Donovan H. Parks

gTranslate predicts prokaryotic translation tables with high accuracy using machine learning, outperforming existing methods and identifying new genetic code variations.

The paper identifies stop codon reassignment of UGA to tryptophan in the Actinomycetota phylum, expanding known bacterial lineages with this genetic code plasticity.

This paper identifies microbiome-derived therapeutics for IBD using a metagenome-guided approach, showing improved disease markers and safety in preclinical models.

AnnoView provides a web tool for exploring and comparing gene neighborhoods across microbial genomes, supporting both large databases and user-uploaded datasets.

CheckM2 improves MAG quality assessment using machine learning, outperforming its predecessor and handling novel lineages and reduced genomes.

Greengenes2 aligns 16S rRNA and shotgun metagenomic data using a whole-genome phylogeny, showing agreement in principal coordinates, taxonomy, and phenotype effect size.

GTDB-Tk v2 improves memory efficiency for taxonomic classification by using a two-step approach with reduced memory usage while maintaining accuracy.

The paper introduces nanotext, a novel representation of genome content for machine learning, and applies it to predict culture media and growth temperatures from metagenome assembled genomes.

AnnoTree visualizes and explores functional annotations across 24,000 bacterial genomes, revealing widespread phylogenetic patchiness in gene families.