Using the Function Prediction Server and Queryome for Functional Protein Design

Accurate computational prediction of protein function is critical for guiding the functional design of proteins in the post-genomic era. The gap between the number of protein sequences deposited in databases and those with experimentally validated functional annotations continues to widen, making automated function prediction tools indispensable for researchers. In this chapter, we describe a suite of complementary computational tools for protein function prediction and literature-based functional analysis. We cover four sequence-based Gene Ontology (GO) term prediction methods available through the Protein Function Prediction web server: PFP, ESG (Extended Similarity Group), Phylo-PFP, and Domain-PFP. Each method leverages distinct algorithmic strategies, from mining distantly related sequences with PSI-BLAST to recursive multilevel sequence searches, phylogenetic distance-based reranking, and function-aware domain embeddings. We also describe Queryome, a multi-agent deep research system that performs evidence-grounded synthesis across 28.3 million PubMed abstracts, which can be used to contextualize and validate function predictions with primary literature evidence. We provide step-by-step protocols for using these tools, interpreting results, and integrating predictions into a workflow for the functional design of proteins. The Protein function prediction server is available at www.kiharalab.org/pfp and Queryome is available as a desktop application at www.queryome.app.