VarRNA is a comprehensive pipeline designed to process RNA-Seq reads from tumor samples, starting with either FASTQ or BAM files. It identifies single nucleotide variants (SNVs) and indels, and classifies these variants as germline, somatic, or artifact. Leveraging Snakemake, VarRNA efficiently tracks each step and processes multiple samples in parallel. The pipeline requires minimal configuration of input sample paths and manages most dependencies through a micromamba (conda) environment.The models are set up to evaluate RNA-Seq data aligned to human reference version GRCh38.
The VarRNA package can run on a standard computer, provided it has sufficient RAM for in-memory operations. However, for optimal performance and efficiency, especially when processing multiple samples simultaneously, we recommend using a high-performance computing environment. This setup will allow you to fully leverage the parallel processing capabilities.
Suggested RAM and CPU specs will be based on the users input file size. The runtimes below are generated using a computer with 28 GB RAM and 4 cores.
git clone https://github.com/nch-igm/VarRNA.git
cd VarRNA
A conda environment is used to manage almost all packages. Please refer to the following documentation to install micromamba (or conda), if not already installed: https://mamba.readthedocs.io/en/latest/installation/micromamba-installation.html.
Create the micromamba environment and activate it.
micromamba env create -f dependencies/mamba_environment.yml
micromamba activate varrna
ANNOVAR is used to add annotations to our VCFs using several databases. You will need to install this yourself since it requires a user agreement license. Please download annovar from their download page, and use the link provided from registering to download to the following directory:
wget <link/to/annovar/tar/file> -P dependencies/
tar -xvzf dependencies/annovar.latest.tar.gz -C dependencies/All the databases that annovar uses will be downloaded in the get_resourches.sh script, later.
Salmon read quantification is used to get TPM values. Below are the instructions to install salmon, create the index, and then point to it's installation location in the config file.
wget https://github.com/COMBINE-lab/salmon/releases/download/v1.9.0/salmon-1.9.0_linux_x86_64.tar.gz -P dependencies/
tar -xvzf dependencies/salmon-1.9.0_linux_x86_64.tar.gz -C dependencies/Note, config/config.yaml references this installation to run salmon. If you download a different version or copy it to a different location, you will need to update that line in the config/config.yaml file (under dependencies). It is not recommended to download a different version.
A number of resources are needed for variant calling and annotation. Download these by running:
bash get_resources.shRNA-Seq samples can be FASTQ files or BAM files, however starting with FASTQ is suggested.
Input FASTQ files are expected to be processed (e.g., run through fastqc, reads are trimmed).
Please use the suggested Alignment scripts first before running the pipeline to get your BAM files.
cd Alignment
# Download STAR
bash get_star.sh
# Build the genome. The sjdbOverhang is dependent on sequencing library and should be your <max(read length) - 1>.
# You may need to edit this value in line 14 of the following script.
bash star_genome_build.sh
# Alignment using the STAR 2-pass method.
# Fastq files are in the format <sample>.R1.fastq.gz. You may need to update the script in lines 6 and 7 to match your fastq file names.
bash star_alignment.sh <sample>It is recommended that the RNA-Seq FASTQ files are used as input so that the reference used for alignment is consistent with variant calling and annotating. However, if you wish to use existing BAM files, please make sure you update the config file (config/config.yaml) with the same reference you used for alignment in the fasta value (line 13). If this step is not done, VarRNA will run into errors during annotation if contigs do not match the expected reference.
You will need to modify the config/samples.csv file to add the sample names and paths to your input RNA BAM files.
Most files generated are temporary. Saved results include annotated VCFs, and features CSV with variant labels predicted as one of Germline, Somatic, or Artifact:
results/<sample>/VCFs/annotated/<sample>.annotated.vcf.gz
results/<sample>/Predictions/<sample>.annotated_predictions.csv
It is recommended to run VarRNA using high performance computing. The following example is using a Sun Grid Engine (SGE) batch system to allocate multiple jobs running at the same time. The scheduler.sh script can be altered to work with other batch system environments.
Making the qsub_logfiles directory will allow snakemake to create a separate logfile for each rule. This can be helpful for understanding what commands are being run, especially when your workflow becomes complex. If this step is ommited it will write all logs to one file.
The --jobs 64 parameter provides snakemake with the maximum number of jobs it can run in parallel. You should update this value to meet your needs or limitations. Please see the snakemake cli documentation for a description of all parameters.
cd workflow
mkdir qsub_logfiles
qsub scheduler.sh
The snakemake pipeline can be started by running the following commands. You can change the --cores value to meet your needs or limitations.
cd workflow
snakemake --cores 2