UGP Variant Pipeline 1.2.1

Utah Genome Project

Nov. 2014 
Variant Calling Pipeline  Version 1.2.1

Software Versions

• cApTUrE is a lightweight NGS pipeline, created for the
 Utah Genome Project (UGP)
• BWA: 0.7.10
• Picard: 1.127 (Broad version)
• GATK: 3.3-0
• SamTools: 0.2.0
• FastQC v0.10.1

Data Source

Data sets used for the variant calling pipeline come from the Broad GSA (GATK) group as the 'GATK resource bundle 2.5' version 2.8

wget -r ftp://gsapubftp-anonymous@ftp.broadinstitute.org/bundle/2.5/b37

Reference Genome (GRCh37):

• human_g1k_v37_decoy.fasta

Call region file generated from NCBI

• GRCh37 GFF3

VCF files for RealignerTargetCreator knowns and dbsnp for BaseRecalibrator.

• known_indel: Mills_and_1000G_gold_standard.indels.b37.vcf
• known_indel: 1000G_phase1.indels.b37.vcf
• known_dbsnp: dbsnp_137.b37.vcf

Background Files

• We have created 1000Genomes (BWA mem/GATK 3.0+) background files to be ran concurrently with the GenotypeGVCFs step.

Groups Currently completed:

• CEU
• GBR
• FIN

Version 1.0.5 background files have been updated to show only the indviduals of each group, not the file names.

This is a complete list of the background individuals for run completed > 1.0.5 [1]

If you would like a copy of the current files, we have made a public AWS s3 bucket
Using s3cmd execute the following command: 
s3cmd get s3://ugp-1k-backgrounds --recursive

Resource files for VariantRecalibrator_SNP

• hapmap_3.3.b37.vcf
• 1000G_omni2.5.b37.vcf
• 1000G_phase1.snps.high_confidence.b37.vcf

Resource files for VariantRecalibrator_INDEL

• Mills_and_1000G_gold_standard.indels.b37.vcf
• 1000G_phase1.indels.b37.vcf

Sequencing

This pipeline is designed for 100 bp (or greater) Illumina HiSeq PE exome or WGS sequence data with Sanger/Illumina 1.9 quality encoding, and uses Illumina naming convention found here [2]

Validate File Integrity with md5sum

An md5sum signature should be provided for each FastQ file by the sequencing center. After the file has been downloaded, locally check the md5sum to be sure that no data corruption occurred during the file transfer.

md5sum file.fastq > file_local.md5
diff file_local.md5 file_provided.md5
Now the pipeline runs md5_check to validation the results and will quit if errors are found. 

If the md5sum signature differs from that provided for the file:

• Check to be sure you have the correct file.
• Check if the md5sum was calculated on that compressed or uncompressed file by the provider and be sure to do the same with the local copy.
• Try the download again.
• Contact the sequence provider.

FastQ File Analyses

fastqc Sample1_L1_R1.txt

From the sumamry.txt report we check

• FAIL sections

From the fastqc_data.txt file we check the following values:

• Encoding (must be Sanger / Illumina 1.9)
• Total Sequences (Currently set to 30000000)
• Filtered Sequences (Currently set to less then 5)
• Sequence length (must be >= 100 bp)
• %GC (45 < x < 55)
• Total Duplicate Percentage (Currently set to 60.0)

The pipeline now runs fastqc_check and output these result into QC-report.txt.

Indexing

The following indexing is required using BWA, Picard and SamTools. GATK requires all three. However this step only needs to be done once "per-machine".

• BWA

bwa index -a bwtsw human_g1k_v37_decoy.fasta

• Picard

java -jar CreateSequenceDictionary.jar R=human_g1k_v37_decoy.fasta O=human_g1k_v37_decoy.dic

• SamTools

samtools faidx human_g1k_v37_decoy.fasta

Alignment

Align reads to the genome with bwa.

