Automated reconstruction of reproducible protein databases with RepDBmaker

RepDBmaker is a reproducible Snakemake pipeline designed for the automated construction of large-scale protein sequence databases. It enables the seamless downloading of sequences from different public databases and the integration of user-specified proteomes. The pipeline also performs standardized taxonomic annotations to all compiled sequences and contamination screening in eukaryotic proteomes. In addition, RepDBmaker generates queryable database files for Basic Local Alignment Search Tool Program (BLASTP), DIAMOND, and MMseqs2, with the option of taxonomy-aware sequence clustering to produce less redundant datasets. To showcase its utility, we used RepDBmaker to build a Tree-of-Life-scale database comprising 482 million protein sequences from 160,045 publicly available genomes and transcriptomes. RepDBmaker offers an off-the-shelf but customizable solution for researchers undertaking large-scale evolutionary projects that require extensive, well-annotated protein databases. The pipeline is available at https://github.com/Gabaldonlab/RepDBmaker.