Script to build new reference files

[skchronicles]

It would be awesome if we had a script to automate the process of building reference files for different/new organisms, something where we could give it a minimal set of inputs (such as a genomic fasta file, annotation, and an optional blacklist, etc.) and it would handle the rest.

[tovahmarkowitz]

This task is now possible to achieve at least to some extent. Example code for all pieces available at: /data/OpenOmics/references/chrom-seek/mm39_basic/.

Requires custom installation of khmer, which is not available on Biowulf.

All needed files are created by chrom-seek_prep.sbatch and create_geneinfo_bed.R.

• create_geneinfo_bed.R needs to be improved/rewritten as the input and output file names are currently hardcoded.

Let's set it up so that all files get 1-2 reference-specific flags. All files should get a reference-specific id (eg, hg38). All files created from the original gtf should have an annotation version id (ie, M39). These will likely need to be manually defined.

calc_frac.py and run_khmer.sbatch are used to grab a specific value that gets added directly to the genome.json.

[tovahmarkowitz]

Change that. Instead of making a new create_geneinfo_bed.R, we just need to take genes.gtf and filter it to create a protein-coding only gtf file.