The 'BWA-mem' program will find the reference coordinates of the input reads (independent of their mate-pair). The following parameters are those used by the 1KG project and GATK for aligning Illumina data.

bwa mem -M -R "read group" human_g1k_v37_decoy.fasta Sample1_L1_R1.fq Sample1_L1_R2.fq | samtools view -bSho BAM_FILE -

BAM File Analyses and Processing

Alignment BAM files are improved in various ways to help increase the quality and speed of subsequent variant calling steps.

BAM Sorting

java -jar -XX:ParallelGCThreads=# -Xmx#g -Djava.io.tmpdir=/tmp /usr/local/picard/dist/picard.jar SortSam 
   INPUT=*.bam 
   OUTPUT=*_sorted.bam 
   VALIDATION_STRINGENCY=LENIENT 
   SORT_ORDER=coordinate 
   MAX_RECORDS_IN_RAM=5000000 
   CREATE_INDEX=True 
   &> SortSam.log-#

Merge lane level BAMs to individual

• This step only needs to run if you have multiple lanes per sample.

java -jar -Xmx#g -XX:ParallelGCThreads=# -Djava.io.tmpdir=/tmp /usr/local/picard/dist/picard.jar MergeSamFiles.jar
    INPUT=#*_sorted.bam
    INPUT=#*_sorted.bam
    INPUT=[ ... ]
    OUTPUT=*.bam                          
    VALIDATION_STRINGENCY: LENIENT
    MAX_RECORDS_IN_RAM: 5000000
    CREATE_INDEX: True
    SORT_ORDER: coordinate
    ASSUME_SORTED: True 
    USE_THREADING: True
    &> MergeSamFiles.log-#

Mark Duplicates

Remove PCR/Optical duplicate reads

java -jar -Xmx#g -XX:ParallelGCThreads=# -Djava.io.tmpdir=/tmp /usr/local/picard/dist/picard.jar MarkDuplicates 
   INPUT=*_sorted.bam 
   OUTPUT=*_sorted_Dedup.bam 
   METRICS_FILE=*_sorted_Dedup.metrics 
   VALIDATION_STRINGENCY=LENIENT 
   MAX_RECORDS_IN_RAM=5000000 
   ASSUME_SORTED=True 
   CREATE_INDEX=True  
   &> MarkDuplicates.log-#

Currently all duplicates are flagged to allow GATK to handle them.

BAM Quality Control

At this point the pipeline has broken the tasks into chromosomal regions.

• CollectMultipleMetrics

java -jar -Xmx#g -XX:ParallelGCThreads=# -Djava.io.tmpdir=/tmp /usr/local/picard/dist/picard.jar CollectMultipleMetrics
   INPUT=*_sorted_Dedup_realign_chr#_recal.bam
   OUTPUT=*_sorted_Dedup_realign_chr#_recal.metrics 
   VALIDATION_STRINGENCY=LENIENT 
   PROGRAM=QualityScoreDistribution  
   REFERENCE_SEQUENCE=human_g1k_v37_decoy.fasta 
   &> CollectMultipleMetrics.log-#

• idxstats

samtools idxstats [dedup bam files] > dedup-bamfile.stats

Now the pipeline will take idxstats ouput and check for unmapped reads.

Local Realignment of Indels

• RealignerTargetCreator

This is where the tasks are broken into chromosomal regions.

java -jar -Xmx#g -XX:ParallelGCThreads=# -Djava.io.tmpdir=/tmp GenomeAnalysisTK.jar 
   -T RealignerTargetCreator 
   -R human_g1k_v37_decoy.fasta 
   -I *_sorted_Dedup.bam 
   --num_threads #  
   --known Mills_and_1000G_gold_standard.indels.b37.vcf
   --known 1000G_phase1.indels.b37.vcf  
   -L chr#_region_file.list 
   -o *_chr#_realign.intervals 
   &> RealignerTargetCreator.log-#

• IndelRealigner

java -jar -Xmx#g -XX:ParallelGCThreads=# -Djava.io.tmpdir=/tmp GenomeAnalysisTK.jar 
   -T IndelRealigner 
   -R human_g1k_v37_decoy.fasta 
   -I *_sorted_Dedup.bam 
   -L chr#_region_file.list
   -targetIntervals *_chr#_realign.intervals 
   -known Mills_and_1000G_gold_standard.indels.b37.vcf 
   -known 1000G_phase1.indels.b37.vcf 
   -o *_sorted_Dedup_realign_chr#.bam 
   &> IndelRealigner.log-1

BaseRecalibration & PrintReads

• BaseRecalibrator

java -jar -Xmx#g -XX:ParallelGCThreads=# -Djava.io.tmpdir=/tmp GenomeAnalysisTK.jar 
   -T BaseRecalibrator 
   -R /human_g1k_v37_decoy.fasta 
   -I *_sorted_Dedup_realign_chr#.bam 
   --num_cpu_threads_per_data_thread #  
   --knownSites dbsnp_137.b37.vcf
   --knownSites Mills_and_1000G_gold_standard.indels.b37.vcf 
   --knownSites 1000G_phase1.indels.b37.vcf  
   -o *_sorted_Dedup_realign_chr#_recal_data.table 
   &> BaseRecalibrator.log-#

• PrintReads

java -jar -Xmx#g -XX:ParallelGCThreads=# -Djava.io.tmpdir=/tmp GenomeAnalysisTK.jar 
   -T PrintReads 
   -R human_g1k_v37_decoy.fasta 
   -I *_sorted_Dedup_realign_chr#.bam
   --num_cpu_threads_per_data_thread #  
   -BQSR *_sorted_Dedup_realign_chr#_recal_data.table 
   -o *_sorted_Dedup_realign_chr#_recal.bam 
   &> PrintReads.log-#

Variant Calling

HaplotypeCaller

• Now HaplotypeCaller handels SNP and INDEL calls.

java -jar -Xmx#g -XX:ParallelGCThreads=# -Djava.io.tmpdir=/tmp GenomeAnalysisTK.jar 
   -T HaplotypeCaller 
   -R human_g1k_v37_decoy.fasta 
   --min_base_quality_score 20 
   --variant_index_parameter 128000 
   --emitRefConfidence GVCF 
   --standard_min_confidence_threshold_for_calling 30.0 
   --num_cpu_threads_per_data_thread # 
   --variant_index_type LINEAR 
   --standard_min_confidence_threshold_for_emitting 30.0  
   -I *_sorted_Dedup_realign_chr*_recal.bam 
   -L chr#_region_file.list
   -o chr#_region_*.raw.snps.indels.gvcf 
   &> HaplotypeCaller.log-#

CatVariants

Collect all individual gvcf files.

java -cp GenomeAnalysisTK.jar org.broadinstitute.gatk.tools.CatVariants 
   -R human_g1k_v37_decoy.fasta 
   --assumeSorted  
   -V chr#_region_*.raw.snps.indels.vcf 
   -V [ ... ]
   -out *.raw.snps.indels.gCat.vcf 
   &> CatVariants.log-*

CombineGVCFs

Collect all the chromosomal cat files.

java -jar -Xmx#g -XX:ParallelGCThreads=# GenomeAnalysisTK.jar  
   -T CombineGVCFs 
   -R human_g1k_v37_decoy.fasta 
   --variant *.raw.snps.indels.gCat.vcf
   --variant [ ... ]
   -o cApTUrE_*_final_mergeGvcf.vcf 
   &> CombineGVCF.log-*

