gtars-uniwig scripts for scoring work

This project contains PEPs and pipelines to explore the use of the scoring flag in gtars-uniwig to create bigwig files with and without scoring used for accumulations.

Scored and No-Scored bw files are then fed into geniml's universe creation module. Various universes are then compared against the source files (the one used to create the bw files) to determine recall of genomic regions.

IGV is helpful in visually assessing bw and universes.

Initial results are located on HPC for 3 different sets of files: /project/shefflab/brickyard/results_pipeline/gtars_uniwig

1. ATACseq ~ 500 files
2. lymphoblasotid ~ 400 files
3. scatlas2 samples ~ 358 files

For 1, there are many more files that could be used for this work; I did a random sampling of 500 files to keep output sizes low for proof of concept experimentation.

Example running these pipelines

ATACSEQ 500 27aug2025

export RESULTS=/project/shefflab/brickyard/results_pipeline/gtars_uniwig/experiment_ATACseq_500

export CHROMSIZESPATH=/home/zzz3fh/files_for_gtars_uniwig/hg38.chrom.sizes

export GTARSPATH=/home/zzz3fh/code/gtars/gtars/target/release/gtars

Create bigwig files

looper run -c .looper_uniwig.yaml --package slurm --compute PARTITION=standard time='01-00:00:00' cores='12' mem='64000' -k 2 -d

Create universes using the above bigwig files

looper run -c .looper_parallel_universe.yaml --package slurm --compute PARTITION=standard time='01-00:00:00' cores='12' mem='64000' -k 16 -d

Once these two pipelines are finished, you can take the stats.py script and point it to the stats_output directory to create figures: /project/shefflab/brickyard/results_pipeline/gtars_uniwig/experiment_ATACseq_500/stats_output