Haplotype Aware Segmentation algorithm for estimating homologue-specific somatic copy number alterations
HapASeg requires a number of dependencies that can be complicated to install. We provide a docker image that builds a container for running HapASeg in hapaseg_local/Dockerfile. To build the container from the root directory, invoke docker build -f ./hapaseg_local/Dockerfile -t hapaseg_image .
HapASeg also relies on a number of reference files. These files can be downloaded automatically by calling docker run -v {./workdir}:{/workdir/} hapaseg_image hapaseg_local_install_ref_files --ref-build {ref-build} /workdir/ref_files/ where ref_build is the reference genome build of the input sample BAMS (hg19 or hg38 or both) and workdir is the local path where the reference files will be saved.
Once the reference files have been downloaded, the HapASeg can be executed by calling docker run -v {workdir}:{/workdir/} hapaseg_image hapaseg_local ... with the desired inputs. Use --help to see the run options. Here, workdir is the local path where the reference files were saved. Directories containing sample BAM files can be mounted with additional -v mount commands if they are not already in workdir.
hapaseg_local has several subroutines that are ammenable to parallelization. Set the maximum number of cpus and memory that you would like HapASeg to using the --max-cpus and --max-mem commands. The default behaviour is to use all available resources. The method requires at least 12GB of memory to run.
Details on the method and relevant benchmarking can be found in our preprint XXX with citation XXX.
HapASeg has been optimized for running on the wolF workflow managment platform. The tasks and full workflow can be found in /wolF. Benchmarking for the aforementioned publication was also done in wolF with source code available in /benchmarking.