Inria / Raweb 2004
Project-Team: MODBIO

Search in Activity Report, year 2004:


Project-Team : modbio

Section: New Results

Keywords: Statistical learning, protein secondary and tertiary structure, kernel engineering, disulphide bridges.

Protein structure prediction

Participants: Yannick Darcy, Yann Guermeur, Frédéric Sur.

Knowing the three-dimensional structure of a protein can greatly help to infer its function. Predicting this tertiary structure from the sequence of amino acids (or primary structure), remains one of the central open problems in structural biology. This is the subject of the «GENOTO3D» project that we coordinate. This year, our main efforts have been concentrated on the implementation of M-SVMs for secondary structure prediction and disulphide bridge prediction. Those efforts have mainly taken two aspects. On one hand, we have derived uniform convergence results of the empirical risk with the aim to use them as objective function in a procedure of model selection. On the other hand, we have started to study the possibility to derive kernels with good discrimination properties from pair-HMMs.

Another contribution to predictive structural biology, although not directly related to the structure of globular proteins, is a collaboration with the team of Gilbert Deléage, at IBCP Lyon. It deals with the identification of amphiphilic helices. This work provides us with the opportunity to assess the efficiency of the M-SVM dedicated to protein sequence processing [15] in a context different from its initial use, which calls for significant changes in the parameterization.