Parsing the Homologene build and building the XML document tree in python takes too much memory. Here, we built a small pipeline to:

1. Simplify the Homologene XML document tree corresponding only to species of interest
2. Extract from the simplified Homologene XML one-to-one ortholog mappings between two species of interest.

Run conda env create -f env.yml to install necessary packages.

We have included a Snakemake pipeline to extract 1-1 ortholog mappings between a specified pairs of species. These pipelines are configured by configuration files. Run snakemake all <configfile> to produce ortholog mappings with the provided config files in configs/.

The provided config files are:

• sc-sp.yml : For S. Cerevisiae vs. S. Pombe ortholog lists (using locus tags)
• human-mouse.yml: For human vs. mouse ortholog lists (using RefSeq IDs)

simplify_homologene.py extracts and outputs XML elements of the Homologene build corresponding only to the provided species Taxa IDs.

The following elements are removed from the XML:

• From HG-Entry elements:
  • HG-Entry_commentaries
  • HG-Entry_cr-date
  • HG-Entry_up-date
• From HG-Gene elements:
  • HG-Gene_domains
  • HG-Gene_location

Command line parameters:

• --input: Path to the Homologene XML
• --output: Destination for output file
• --tax_ids: Space separated list of taxonomy IDs for species of interest.

extract_homologs.py extracts and saves to disk (as a TSV) 1-1 ortholog mapping from a Homologene XML document. 1-1 homolog mappings are generated by selecting pairs with reciprocol best bitscores. (We recommend only performing this on simplified Homologene XML files since building XML document trees in memory for all species in Homologene takes an enormous amount of memory.)

Command line parameters:

• --input: Path to the Homologene XML
• --output: Destination for output file
• --tax_ids: Space separated list of two [taxonomy IDs] for species of interest
• --use_refseq_id: An optional flag. When this flag is used, RefSeq IDs will be used, otherwise locus tags will be used. (Note: if locus tags are not available for all genes w.r.t the species of interest then this flag must be used).