Thousands of predicted protein structures released

Researchers have deposited over 6,300 protein structures in a freely available database. The models result from protein structure predictions completed by David Baker’s group at the University of Washington. The team hopes that access to predicted structures will allow researchers to answer biological questions about their proteins more quickly. The structures are available in the protein family (Pfam) database, which arranges proteins by sequence alignment and function. Baker’s team used amino acid sequences of functional domains shared between proteins to make their structural models using the trRosetta algorithm. Alex Bateman, head of protein sequence resources at the European Bioinformatics Institute, likens Pfam to a periodic table, but for proteins. He hopes the data will be useful to many biologists, including those who want to explore how protein domains and functions evolved, as well as structural biologists, who might apply the predicted structures to translate experimental data into other models. These new models mean that 88% of the protein families in the database now have an experimentally derived structure or one modeled by the Baker group. Bateman says he can also use the structures to classify proteins in the database. Instead of “fumbling around” in the dark, he says, someone has turned on the light.