Identification of Protein Homologs and Domain Boundaries by Iterative Sequence Alignment

Evolutionary domains are protein regions with observable sequence similarity to other known domains. Here we describe how to use common sequence and profile alignment algorithms (i.e., BLAST, HHsearch) to delineate putative domains in novel protein sequences, given a reference library of protein domains. In this case, we use our database of evolutionary domains (ECOD) as a reference, but other domain sequence libraries could be used (e.g., SCOP, CATH). We describe our domain partition algorithm along with specific notes on how to avoid domain indexing errors when working with multiple data sources and software algorithms with differing outputs.