Skip to contents

Installation

Source: vignettes/installation.Rmd
installation.Rmd

Data

PCGR requires the following data to run successfully:

  1. Reference bundle - containing data from multiple knowledge resources, including information on molecular biomarkers, targeted cancer therapies, variant frequencies etc. Key datasets include CIViC, CGI, Open Targets Platform, TCGA, and ClinVar.
  2. Ensembl VEP data cache - needed for variant annotation with VEP (Variant Effect Predictor)
  3. User-supplied sample-specific inputs - e.g. somatic variant calls in VCF format

PCGR supports both the GRCh37 and GRCh38 human genome assemblies. All the data above need to match the chosen assembly.

1. Reference Bundle

Reference bundles are generated semi-automatically (by the PCGR author) and are versioned based on their release date. Keep in mind that the bundles support only certain Ensembl VEP versions, and importantly also specific software versions of PCGR. Upgrading the PCGR software without upgrading the bundle specified here is thus not a recommended installation strategy. The latest (v20250314) genome-specific bundles can be downloaded directly from below (size: ~5G):

Assembly Download Link
GRCh38 https://insilico.hpc.uio.no/pcgr/pcgr_ref_data.20250314.grch38.tgz
GRCh37 https://insilico.hpc.uio.no/pcgr/pcgr_ref_data.20250314.grch37.tgz

Tip 1: The data/grch3x/.PCGR_BUNDLE_VERSION file within the downloaded bundle indicates the bundle version for reporting purposes.

Tip 2: The data/grch3x/data_overview.grch3x.html file provides a report with an overview and statistics of the key resources included in the reference bundle.

Bash Example 

BUNDLE_VERSION="20250314"
GENOME="grch38" # or "grch37"
BUNDLE="pcgr_ref_data.${BUNDLE_VERSION}.${GENOME}.tgz"
wget https://insilico.hpc.uio.no/pcgr/${BUNDLE}
gzip -dc ${BUNDLE} | tar xvf -

mkdir ${BUNDLE_VERSION}
mv data/ ${BUNDLE_VERSION}

2. VEP Cache

VEP requires a data cache which is available from the Ensembl FTP site (search there for files starting with homo_sapiens_vep_). The latest Ensembl VEP version we support is v113.

Tip: PCGR needs to be pointed to the parent directory containing the downloaded homo_sapiens/xyz_GRCh3x/ cache, which is usually called .vep if you’ve followed the VEP cache download instructions.

Bash Example 

VEP_VERSION="113"
GENOME="GRCh38" # or "GRCh37"
CACHE="homo_sapiens_vep_${VEP_VERSION}_${GENOME}.tar.gz"

wget https://ftp.ensembl.org/pub/release-${VEP_VERSION}/variation/indexed_vep_cache/${CACHE}
gzip -dc ${CACHE} | tar xvf -

3. Sample Inputs

See the Inputs article.

Software

The PCGR workflow can be installed with any of the following:

A. Conda

There is Conda support for both Linux and macOS machines. The following process can take anywhere from 10 up to 40 minutes when installing from scratch, mostly depending on the user’s and server’s internet connection. Most of the time is spent on downloading the {BSgenome.Hsapiens.UCSC.hg19} and {BSgenome.Hsapiens.UCSC.hg38} R packages (which happens at the very end of the conda environment creation).

PCGR_VERSION="2.2.1"
# set up variables
PCGR_REPO="https://raw.githubusercontent.com/sigven/pcgr/v${PCGR_VERSION}/conda/env/lock/"
PLATFORM="linux"
# create conda envs in local directory
mkdir pcgr_conda
conda create --prefix ./pcgr_conda/pcgr --file ${PCGR_REPO}/pcgr-${PLATFORM}-64.lock
conda create --prefix ./pcgr_conda/pcgrr --file ${PCGR_REPO}/pcgrr-${PLATFORM}-64.lock
# you need to specify the directory of the conda env when using --prefix
conda activate ./pcgr_conda/pcgr
# test that it works
pcgr --version
pcgr --help

