---

Fellow data members: Please send comments / updates / suggestions to the Issues section

• June 2017: WES QC Process (last updated June 2017)
  1. Files
  2. QC comparisons
  3. Sample QC Parameters
  4. Variant QC Parameters
  5. QC pipeline steps
• QC pipeline examples

WES QC process

Files

• BAM_metrics.R - a quick way to get relevant BAM metrics from your VCF samples sequenced at the Broad
• DBS pipeline 2017-03-21.py - Kyle Satterstrom's DBS pipeline python script
• pruned_9k_common_variants_t.bim - TJ Singh's 9k common variant list (hg19) for getting ancestry/relatedness from exome variants

QC comparisons

Categorical Groupings

• Cohorts/waves
• C-Project
• Capture platform
• Sex
• Affection status
• Project specific categories

Quantitative parameters

• Top Principal components
• % Coverage (callRate)
• Mean Depth (dpMean)
• Singleton Synonymous rate (nSingleton when restricting to synonymous variants)
• Non-ExAC / non-discovEHR singleton rate * 92k ExAC/GnomAD sites with Psych-exomes and MiGEN removed: gs://exome-qc/resources/gnomad.r2.0.1/gnomad.r2.0.1.nonpsych.variants.tsv
  • discovEHR sites /humgen/atgu1/fs03/shared_resources/discovEHR/GHS_Freeze_50.L3DP10.pVCF.frq.hail.vcf
  • ExAC sites /humgen/atgu1/fs03/shared_resources/ExAC/release0.3.1/ExAC.r0.3.1.sites.vep.vcf.gz
  • Non-Psych ExAC sites ftp://ftp.broadinstitute.org/distribution/ExAC_release/release1/subsets/ExAC_nonpsych.r1.sites.vep.vcf.gz

Sample QC Parameters

From GATK/Picard metadata

• Freemix Contamination
• Chimeric read percentage
• PCT_TARGET_BASES_10X
• PCT_TARGET_BASES_20X

From Hail Sample QC Annotation Table Primary QC parameters

• callRate
• rTiTv
• rHetHomVar
• rInsertionDeletion
• nSingleton
• dpMean

Secondary QC parameters

• nCalled
• nNotCalled
• nHomRef
• nHet
• nHomVar
• nSNP
• nInsertion
• nDeletion
• nTransition
• nTransversion
• dpStDev
• gqMean
• gqStDev
• nNonRef

Variant QC Parameters

Variant Quality

• VQSLOD - Variant Quality Score in LOD space (or new RF posterior probabilities)
• pHWE - Hard-Weinberg Equilibrium
• AC - allele count
• Median depth
• QD - quality by depth

Genotype Quality

• Depth
• PHRED likelihood (PL)
• Genotype Quality (GQ - same as PL in joint-called VCF)
• Allele Depth (AD)

Well-covered region intervals

• Regions that show minimum 10x depth in 80% of samples from 47 sequencing batches (20 are Agilent, 18 are ICE, 9 others) at the Broad
• gs://exome-qc/resources/coverage/agilent_regions_20170703.bed.bgz
• gs://exome-qc/resources/coverage/ice_regions_20170703.bed.bgz

QC pipeline steps

0. Phenotype file QC

   • GOAL: Understanding the phenotype data you are working with
   • List and understand all sample phenotypes provided
   • Match up phenotype file IDs with VCF IDs
     • Resolve any inconsistencies before moving forward
   • Start spreadsheet/table of datasets, capture platforms
     • Split data up into category with the most designations
   • Write up paragraph of sample collection and descriptives
   • Read VCF and sample file into Hail
   • Generate sample QC metrics from raw VCF

1. Pre-filtering

   • GOAL: Remove variants that are highly unlikely to be analyzed
   • Remove variants that fail VQSR
   • Remove variants outside the intersection of capture target intervals
   • Remove variants in low-complexity regions
   • Generate sample QC metrics from pre-filtered VCF
   • Create sites-only VDS and annotate with VEP (only need synonymous annotation for QC purposes)

2. Outlier sample QC: part 1

   • GOAL: Remove samples that are contaminated or have poor sequencing levels
   • Use pre-filtered VCF
   • Plot values below before using DEFAULT filters to ensure you are not throwing away large amounts of samples
   • freemix contamination filtering (DEFAULT > 5%)
   • Chimeric read filtering (DEFAULT > 1.4% )
   • Call rate filtering (DEFAULT < 90%)
   • Mean Depth coverage filtering (DEFAULT < 20)

