STAR-Fusion for Treehouse RNA-seq analysis

Overview

Gene fusions play a major role in tumorigenesis, so it is crucial that Treehouse has a pipeline for detecting them. We have built a docker container that runs STAR-Fusion and filters the output against a list of known cancer fusion genes. There is also an option to run additional filters and generate de novo assembled fusion transcripts using the FusionInspector program.

Docker and usage

Image located on hub.docker.com

REPOSITORY: ucsctreehouse/fusion

TAG: 0.3.0

Input files

The pipeline requires paired-end fastq files, the output directory, and the genome library directory. The genelist is already baked into the docker container, but there is an option to include a different genelist. Please refer to the STAR-Fusion documentation for creating a genome library.

Wraps STAR-Fusion program and filters output using FusionInspector.

optional arguments:
  -h, --help            show this help message and exit
  --left-fq R1          Fastq 1
  --right-fq R2         Fastq 2
  --output-dir OUTPUT_DIR
                        Output directory
  --tar-gz              Compresses output directory to tar.gz file
  --genome-lib-dir GENOME_LIB_DIR
                        Reference genome directory (can be tarfile)
  --CPU CPU             Number of jobs to run in parallel
  --genelist GENELIST
  --skip-filter         Skips gene-list filter
  -F, --run-fusion-inspector
                        Runs FusionInspector on STAR-Fusion output
  --star-fusion-results STAR_FUSION_RESULTS
                        Skips STAR-Fusion and runs FusionInspector
  --save-intermediates  Does not delete intermediate files
  --root-ownership      Does not change file ownership to user
  --test                Runs the pipeline with dummy files
  --debug               Prints tool command line arguments

Run command

docker run -it --rm -v `pwd`:/data ucsctreehouse/fusion:0.3.0 \
                                   --left-fq 1.fq.gz \
                                   --right-fq 2.fq.gz \
                                   --output-dir fusion_output \
                                   --genome-lib-dir STARFusion-GRCh38gencode23 \
                                   --run-fusion-inspector

Output

STAR-Fusion Output

STAR-Fusion candidates and fusion fragments per million mapped read values
- star-fusion-non-filtered.final
Fusions where donor and acceptor genes passed the gene-list filter
- star-fusion-gene-list-filtered.final

FusionInspector Output

FusionInspector candidates and fusion fragments per million mapped read values where donor and acceptor passed gene-list filter
- fusion-inspector-results.final
IGV input files for visualization
- FusionInspector.fa
- FusionInspector.gtf
- FusionInspector.junction_reads.bam
- FusionInspector.spanning_reads.bam

Fusion predictions with a large anchor support (YES_LDAS) and total FFPM > 0.1 are considered significant.

Output conditions

Different output files are generated depending on the amount of evidence supporting the fusion calls. If star-fusion-gene-list-filtered.final contains no fusions, FusionInspector does not run and no fusion-inspector-results.final output file is generated. If star-fusion-gene-list-filtered.final contains fusions, FusionInspector runs. If the fusions meet FusionInspector's criteria, an output file is generated containing fusions. If they do not, an output file is generated containing column headings but no fusion data.

Version disambiguation

This repo ("fusion-for-core") contains a change in star_fusion_pipeline.py that renames FusionInspector.fusion_predictions.final.abridged.FFPM to fusion-inspector-results.final, which is consistent with the output generated by the docker we use (ucsctreehouse/fusion@sha256:3faac562666363fa4a80303943a8f5c14854a5f458676e1248a956c13fb534fd) and not with the https://github.com/UCSC-Treehouse/treehouse-fusion repository.

UCSC-Treehouse / fusion-for-core