Section: Research Program

Axis 3: Protein 3D structure

The three-dimensional (3D) structure of proteins tends to be evolutionarily better preserved during evolution than its sequence. Finding structural similarities between proteins gives deep insights into whether these proteins share a common function or whether they are evolutionarily related. Structural similarity between two proteins is usually defined by two functions – a one to- one mapping (also called alignment) between two subchains of their 3D representations and a specific scoring function that assesses the alignment quality. The structural alignment problem is to find the mapping that is optimal with respect to the scoring function. Protein structures can be represented as graphs, and the problem reduces to various combinatorial optimization problems that can be formulated in this framework: for example finding the maximum weighted path [1] or finding the maximum cardinality clique/pseudo-clique [6].

In most cases, however, suitable conformations for a given protein are unknown. To support this statement, we point out that the number of deposited protein conformations on the Protein Data Bank (PDB (http://www.rcsb.org/)) recently reached the threshold of 110,000 entries, while the UniProtKB/TrEMBL (http://www.ebi.ac.uk/uniprot/TrEMBLstats) database contains more than 50 million sequence entries, all of them potentially capable for coding for a new protein. In this context, distance geometry provides powerful methods and algorithms for the identification of protein conformations from Nuclear Magnetic Resonance (NMR) data, which basically consist of a distance list concerning atom pairs of the protein. We are working on the discretization of the distance geometry, so that its search space becomes discrete (and finite!), for making it possible to perform an exhaustive exploration of the solution set.