The bam-aggregate-task is Kotlin workflow task which aggregates the 5' end of reads from an alignment file across sets of input regions in BED 6+ format, producing corresponding vectors of read counts in JSON format. All the regions in each input file are first resized to a constant width, which defaults to 2000 basepairs, around their center points. The task requires the input BAM to have a corresponding index, located at <input>.bam.bai. The task is strand-aware: when a region in a BED 6+ input file lies on the minus strand, the ordering of read counts around its center is reversed. If your input regions are not strand-specific, you can use a strand designation of . in the sixth field of each BED file.

To run the task, first checkout the GitHub repository and build an executable JAR:

git clone https://www.github.com/krews-community/bam-aggregate-task
cd bam-aggregate-task && scripts/build.sh

Then, to run the task:

java -jar build/*.jar \
    --region-file input1.bed
    --alignments input.bam
    --expansionSize 4000
    --outputDir output

This will produce an output file at output/input1_input.aggregate.json containing a vector of aggregated read counts in JSON format. The above example uses a custom width of +/- 2000 basepairs around each peak summit, rather than the default +/- 1000.

The scripts/ directory contains utilities you can use to build, test, and deploy.

To build the docker container with a version tag, use scripts/build-image.sh vX.X.X. To build the JAR locally, use scripts/build-local.sh. To run the built JAR, use scripts/run-local.sh.

To run automated tests from command line, use scripts/test.sh.

To deploy the image to our docker repository, use scripts/push-image.sh.