Inverse statistical physics approaches for the modeling of protein families
| When |
Sep 14, 2015
from 11:00 to 12:00 |
|---|---|
| Where | Amphi Schrödinger |
| Attendees |
Simona Cocco |
| Add event to calendar |
vCal
iCal
|
Massive sequencing techniques make available ever-increasing numbers of protein sequences. A fundamental problem is to extract structural and functional information from those sequence data, and to identify sequences sharing common structure and/or function (corresponding to a given family). In this talk I will review how statistical physics tools can be helpful to model the (wide) distribution of sequences in a family, and how maximum-entropy approaches allow us to infer the 'energetic' parameters of models from the sequence data. I will show how the inferred models provide information on the three-dimensional structures of the proteins, and could be useful to design new, artificial proteins sharing the features common to the family. I will also address more conceptual questions about the validity of the maximum-entropy approaches on synthetic data corresponding to lattice-based proteins, which are controlled albeit 'realistic' models of proteins.