Phyling: phylogenetic inference from annotated genomes

Abstract Phyling is a fast, scalable, and user-friendly tool supporting phylogenomic reconstruction of species phylogenies directly from protein-encoded genomic data. It identifies orthologous genes by searching protein sequences against a curated set of hidden Markov model profiles, consisting of single-copy orthologs derived from the BUSCO database. To optimize the speed of the final inference, Phyling includes a module to filter aligned orthologs based on their phylogenetic informativeness. Finally, Phyling provides a companion wrapper for automated species tree construction using either consensus or concatenation strategies. Phyling efficiently resolves large phylogenies by optimizing memory usage and data processing. Its checkpoint system enables users to incrementally add or remove samples without repeating the entire search process. For analyses involving closely related taxa, Phyling supports the use of nucleotide coding sequences, which may capture phylogenetic signals missed by protein sequences. The benchmark results show that Phyling substantially runs faster than OrthoFinder, a reciprocal best hit based method, while achieving equal or better accuracy.