Kids First DRC Tumor Only Pipeline

This repository contains tools and workflows for processing of tumor-only samples. The Kids First DRC recommends running the tumor only pipeline ONLY when no matched normal sample is available. If your data has matched normals we recommend running the Kids First DRC Somatic Variant Workflow instead. This workflow is not a traditional production pipeline run on all data, but rather is run at the user's request.

When comparing the SNV outputs of this workflow to those of the somatic workflow, we have found the outputs to be considerably more noisy. To cut down on this noise, we have included some recommended inputs, parameters, and filters for Mutect2 in our docs. In short we recommend:

• Restrict the callable regions with a blacklist and Panel of Normals (PON)
• Remove low support reads:
  • Allele Depth (AD) == 0: WGS uninformative reads
  • Variant Allele Frequency (VAF) < 1%: WXS noise
• Remove potential germline variants: gnomAD AF > 0.00003
• Only keep variants that are PASS
• Rescue any variants that fall in hotspot regions/genes

Benchmarking results of SNV calling used to inform our filtering criteria can be found in this README It can also be used to process PDX data by first pre-processing reads using the Xenome tool, explained more here in documentation.

Import info on cloning the git repo

This repository takes advantage of the git submodule feature. The Single Nucleotide Variant annotation workflow is maintained in our Annotation Tools Repository. Therefore, in order to get the code for a submodule you can either:

• Clone the repository recursively with git clone --recursive
• After cloning, run: git submodule init && git submodule update More info on how this worked here

Main workflow

The wrapper workflow that runs most of the tools is found here.

Tools run

Single Nucleotide Variant (SNV)

• Mutect2 from GATK 4.2.2.0
• Annotation using the Kids First DRC Somatic SNV Annotation Workflow

  Copy Number Variant (CNV)

• ControlFREEC v11.6

  Structural Variant (SV)

• Manta v1.4.0

Inputs

Most inputs have recommended values that should auto import both files and parameters

Recommended file/param defaults:

• indexed_reference_fasta: FAI and DICT indexed Homo_sapiens_assembly38.fasta
• mutect2_af_only_gnomad_vcf: af-only-gnomad.hg38.vcf.gz
• mutect2_exac_common_vcf: small_exac_common_3.hg38.vcf.gz
• gem_mappability_file: hg38_canonical_150.mappability. If you don't have one for your reference and read length, you can first run the GEM indexer tool, then concatenate those results and convert to a mappability file using the GEM mappability tool.
• b_allele: dbSNP_v153_ucsc-compatible.converted.vt.decomp.norm.common_snps.vcf.gz. dbSNP v153 was obtained from the ftp site. Then, using a awk/perl/bash script of your choice, convert NCBI accession names to UCSC-style chromosome names using this table. Next, run the VCF normalization tool, then use bcftools to extract only common snps: bcftools view --include INFO/COMMON=1 --types snps dbSNP_v153_ucsc-compatible.converted.vt.decomp.norm.vcf.gz -O z -o dbSNP_v153_ucsc-compatible.converted.vt.decomp.norm.common_snps.vcf.gz. Lastly, use tabix to index the resultant file.
• vep_cache: homo_sapiens_merged_vep_105_indexed_GRCh38.tar.gz
• genomic_hotspots: tert.bed # bed file with TERT gene promoter region
• protein_snv_hotspots: kfdrc_protein_snv_cancer_hotspots_20240718.txt
• protein_indel_hotspots: protein_indel_cancer_hotspots_v2.ENS105_liftover.tsv
• echtvar_anno_zips: gnomad.v3.1.1.custom.echtvar.zip

  Necessary for user to define:

• input_tumor_aligned: Indexed BAM/CRAM/SAM file
• input_tumor_name: sample name, should match read group sample name in input_tumor_aligned
• panel_of_normals: Mutect2 Panel of Normals
• wgs_or_wxs: Choose whether input is Whole Genome Sequencing (WGS) or Whole Exome Sequencing or Panel (WXS)
• calling_regions:
  • For WGS: wgs_canonical_calling_regions.hg38.bed
  • For WXS: Unpadded experimental bait capture regions
• blacklist_regions:
  • For WGS: hg38-blacklist.v2.bed.gz
  • For WXS: none
• cnv_blacklist_regions:
  • For WGS: somatic-hg38_CNV_and_centromere_blacklist.hg38liftover.bed
  • For WXS: none
• i_flag: for CNV calling, whether to intersect b allele file. Set to N skip
• cfree_sex: for CNV calling, set to XX for female, XY for male
• cfree_ploidy: Array of ploidy possibilities for ControlFREEC to try. Recommend [2,3,4]
• filtermutectcalls_extra_args: "--min-allele-fraction 0.01"
• gatk_filter_name: ["GNOMAD_AF_HIGH", "ALT_DEPTH_LOW"]
• gatk_filter_expression: ["gnomad_3_1_1_AF != '.' && gnomad_3_1_1_AF > 0.001 && gnomad_3_1_1_FILTER == 'PASS'", "vc.getGenotype('<input_tumor_name>').getAD().1 < 1"]
• output_basename: String value to use as basename for outputs

Output Files

Mutect2

• mutect2_protected_outputs: VCF with SNV, MNV, and INDEL variant calls and of pipeline soft FILTER-added values in MAF and VCF format with annotation, VCF index, and MAF format output
• mutect2_public_outputs: Protected outputs, except MAF and VCF have had entries with soft FILTER values removed
• mutect2_bam: BAM generated will be written as BAM. Useful for debugging

  ControlFREEC CNV

• ctrlfreec_pval: Copy number call with GT (if BAF provided) and p values. Most people want this
• ctrlfreec_config: Config file used to run
• ctrlfreec_pngs: Visualization of CN and BAF
• ctrlfreec_bam_ratio: Calls as log2 ratio
• ctrlfreec_bam_seg: Custom made microarray-style SEG file
• ctrlfreec_baf: b allele frequency file
• ctrlfreec_info: Calculated information, like ploidy, if a range was given

  Manta SV

• manta_pass_vcf: VCF file with SV calls that PASS
• manta_prepass_vcf: VCF file with all SV calls
• annotsv_annotated_calls: Manta calls annotated with AnnotSV
• annotsv_unannotated_calls: Manta calls not annotated with AnnotSV