nf-core pipeline · archived · deep-variant, dna, google
nf-core/deepvariant
A Nextflow pipeline running Google DeepVariant to identify genetic variants from BAM alignment files.
What nf-core/deepvariant is
nf-core/deepvariant is a Nextflow bioinformatics pipeline designed to execute Google DeepVariant, a variant caller released by the Google Brain Team in December 2017. DeepVariant detects genetic differences by treating alignment data as images and applying deep learning image recognition models. The pipeline was originally developed at Lifebit to streamline and reduce the cost of variant calling analyses, with contributions from the nf-core community.
The workflow automates prerequisite preprocessing steps, including the indexing and compression of required input files, before running variant calling across single or multiple BAM files in parallel. It runs each step inside Docker containers to ensure computational reproducibility and outputs variant predictions in standard Variant Call Format (VCF).
The pipeline is now archived and no longer maintained. Researchers are advised to consider using or contributing to the nf-core/sarek pipeline instead.
What the pipeline does
The workflow accepts a reference genome alongside one or more BAM alignment files, with optional target BED files. During initial preprocessing steps, the pipeline checks for required compressed and indexed files; if these are missing, it creates them automatically. If they are already provided, the preprocessing phase is skipped.
For the variant calling phase, DeepVariant constructs images representing aligned reads from each BAM file. A deep learning image recognition model processes these images to detect variants, and the resulting predictions are converted into individual VCF files. Each input BAM file is processed independently, allowing Nextflow to parallelise execution across available compute cores.
Key facts
- Pipeline engine
- Nextflow
- Status
- Archived (no longer maintained)
- Latest release
- 1.0 (2018-11-19)
- Inputs
- BAM files, FASTA reference genome, optional BED file
- Outputs
- VCF files
- Container support
- Docker
- Original developer
- Lifebit
What you can use it for
- Whole genome variant calling
- Calling genetic variants across whole genome sequencing BAM datasets using reference genomes such as hg19.
- Whole exome variant calling
- Identifying variants from targeted whole exome sequencing data using BAM files and specified target BED regions.
- Multi-sample batch processing
- Running variant calling simultaneously across multiple independent BAM files in parallel to generate separate VCF files.
Strengths and limits
Good at
- Automatically prepares and indexes secondary input files if they are not supplied by the user.
- Parallelises execution across multiple BAM files independently to maximise compute resource utilisation.
- Executes pipeline steps within Docker containers to ensure clean, reproducible outputs.
Watch out for
- The repository is archived and no longer maintained, with users directed to nf-core/sarek instead.
- Running DeepVariant can be very computationally intensive.
Questions
Can I run nf-core/deepvariant without my own cluster?
The research agent runs nf-core/deepvariant with Nextflow on its own machine — up to 32 cores, 256 GiB of memory and 24 hours per run, which covers the test profile and real single-sample runs. It builds the samplesheet, checks it against the pipeline's schema and reads the results back. When a run is too big for one machine, it gives you the exact command for your own cluster instead.
What licence is nf-core/deepvariant under?
MIT, like every nf-core pipeline.
Is nf-core/deepvariant actively maintained?
No, the repository is archived and no longer maintained. Users are encouraged to use or contribute to nf-core/sarek.
How does DeepVariant identify variants?
It builds images from BAM alignment files, analyses those images using a deep learning image recognition approach to predict variants, and outputs the calls in standard VCF format.
Can the pipeline process whole exome data?
Yes, whole exome data is supported by providing an exome flag along with a BED file defining the target regions.
What happens if required index files are not provided?
The pipeline includes preprocessing steps that automatically generate missing indexed and compressed files before running DeepVariant.
Written from nf-core/deepvariant's own README. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.