AlphaFold Database Expands to Include Protein Complexes of Common Viruses

Thousands of protein pairs from disease-causing viruses like monkeypox have been added to the AlphaFold protein-structure database to help with pandemic preparedness. Researchers at various organizations analysed the sequences of 41,774 proteins from around 2,800 viruses, including the ones that cause mpox, measles and hepatitis B.

Database Expansion

Earlier this year, scientists added predicted structures for 1.7 million pairs of interacting proteins from 20 organisms, which marked the first time protein complexes were included in the database maintained by the European Molecular Biology Laboratory's European Bioinformatics Institute in Hinxton, UK. However, viruses have remained a blind spot because many viral proteins act in concert with partners and lack high-quality entries. When certain viruses replicate, their RNA molecules translate into a polyprotein that gets cut into individual functional molecules, making it hard to see where a viral protein begins and ends from its genetic sequence alone.

Predictions And Limits

To fix this, researchers at the Swiss Institute of Bioinformatics in Geneva found accurate sequences for thousands of viral proteins cut from polyprotein chains. Using AlphaFold2, teams predicted structures for 40,746 homodimers of identical protein strands and nearly 1.7 million heterodimers of different molecules. Out of these, only 2,749 homodimers and 5,279 heterodimers met the accuracy bar for the official database, though all predictions are publicly available. The models still lack sugar molecules that help viral proteins hide from the immune system, and some proteins work in larger groups like trimers that require further testing.