The Bosse Group at CSSB has developed a structure prediction database covering the proteomes of all human Herpesviruses. A description of their database, HerpesFolds, and insights it has revealed about new protein families was recently published in the scientific journal Nature Communications.
There are nine Herpesviruses known to infect humans, which cause various common diseases such as cold sores, chicken pox, infectious mononucleosis, and congenital disorders. In fact, more than 90% of adults have been infected with at least one of the nine human herpesviruses, making them medically important human pathogens.
Understanding the interactions between herpesviruse and human host cells is critical for the identification of novel targets and the development of new antiviral therapies. To understand the molecular mechanisms at play in these interactions, researchers need the structural information for each of the 70 to 180 viral proteins in a single herpes virus. Experimental structural information is, however, not available for most of the viral proteins, and viral structures are absent from the EMBL/DeepMind AlphaFold protein structure prediction database.
To remedy this, the Bosse group predicted the structures of all nine human herpesvirus proteomes, totaling 844 individual proteins, and made them available in their open-access database HerpesFolds. “Not only did we critically evaluate the accuracy of our predictions with stringent quality scores and thresholds, but we also present the structural information in a format that is easily searchable to other researchers,” notes the paper’s first author, Timothy Soh.
In addition to providing the viral structural predictions, the HerpesFolds database also clusters these predictions into structure similarity groups. “This feature of the database enabled us to identify new families, such as the HCMV UL112-113 cluster, which is conserved in alpha- and beta-herpesviruses,” notes the paper’s corresponding author Jens Bosse. Additionally, using structure similarity searches against cellular proteins, the researchers were able to propose new functions for uncharacterized viral proteins. “Overall, we show that system-wide structure predictions can reveal similarities between viral species and identify potential protein functions,” explains Soh.
The HerpesFolds database has been open-access since its inception, and the Bosse group has actively incorporated community feedback into the database design. “HerpesFolds already receives several thousand hits per month,” explains Bosse, “We are happy to have developed a tool that is being widely used in the herpesvirus research community.” Building on HerpesFolds’ success, the Bosse group has already begun to develop their next protein structure predictions database. The group recently released structure predictions for multiple poxviruses on their PoxFolds database.
Original Publication:
Soh, T.K., Ognibene, S., Sanders, S. et al. A proteome-wide structural systems approach reveals insights into protein families of all human herpesviruses. Nat Commun 15, 10230 (2024).