Improved maximum growth rate prediction from microbial genomes by integrating phylogenetic information

Abstract

Microbial maximum growth rates vary widely across species and are key parameters for ecosystem modeling. Measuring these rates is challenging, but genomic features like codon usage statistics provide useful signals for predicting growth rates for as-yet uncultivated organisms,s though current predictors often show high variance. To improve accuracy, we integrate phylogenetic signals, leveraging the evolutionary relationships among species to refine trait predictions. We present Phydon, which combines codon statistics and phylogenetic information to enhance the precision of maximum growth rate estimates, especially when a close relative with a known growth rate is available. We construct the largest and most taxonomically broad database of temperature-corrected growth rate estimates for 111,349 microbial species. The results reveal a bimodal distribution of maximum growth rates, resolving distinct groups of fast and slow growers. Our hybrid approach advances the accuracy of genome-based growth rate predictions and presents a new framework for accurate genome-based trait prediction.