Christophe DessimozFeb 1, 2024
Our lab is excited to release FastOMA, a rewrite of OMA orthology algorithm that scales linearly, while maintaining OMA’s high accuracy. FastOMA can process all >2000 eukaryotic UniProt ref proteomes within 24 hours. Try it out https://github.com/DessimozLab/FastOMA
Preprint https://www.biorxiv.org/content/10.1101/2024.01.29.577392
GitHub - DessimozLab/FastOMA: FastOMA is a scalable software package to infer orthology relationship.

FastOMA is a scalable software package to infer orthology relationship. - GitHub - DessimozLab/FastOMA: FastOMA is a scalable software package to infer orthology relationship.

GitHub
Show threadChristophe Dessimoz
Since introducing OMA 20y ago, we’ve strived to maintain high accuracy. However, scaling became a challenge with increasing data. We have been in good company: even fast methods like SonicParanoid and OrthoFinder scale quadratically, struggling beyond a few hundred genomes.
Feb 1, 2024 at 11:31amWeb
Show threadChristophe DessimozFeb 1, 2024
After trying many different ideas over years, we're happy to have found a scalable solution without sacrificing accuracy. It works by clustering homologs using k-mer based placement, followed by a highly parallelised taxonomy-guided processing of families using subsampling.
Show threadChristophe DessimozFeb 1, 2024
Little note to fellow orthology geeks: OMA now relies on MSAs and trees rather than pairwise comparisons! The details are in the preprint https://www.biorxiv.org/content/10.1101/2024.01.29.577392. Would love to hear your feedback. Want to be part of the team? We are hiring http://tinyurl.com/BioinfPosition