3. Sex check

   • GOAL: remove samples where genotype sex does not equal reported sex
   • Filter out variants within PAR coordinates (https://en.wikipedia.org/wiki/Pseudoautosomal_region)
   • Reported males shoud have X chromosome F-statistic from 0.8 to 1
   • Reported females shoud have X chromosome F-statistic from -0.2 to 0.4
   • Remove samples with ambiguous imputed sex (X chromosome inbreeding coefficient between 0.4 and 0.8)
   • Large-scale sex check errors are indicative of ID mismatching or upstream data mishandling

4. Principal components analysis

   • GOAL: Determine general ancestry of cohort
   • Use raw VCF (so as not to exclude any common variants)
   • Subset to common variant VCF (use 9k set: pruned_9k_common_variants_t.bim or generate your own)
   • Run Hail PCA with 1K Genomes samples
   • Create plots of PCs
     • Case / control coloring
     • Cohort coloring
   • Assigning samples to a particular ancestry
     • Random forest approach (used by TJ Singh)
     • SNPWEIGHTS: https://www.hsph.harvard.edu/alkes-price/software/ (used by Chia-Yen Chen)

5. Outlier sample QC: part 2

   • GOAL: remove samples within cohort that have unusual variant properties
   • Use pre-filtered VCF
   • Examine per-cohort variation in:
     • TiTv ratio
     • Het/Hom ratio
     • Insertion/Deletion ratio
   • Plots with colors defined by assigned ancestry
     • Different ancestries have significant mean differences in these ratios
   • Filter out within cohort outliers (DEFAULT > 4 Std. deviations within a particular ancestry)

6. Principal components filtering

   • GOAL: match case and controls within a common genetic ancestry
   • Run Hail PCA without 1K Genomes samples
   • If retaining multiple ancestries, make sure to define ancestry groups in phenotype file
   • PCA filtering (no DEFAULT filtering parameters)
     • 2-dimensional centroid approach (using distance from 1KG ancestry mean PC; used by Chia-Yen Chen)
     • pair-matching cases to controls (R package: optmatch; R function: pairmatch(); used by Mitja Kurki)
   • Within each ancestry group, re-run PCA
     • Re-evaluate PC dispersion
     • Add these PCs as covariates to phenotype file

7. Identity-by-descent (IBD) filtering

   • GOAL: remove 1st and 2nd degree relatives from population based sample
   • Within each ancestry group, calculate IBD using common variant VDS
   • Plot proportion of 0 shared and 1 shared alleles
   • IBD filtering on PI-HAT value (DEFAULT > 0.2)

8. Variant QC

   • GOAL: Remove low quality/somatic variants
   • Use pre-filtered VCF with outlier samples removed
   • Filter variants with low call rate (DEFAULT < 95%)
     • Split variant call rate by capture, case/control status, or other category
   • Remove monoallelic variants: no alt or no ref calls
   • Genotype filtering (set to filtered variants to missing)
     • Filter by Depth (DEFAULT < 10)
     • Filter by GQ (DEFAULT < 25)
     • HET calls: Filter by allele depth/balance (AB/AD; DEFAULT < 0.25)
       • Additional pAB filter on higher depth (binomial test - DEFAULT p < 1e-9)
     • HOMREF calls: (AB/AD; DEFAULT > 0.1)
     • HOMALT calls: (AB/AD; DEFAULT < 0.9)
   • Remove variants not in HWE (DEFAULT p < 1e-6)
   • Generate sample QC metrics from variant-QC'ed VCF

9. Assessing variant QC

   • GOAL: Determine if more stringent variant QC is needed
   • Examine QC parameters across 3 filtering steps:
     • Raw VCF
     • Pre-filtered VCF
     • Variant QC'ed VCF
   • QC parameters:
     • Number of SNPs / Indels
     • TiTv ratio
     • Het/Hom ratio
     • Ins/Del ratio
     • Singleton synonymous rate
   • Primary categories:
     • Case/control (should be equal between groups)
     • Cohort (should vary predictably)
   • Determine whether additional variant filtering needs to be done

10. Final sample QC

    • GOAL: See if any samples are outliers after variant QC
    • Call rate
    • Median Depth
    • TiTv
    • Singleton synonymous rate

QC pipeline examples

Andrea Ganna's steps for running QC and analysis in WGS data:

Analysis Plan

Includes:

• Normalizing phenotypes
• pyhail quality control
• VEP annotation
• Score generation
• Association analysis

WGSPD

Includes:

• Data structure on the cloud
• Downloading files/data from the cloud
• Running Hail on the cloud
• QC filtering
• Generating PCs
• Annotation