Evaluating an Alphafold-based platform for the discovery and validation of dominant-negative protein fragments

Authors

• AC Castroverde
• JH Williams
• AJ Hu
• JL Meier

Abstract

Peptide tiling is a method in which small segments of a protein interactor are screened to identify fragments capable of exerting a regulatory function, for instance, inhibition. One challenge of this approach is that – due to the vast size of protein sequence space – screening large libraries of fragments from potential interactors is often prohibitively costly in terms of time and resources. New protein structure prediction technologies, such as Google DeepMind’s AlphaFold, have the potential to decrease experimental hurdles by allowing researchers to focus on high confidence interactors. However, few studies have defined the correlation between the confidence of predicted interaction of peptide fragments and their corresponding functional effects. In this project, we are developing a high-throughput approach to predict dominant negative protein fragments using AlphaFold and then correlating those predictions with growth detects caused by essential fragments in cell culture. Our initial studies have focused on compiling a catalog of enzymes implicated in cancer pathogenesis, their interactors, and protein fragments. Focusing on the Ras-Raf1 interaction, AlphaPulldown identified specific sequences predicted to interact that lie at known protein-protein interfaces and highlighted the importance of mpDock/pDock scoring model in predicting experimentally validated interfaces. These studies set the stage for the large-scale discovery of dominant negative protein fragments and their use as biological probes and in proof-of-concept proximity medicine platforms.

Scientific Focus Area: Computational Biology

This page was last updated on Tuesday, August 6, 2024