ProteinFP is an open-source computational pipeline that predicts protein function from structure + sequence evidence, no experimental annotations required.

What it predicts:

• Active sites and catalytic residues
• Binding interfaces and pockets
• Allosteric sites
• Chemical environment per residue
• Protein-protein interactions
• GO term annotations (MF, BP, CC)
• Drug target modality recommendations (small molecule, ADC, CAR-T, PROTAC)
• Small Molecule and Antibody Evolutionary Engine

How it works: Input is a UniProt ID. The pipeline fetches the AlphaFold structure, runs homology search (BLAST vs Swiss-Prot + InterProScan), and passes results through 13 chained modules to produce a full function report. Validation: Tested on 75+ well-characterised proteins spanning kinases, proteases, transcription factors, and GTPases. Mean GO recall >0.94 across MF/BP/CC. Quick start:

pip install proteinfp
proteinfp --uniprot P28593   # Trypanothione reductase (Chagas disease)

PyPi: ProteinFP PyPi | GitHub: ProteinFP GitHub | Feedback and contributions welcome!