For macOS M1 machines, you need to include CONDA_SUBDIR=osx-64 before the conda create command - see https://github.com/conda-forge/miniforge/issues/165#issuecomment-860233092:

PCGR_VERSION="2.2.1"
# set up variables
PCGR_REPO="https://raw.githubusercontent.com/sigven/pcgr/v${PCGR_VERSION}/conda/env/lock/"
PLATFORM="osx"
# create conda envs in local directory
mkdir pcgr_conda
CONDA_SUBDIR=osx-64 conda create --prefix ./pcgr_conda/pcgr --file ${PCGR_REPO}/pcgr-${PLATFORM}-64.lock
CONDA_SUBDIR=osx-64 conda create --prefix ./pcgr_conda/pcgrr --file ${PCGR_REPO}/pcgrr-${PLATFORM}-64.lock
# you need to specify the directory of the conda env when using --prefix
conda activate ./pcgr_conda/pcgr
# test that it works
pcgr --version
pcgr --help

B. Docker

The PCGR Docker image is available on Docker Hub. Pull the latest v2.2.1 image with:

docker pull sigven/pcgr:2.2.1

Example Run 

PCGR_VERSION="2.2.1"
docker container run -it --rm \
    -v /Users/you/dir1/.vep:/mnt/.vep
    -v /Users/you/dir1/bundle:/mnt/bundle \
    -v /Users/you/dir1/pcgr_inputs:/mnt/pcgr_inputs \
    -v /Users/you/dir1/pcgr_outputs:/mnt/pcgr_outputs \
    sigven/pcgr:${PCGR_VERSION} \
    pcgr \
      --input_vcf "/mnt/pcgr_inputs/T001-BRCA.grch38.vcf.gz" \
      --vep_dir "/mnt/.vep" \
      --refdata_dir "/mnt/bundle" \
      --output_dir "/mnt/pcgr_outputs" \
      --genome_assembly "grch38" \
      --sample_id "SAMPLE_B" \
      --tumor_dp_tag "TDP" \
      --tumor_af_tag "TAF" \
      --assay "WGS" \
      --vcf2maf

NOTE: If you need to run the Docker-based version of PCGR as a non-root user, you may need to explicitly add options for quarto to work properly, i.e. --env "XDG_CACHE_HOME=/tmp/quarto_cache_home" (same as for Singularity/Apptainer below, see also issue #246).

C. Singularity/Apptainer

The PCGR Singularity/Apptainer image is available on GitHub Container Registry. Pull the latest v2.2.1 image with:

apptainer pull oras://ghcr.io/sigven/pcgr:2.2.1.singularity

This will download a Singularity Image File (SIF) called pcgr_2.2.1.singularity.sif that can be run with Singularity or Apptainer.

Example Run 

PCGR_VERSION="2.2.1"
apptainer exec \
  --writable-tmpfs \
  --no-home \
  --env "XDG_CACHE_HOME=/tmp/quarto_cache_home" \
  -B /Users/you/dir1/.vep:/mnt/.vep \
  -B /Users/you/dir1/bundle:/mnt/bundle \
  -B /Users/you/dir1/pcgr_inputs:/mnt/pcgr_inputs \
  -B /Users/you/dir1/pcgr_outputs:/mnt/pcgr_outputs \
  pcgr_${PCGR_VERSION}.singularity.sif \
  pcgr \
    --input_vcf "/mnt/pcgr_inputs/T001-BRCA.grch38.vcf.gz" \
    --vep_dir "/mnt/.vep" \
    --refdata_dir "/mnt/bundle" \
    --output_dir "/mnt/pcgr_outputs" \
    --genome_assembly "grch38" \
    --sample_id "SAMPLE_B" \
    --assay "WGS" \
    --tumor_dp_tag "TDP" \
    --tumor_af_tag "TAF" \
    --vcf2maf

Note: For any Apptainer/Singularity issues not directly related to PCGR, we would recommend reaching out to the Apptainer community (e.g. https://github.com/apptainer/apptainer) since we have limited experience with Apptainer/Singularity.