DNV phasing with long read technology

• Authors: Allison Seiden, Felix Richter

Project overview

1. Use 5x PacBio whole genome sequencing (WGS) to phase de novo variants (DNVs) previously identified with 30x Illumina WGS.
   • Python 3 wrappers around WhatsHap to phase variants using PacBio (no indels) and/or Illumina WGS
   • Python 3 programs to assign DNVs to parent-of-origin based on informative variants (i.e., those uniquely inherited from mom or dad) on the same reads
2. Analyze properties of these DNVs, overall and between the following categories
   • Phased vs unphased
   • Illumina vs Illumina+PacBio phasing
   • Maternal vs paternal
   • Within clusters
   • As a function of parental age

Dependencies

• Python 3
• WhatsHap
• tabix and samtools

Directory descriptions

• cluster_analysis: code for analyzing DNV clusters
• data: pedigree files for all trios
• IGV: salient IGV plots
• notes: lab notebook of day-to-day commands
• phasing: core programs to run WhatsHap and assign variants to parent-of-origin or classify as unphased
• phasing_analysis: scripts and programs for downstream DNV analyses

Core phasing pipeline

• chromosome_split.py: splits a VCF by chromosome in (so that Whatshap is run separately by chromosome)
• whasthap_bsub.sh (or illumina_whatshap_int1.py): Run whatshap with indels per chromosome
  • pacbio_whatshap.py: run phasing analysis for PacBio data
• whasthap_output_check.py: Check whatshap output file length and delete if not consistent with input (i.e., re-run)
• clean_whatshap_vcf.py: Remove rows without variants and move VCF to new, smaller directory
• get_gtf.py: Run the Whasthap GTF command to get phased haplotype blocks
• get_ID_dataframes.py: wrapper around the PhasedData.py class, creates a PhasedData object for every ID/chromosome
• Other files in the phasing/ directory:
  • check_discontinuities.py: tests the discontinuities functions that are part of PhasedData.py
  • get_results.py: review the phasing results
  • sort_and_index.py: preprocess the PacBio BAM (which had weird ASCII characters that had to be removed)
  • split_trios.py: split the large single VCF by trio
  • utilsy.py: various utility functions used across scripts

Analysis file descriptions

• PacBio (N=10) data results
  1. phasing_analysis/results_phasing/indels_df.txt: indels phased with heuristic/IGV approach
  2. indel_analysis/classified_indels.txt: sorting-hat output (contains indel classes and repeat track overlaps)
  3. indel_analysis/all_indel_info.txt: joins of (1) and (2)
• Illumina (N=308) data results
  1. phasing_analysis/results_phasing/indels_df_ilmn_2018_12_08.txt: contains both indel phasing and sorting hat output classifications
• counts_per_id_ilmn_pb_2019_01_22.txt: Joined counts per ID from Illumina and PacBio