d3b-center/tumor-loh-app-dev

Development repo for LOH

Kids First Loss of Heterozygosity (LOH)

The Kids First Loss of Heterozygosity Preprocessing (aka LOH) is a CWL workflow that assesses the loss of heterozygosity in the tumor for rare germline calls filtered by gnomad_3_1_1_AF_popmax (typically < 0.01) or when gnomad_3_1_1_AF_popmax is not defined. This preprocessing is designed to compute variant allele frequency (VAF) for multiple proband tumor samples and can also map germline VAF for family trios if trio germline VCF file is provided.

Basic info

• Dockerfile: https://github.com/d3b-center/bixtools/tree/master/LOH/1.0.1
• tested with
  • Seven Bridges Cavatica Platform: https://cavatica.sbgenomics.com/
  • cwltool: https://github.com/common-workflow-language/cwltool/releases/tag/3.1.20221201130942

Application Description

The Kids First Loss of Heterozygosity application is divided into two tools: Germline tool and Tumor tool.

Germline Tool

Germline tool filters germline annotations to retain variants based on gnomad_3_1_1_AF_popmax (typically < 0.01) or when gnomad_3_1_1_AF_popmax is not defined. It requires vcf file, proband sample id, ram as required inputs and peddy file as optional input which is required for family trios. It outputs variant information such as gene, chr, start, stop, ref/alt alleles, ref/alt allele depths, variant allele frequency and list of coordinates that will be an input to tumor tool.

Tumor Tool

Search in paired proband tumor sample for aligned reads in the regions where rare variants from the germline exist and exact allele/reference count, allele/reference depth, and calculate variant allele frequency(VAF). Tumor tool has the capability to search in multiple tumor samples for proband and if applicable, paternal and maternal tumor samples. To extract reads from the bam/cram files, this tool utilizes bam-readcount and wraps it with a python script to shape the output in a tabular format.

LOH Inputs

Germline tool
  # Required  
  BS_ID:{ doc: provide BS id for germline normal, type: string }
  frequency:{ doc: provide popmax cutoff for rare germline variants, type: 'float?', default: 0.01 }
  # Optional
  ram_germline:{  doc: Provide ram (in GB) based on the size of vcf, type: 'int?', default: 8}
  # Required for family trios otherwise not required
  peddy_file:{ doc: provide ped file for the trio, type: 'File?' }

Tumor tool
  # Required
  participant_id:{ doc: provide participant id for this run, type: string }
  bamscrams:{ doc: tumor input file in cram or bam format with their index file, type: 'File[]' , secondaryFiles: [ { pattern: ".crai", required: false }, { pattern: ".bai", required: false } ] }
  reference:{ doc: human reference in fasta format with index file, type: File,secondaryFiles: [ .fai ] }
  sample_vcf_file: { doc: provide germline vcf file for this sample, type: File }
  # Optional
  minDepth:{ doc: provide minDepth to consider for tumor reads, type: 'int?', default: 1 }
  bamcramsampleIDs: { doc: provide unique identifers (in the same order) for cram/bam files provided under bamcrams tag. Default is sample ID pulled from bam/cram files., type: 'string[]?' }
  ram_tumor:{  doc: Provide ram (in GB) for tumor tool based on the number cram/bam inputs, type: 'int?', default: 16} 
  minCore:{ type: 'int?', default: 16, doc: "Minimum number of cores for tumor tool based on the number cram/bam inputs" }

LOH schematic

LOH Output

LOH application will output a tab-separated values file mapped data from germline tool and tumor tool.

output_file:{ type: File, doc: A tsv file with gathered data from germline and tumor tool }

Output headers

Preprocessing LOH will generate a tab-separated values file with following headers:

Headers	Description
BS_ID	Sample Id for germline sample
gene	Gene
chr	Chromosome
start	Start position
end	End position
ref	Reference allele
alt	Alternate Allele
proband_germline_ref_depth	Reference depth from germline from germline proband
proband_germline_alt_depth	Alternate depth from germline from germline proband
proband_germline_depth	Total number of reads overlapping a site from germline proband
proband_germline_vaf	Fraction of reads with the alternate allele from germline proband
paternal_germline_ref_depth	Reference depth from germline from paternal germline
paternal_germline_alt_depth	Alternate depth from germline from paternal germline
paternal_germline_depth	Total number of reads overlapping a site from paternal germline
paternal_germline_vaf	Fraction of reads with the alternate allele from paternal germline
maternal_germline_ref_depth	Reference depth from maternal germline
maternal_germline_alt_depth	Alternate depth from maternal germline
maternal_germline_depth	Total number of reads overlapping a site from mother
maternal_germline_vaf	Fraction of reads with the alternate allele from mother
proband_sample_id_tumor_vaf	Proband variant allele frequency from specific tumor sample
proband_sample_id_tumor_depth	Depth of coverage from specific tumor sample
proband_sample_id_tumor_alt_depth	Allele count at site from specific proband tumor sample
proband_sample_id_tumor_ref_depth	Reference count at site from specific proband tumor sample

More information can be found here

Running it locally on a laptop?

It is recommended to run this CWL workflow on a system with a high number of CPUs and memory (>=16 GB). The basic requirement is a running docker engine and CWL tools. Command line to run the LOH workflow locally is:

cwltool workflow/run_LOH_app.cwl sample_input.yml

Note: Inputs to the workflow need to be defined in sample_input.yml.

tumor-loh-app-dev's Issues

Update README to list the output file headers with explanations

Please add a section of the README which describes the output file with headers and what each means.
Thanks!