GenotypeGVCFs

java -Xmx10g -Djava.io.tmpdir=/tmp GenomeAnalysisTK.jar
  -T GenotypeGVCFs
  -R human_g1k_v37_decoy.fasta
  -variant [sampe_mergeGvcf.vcf]
  -variant [1000G pre-combined .gvcf files]
  -o [sample+background_genotyped.vcf]
  &> GATK_GenotypeGVCF.log
  -o [sampe_mergeGvcf.vcf]

VariantRecalibrator

• SNP Recalibration

java -Xmx10g -Djava.io.tmpdir=/tmp GenomeAnalysisTK.jar
  -T VariantRecalibrator
  -R human_g1k_v37_decoy.fasta
  -resource:hapmap,known=false,training=true,truth=true,prior=15.0 /data/GATK_Bundle/hapmap_3.3.b37.vcf 
  -resource:omni,known=false,training=true,truth=true,prior=12.0 /data/GATK_Bundle/1000G_omni2.5.b37.vcf 
  -resource:1000G,known=false,training=true,truth=false,prior=10.0 /data/GATK_Bundle/1000G_phase1.snps.high_confidence.b37.vcf 
  -an QD -an MQRankSum -an ReadPosRankSum -an FS
  -input [sample+background_genotyped.vcf]
  -recalFile [sample_snp_recal]
  -tranchesFile [sample_snp_tranches]
  -rscriptFile [sample_snp_plots.R]
  -mode SNP 
  &> GATK_VariantRecalibrator_SNP.log

• INDEL Recalibration

java -Xmx10g -Djava.io.tmpdir=/tmp GenomeAnalysisTK.jar
  -T VariantRecalibrator
  -R human_g1k_v37_decoy.fasta
  -resource:mills,known=false,training=true,truth=true,prior=12.0 /data/GATK_Bundle/Mills_and_1000G_gold_standard.indels.b37.vcf 
  -resource:1000G,known=false,training=true,truth=true,prior=10.0 /data/GATK_Bundle/1000G_phase1.indels.b37.vcf 
  -an MQRankSum -an ReadPosRankSum -an FS
  -numBadVariants: 5000
  -mode INDEL
  -input [sample+background_genotyped.vcf]
  -recalFile [sample_indel_recal]
  -tranchesFile [sample_indel_tranches]
  -rscriptFile [sample_indel_plots.R]
  -mode INDEL

ApplyRecalibration

• SNP Apply

java -Xmx10g -Djava.io.tmpdir=/tmp GenomeAnalysisTK.jar
  -T ApplyRecalibration
  -input [genotyped.vcf]
  -o [recal_SNP.vcf]
  -R human_g1k_v37_decoy.fasta
  -ts_filter_level 99.0
  -excludeFiltered
  -tranchesFile [sample.snp.tranches]
  -recalFile [sample.snp.recal]
  -mode SNP 
  &> ApplyRecalibration_SNP.report

• INDEL Apply

java -Xmx10g -Djava.io.tmpdir=/tmp GenomeAnalysisTK.jar
  -T ApplyRecalibration
  -input [genotyped.vcf]
  -o [recal_INDEL.vcf]
  -R human_g1k_v37_decoy.fasta
  -ts_filter_level 99.0
  -excludeFiltered
  -tranchesFile [sample.indel.tranches]
  -recalFile [sample.indel.recal]
  -mode INDEL
  &> ApplyRecalibration_INDEL.report

CombineVarients

These command will combine INDEL and SNP files into a single VCF file.

• CombineVarients

java -Xmx10g -Djava.io.tmpdir=/tmp GenomeAnalysisTK.jar
  -T CombineVariants
  -R human_g1k_v37_decoy.fasta
  --variant cleaned_recal_SNP.file
  --variant cleaned_recal_INDEL.file
  -o [sample+background_Final.vcf]
  &> CombineVarients.report

Variant File QC

Quality Metrics on variants

• Ti/Tv Ratio (2.1 for WGS ~2.8 for exome)
• HapMap concordance
• SNV/Indel Counts
• Rare variant enrichment
• DP
• Q
• GQ