PharmGKB NGS Pipeline

This repository contains a simplified version of our work-in-progress GRCh38 NGS pipeline for pharmacogenomics. The primary goal of this pipeline is to generate VCF data suitable for use by PharmCAT.

Quick Overview

The pipeline takes FASTQ files, aligns them using BWA, and then calls variants using GATK.

It is optimized for WGS and Exome analysis, and currently does not handle trios. VQSR filtering has shown to be effective for WGS data, but you may want to switch to hard filtering if using other data sources.

Take a look at run.sh to see the steps of our pipeline and the parameters used. The input is a pair of FASTQ files from an Illumina NGS run and the output is a complete VCF file containing genotypes for every position, which is useful for pharmacogenomics analysis and PharmCAT.

Setting Up

This pipeline uses Bwakit, GATK and Picard. Follow the GATK installation instructions to install bwa (0.7.13), picard (2.0.1), samtools and gatk (3.5.0). Please take particular note to use Java 8, and to install the R libraries, as these are needed. Update the paths in the run.sh if necessary.

We recommend placing using the following file structure:

pipeline:
|-- run.sh
|-- ext_data:
|   |-- hgbundle
|   '-- genome
'-- utils
    |-- bwa-0.7.12
    |-- picard-2.0.1
    '-- gatk-3.5-0

For /ext_data:

uncompress the GATK hg38 bundle from ftp://gsapubftp-anonymous@ftp.broadinstitute.org/bundle/hg38/hg38bundle.tar.gz

For /ext_data/genome:

uncompress the GRCh38 (no-alt) FASTA data from ftp://ftp.ncbi.nlm.nih.gov/genomes/all/GCA_000001405.15_GRCh38/seqs_for_alignment_pipelines.ucsc_ids/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.gz
download GRCh38 (no alt) FASTA index from ftp://ftp.ncbi.nlm.nih.gov/genomes/all/GCA_000001405.15_GRCh38/seqs_for_alignment_pipelines.ucsc_ids/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.fai

You will need to also need to index the genome for BWA. The following commands should automate the download and indexing:

> curl -o external_data/genome/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.fai ftp://ftp.ncbi.nlm.nih.gov/genomes/all/GCA/000/001/405/GCA_000001405.15_GRCh38/seqs_for_alignment_pipelines.ucsc_ids/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.fai
> curl -o external_data/genome/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.gz ftp://ftp.ncbi.nlm.nih.gov/genomes/all/GCA/000/001/405/GCA_000001405.15_GRCh38/seqs_for_alignment_pipelines.ucsc_ids/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.gz
> bwa index external_data/genome/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.gz
> curl -o external_data/grc38.tar.gz ftp://gsapubftp-anonymous@ftp.broadinstitute.org/bundle/hg38/hg38bundle.tar.gz
> tar -zxvf external_data/grc38.tar.gz -C external_data/

Testing the Pipeline

The following commands can be used to download some test files to check if the pipeline is working:

> mkdir test
> curl -o test/fasta1.gz ftp://ftp-trace.ncbi.nlm.nih.gov/giab/ftp/data/NA12878/Garvan_NA12878_HG001_HiSeq_Exome/NIST7035_TAAGGCGA_L001_R1_001.fastq.gz
> curl -o test/fasta2.gz ftp://ftp-trace.ncbi.nlm.nih.gov/giab/ftp/data/NA12878/Garvan_NA12878_HG001_HiSeq_Exome/NIST7035_TAAGGCGA_L001_R2_001.fastq.gz
> ./run.sh test/fasta1.gz test/fasta2.gz "@RG\tID:fasta1.gz\tPL:ILLUMINA\tSM:1" test run1

Running

The pipeline can be run using the following command:

> ./run.sh fasta1.q fasta2.q @RG\tID:fasta1.gz\tPL:ILLUMINA\tSM:1 working_dir output_prefix

Where:

fasta1.q and fasta2.q are pair reads from Illumina sequencing,
RG is the read group header in the format @RG\tID:fasta1.gz\tPL:ILLUMINA\tSM:1,
working_dir is the directory where the input FASTQ files are and where the output will be put,
and output-prefix will be added to beginning of each of the output files.

This script has been tested on Ubuntu, takes about 5 days for 30x WGS data, and requires roughly 200GB of space.

Docker version

The pipeline can also be installed using this docker file. To run this version check out the project and cd into the main directory then create the docker image:

> docker build -t pharmngs .

Then create a container using this image, with a shared volume for the data:

> docker run -t -v <folder_with_ngs_data>:/usr/share/docker_data pharmngs /bin/bash

This gives you command line access to the container. If you would like to try running the test data type: