla, Cyberduck etc.), you can download from ftp.broadinstitute.org, using the username gsapubftp-anonymous and leaving the password blank.
  • If you're using a web browser (e.g. Chrome, FireFox), you can download from ftp://ftp.broadinstitute.org/bundle.

Bundle contents

The bundle directory contains a folder for each release (e.g. 2.8). Inside each folder there are four other directories: b36, b37, hg18 and hg19.

bundle
|--2.8
  |--b36
  |--b37
  |--hg18
  |--hg19
  |--exampleFASTA
|--2.5
  |--b36
  |--b37
  |--hg18
  |--hg19
  |--exampleFASTA

These are the contents of b37 (hg19 is very similar, except that the reference sequence is called ucsc.hg19.fasta.gz and the dbsnp_138 version is used):

Reference Sequence
|--human_g1k_v37.fasta.gz
   human_g1k_v37.fasta.fai.gz
   human_g1k_v37.dict.gz
|--human_g1k_v37_decoy.fasta.gz
   human_g1k_v37_decoy.fasta.fai.gz
   human_g1k_v37_decoy.dict.gz

dbSNP recent release
|--dbsnp_137.b37.vcf.gz
   dbsnp_137.b37.vcf.idx.gz

dbSNP with sites discovered in or before dbSNPBuildID 129, which excludes the
impact of the 1KG project and is useful for evaluation of dbSNP rate and Ti/Tv
values at novel sites.
|--dbsnp_137.b37.excluding_sites_after_129.vcf.gz
   dbsnp_137.b37.excluding_sites_after_129.vcf.idx.gz

HapMap genotypes and sites
|--hapmap_3.3.b37.vcf.gz
   hapmap_3.3.b37.vcf.idx.gz

OMNI 2.5 genotypes for 1KG samples and sites
|--1000G_omni2.5.b37.vcf.gz
   1000G_omni2.5.b37.vcf.idx.gz

The current best set of known indels for local realignment (note that dbSNP
isn't used for this anymore)
|--1000G_phase1.indels.b37.vcf.gz
   1000G_phase1.indels.b37.vcf.idx.gz
|--Mills_and_1000G_gold_standard.indels.b37.vcf.gz
   Mills_and_1000G_gold_standard.indels.b37.vcf.idx.gz

Large-scale standard single sample BAM file for testing
|--NA12878.HiSeq.WGS.bwa.cleaned.raw.subset.b37.sites.vcf.gz
   NA12878.HiSeq.WGS.bwa.cleaned.raw.subset.b37.sites.vcf.idx.gz
   NA12878.HiSeq.WGS.bwa.cleaned.raw.subset.b37.vcf.gz
   NA12878.HiSeq.WGS.bwa.cleaned.raw.subset.b37.vcf.idx.gz
   NA12878.knowledgebase.snapshot.20131119.b37.vcf.gz
   NA12878.knowledgebase.snapshot.20131119.b37.vcf.idx.gz
|--CEUTrio.HiSeq.WGS.b37.NA12878.bam
   CEUTrio.HiSeq.WGS.b37.NA12878.bam.bai.gz
   CEUTrio.HiSeq.WGS.b37.NA12878.vcf.gz
   CEUTrio.HiSeq.WGS.b37.NA12878.vcf.idx.gz

(Not mentioned)
|--1000G_phase1.snps.high_confidence.b37.vcf.gz
   1000G_phase1.snps.high_confidence.b37.vcf.idx.gz
|--Broad.human.exome.b37.interval_list.gz
|--CEUTrio.HiSeq.WGS.b37.bestPractices.phased.b37.vcf.gz
   CEUTrio.HiSeq.WGS.b37.bestPractices.phased.b37.vcf.idx.gz

Differences between b37 and hg19

The Genome Reference Consortium (GRC) periodically releases a version of the human genome e.g. in February 2009 they released "build 37", which is known as GRCh37.

UCSC Version Release Date Release Name
hg18 Mar 2006 NCBI b36
hg19 Feb 2009 GRCh37
hg38 Dec 2013 GRCh38

Each version includes the following sequences:

  • 24 "relatively complete" sequences for chromosomes 1-22, X and Y;
  • A complete mitochondrial sequence;
  • Several "unlocalised sequences" - known chromosome, unknown coordinates;
  • Several "unplaced sequences" - known from human, unknown chromosome;
  • Several "alternate loci" - sequences with alternate representations of specific human regions;

The GRC didn't provide a set of naming rules for these sequences, so different teams adopted different rules:

  • The b37 conventions were used by the 1KG Project. GATK and IGV use this name. The ENSEMBL genome browser, the NCBI dbSNP database (in VCF files) and the Sanger COSMIC database (in VCF files) are among those preferring these conventions. These are also the preferred standards for new projects.
  • The hg19 conventions were used by the UCSC genome browser. Despite the old mitochondrial sequence, the nonstandard naming and the inclusion of alternate loci (which are undesirable for read mapping), hg19 has gained popularity due to its exposure via the UCSC genome browser, and is often the convention used by vendors when reporting exome enrichment kit coordinates.

In summary the differences are:

  1. Naming conventions:
  • b37: 1-22, X, Y, MT;
  • hg19: chr1-chr22, chrX, chrY, chrM;
  1. Mitochondrial sequence:
  • b37 has an updated one;
  • hg19 has an older one from build 36;
  1. Unlocalised and unplaced sequences:
  • b37: named after their accession numbers e.g. "GL000191.1";
  • hg19: given custom names e.g. "chr1_gl000191_random", "chrUn_gl000221";
  1. Alternate loci:
  • b37: not included;
  • hg19: included;

Which version should I use?

It's suggested to use the extended "decoy" version of b37 which includes human herpesvirus sequence (Epstein-Barr virus (EBV), NC_007605) and a decoy sequence derived from HuRef (Craig Venter genome), human BAC and Fosmid clones, and NA12878 (named "hs37d5"). It also has the pseudo-autosomal regions (PAR) of chromosome Y masked out (replaced with "N") so that the respective regions in chromosome X may be treated as diploid. In summary:

  • used by the 1KG in Phase II;
  • provides better results due to the PAR masking and the addition of the decoy sequences;
  • supported by GATK and IGV;
  • compatible with all the annotations (dbSNP etc.) that are reported using the "b37" conventions.

Resources