Keynote
PIPER-FlexPepDock - a Novel, Fragment-Based Approach for Blind, Atom-Resolution Peptide Docking

Protein interactions mediated by short linear segments, or peptides, play crucial roles in cellular communication and regulation. Molecular details are important for our ability to study, and modulate, such interactions. However, experimental as well as computational characterization of these transient, often weak associations has been challenging.

In order to model a peptide-protein complex structure starting from a known receptor structure and peptide sequence only, a vast conformational space needs to be sampled, including both rigid body orientation as well as internal degrees of freedom of the peptide, and accurate scoring functions need to be developed to distinguish the correct conformation.

In our novel PIPER-FlexPepDock approach, we have been able to significantly advance in this challenging task by decomposing it into separate, simple steps. The major advance is due to our observation that a representative ensemble of peptide conformations that contains a bound-like conformation can be extracted based on sequence and (predicted) structure-similar fragments cut out from solved monomer protein structures (Step I in the Figure below). This allows the de-coupling of the modeling of the peptide conformation from docking. Exhaustive docking of the peptide ensemble onto the receptor structure using FFT-based PIPER (Step II), and consequent refinement using Rosetta FlexPepDock (including receptor flexibility) (Step III) allows us to generate models of unprecedented quality, starting from the peptide sequence only.

This new protocol opens up the detailed structural modeling of many new interactions, and improves our understanding of underlying principles of peptide-protein recognition.