pbRUGD-workflow

Workflow for the comprehensive detection and prioritization of variants in human genomes with PacBio HiFi reads

Note: Workflow is committed. Web app code to come.

Authors

• William Rowell (@williamrowell)
• Aaron Wenger (@amwenger)

Description

This repo consists of three Snakemake workflows:

1. process_smrtcells
2. process_samples
3. process_cohorts

process_smrtcells

• find new HiFi BAMs or FASTQs under smrtcells/ready/*/
• align HiFi reads to reference (GRCh38 by default) with pbmm2
• calculate aligned coverage depth with mosdepth
• calculate depth ratios (chrX:chrY, chrX:chr2) from mosdepth summary to check for sample swaps
• calculate depth ratio (chrM:chr2) from mosdepth summary to check for consistency between runs
• count kmers in HiFi reads using jellyfish, dump and export modimers

process_sample

• launch once sample has been sequenced to sufficient depth
• discover and call structural variants with pbsv
• call small variants with DeepVariant
• phase small variants with WhatsHap
• merge per SMRT Cell BAMs and tag merged bam with haplotype based on WhatsHap phased DeepVariant variant calls
• merge jellyfish kmer counts
• assemble reads with hifiasm and calculate stats with calN50.js
• align assembly to reference with minimap2
• check for sample swaps by calculate consistency of kmers between sequencing runs

process_cohort

• launched once all samples in cohort have been processed
• if multi-sample cohort
  • jointly call structural variants with pbsv
  • jointly call small variants with GLnexus
• using slivar
  • annotate variant calls with population frequency from gnomAD and HPRC variant databases
  • filter variant calls according to population frequency and inheritance patterns
  • detect possible compound heterozygotes, and filter to remove cis-combinations
  • assign a phenotype rank (Phrank) score, based on Jagadeesh KA, et al. 2019. Genet Med.

Dependencies

• some tools (e.g. pbsv) require linux
• conda
• singularity >= 3.5.3 installed by root
• environment.yaml

Configuration

• config.yaml contains file paths and version numbers for docker images
• reference.yaml contains file paths and names related to reference
• *.cluster.yaml contains example cluster configuration for a slurm cluster with a compute queue for general compute and a ml queue for GPU.