patidarr / ngs_pipeline


Introduction

This is the implementation of KhanLab NGS Pipeline using Snakemake.

Installation

The easiest way to get this pipeline is to clone the repository.

git clone https://github.com/patidarr/ngs_pipeline.git

This pipeline is available on NIH biowulf cluster, contact me if you would like to do a test run. The data from this pipeline could directly be ported in OncoGenomics-DB, an application created to visualize NGS data available to NIH users.

Requirements

mutt
gnu parallel
SLURM or PBS for resource management
Bioinformatics Tools Listed in config files

Following R Packages

• OmicCircos
• RColorBrewer
• Rsubread
• edgeR
• limma
• optparse
• pracma
• sequenza
• stats
• stringr
• deconstructSigs

Conventions

• Sample names cannot have "/" or "." in them
• Fastq files end in ".fastq.gz"
• Fastq files are stored in DATA_DIR (Set as Environment Variable)

DNASeq:

• QC
• BWA, Novoalign
• Broad Standard Practices on bwa bam
• Haplotype Caller, Platupys, Bam2MPG, MuTect, Strelka
• snpEff, Annovar, SIFT, pph2, Custom Annotation
• Coverage Plot, Circos Plot, Hotspot Coverage Box Plot
• Create input format for oncogenomics database (Patient Level)
• Make Actionable Classification for Germline and Somatic Mutations
• Copy number based on the simple T/N LogRatio (N cov >=30), Corrected for Total # Reads
• Copy number, tumor purity using sequenza
• LRR adjusted to center
• Contamination using conpair
• HLA Typing
  • HLAminer v1.3.1
  • seq2HLA
• Neoantigen Prediction
  • pVAC-Seq methods: NNalign,NetMHC,NetMHCIIpan,NetMHCcons,NetMHCpan,PickPocket,SMM,SMMPMBEC,SMMalign
    epitope length: 8,9,10,11

RNASeq:

• QC
• Tophat, STAR
• Broad Standard Practices on STAR bam
• fusion-catcher, tophat-fusion, deFuse
• Cufflinks (ENS and UCSC)
• Rsubread TPM (ENS, UCSC), Gene, Transcript and Exon Level
• In-house Exon Expression (ENS and UCSC)
• Haplotype Caller
• snpEff, Annovar, SIFT, pph2, Custom Annotation
• Actionable Fusion classification

Patient:

• Genotyping On Patient. 1000g sites are evaluated for every library and then compared (all vs all) If two libraries come from a patient the match should be pretty good >80%
• Still to develop: If the match is below a certain threshold, break the pipeline for patient.

Rulegraph

DAG for example Sample