Section: New Results

Axis 2 : Integrative Multi-Component Assembly and Modeling

EROS-DOCK algorithm and its extensions

We have adapted our EROS-DOCK protein-protein docking software [35], [19] to account for experimental knowledge on the protein-protein interface to be modeled. Indeed, structural biology experiments can identify pairs of amino-acids from each protein in a protein-protein interface that are likely to be in close contact. This additional restraint is used to pre-prune the 3D rotational space of one protein toward another, by eliminating cones of rotations that cannot fulfill the distance between the two points at the protein surfaces. Using a single restraint permits to decrease the average execution time by at least 90 percent.

We also developed a new version of EROS-DOCK for multi-body docking (modeling assemblies of more than 2 proteins), using a combinatorial approach. We assembled trimers by docking in a first stage all possible combinations of pairs of proteins involved in the multi-body complex. Possible trimer solutions are assembled by fixing one protein, the “root-protein” (protein A, say) at the origin and by placing the other two around it using the transformations,T[AB] and T[AC], from the corresponding pairwise solution lists returned by EROS-DOCK. If the three transformations together form a near-native (biologically relevant) trimer structure, then it is natural to suppose that T[BC] should be found in the list of B-C pairwise solutions.

Both extensions of the EROS-DOCK algorithm reported last year and published early this year [19] have been presented by Maria-Elisa Ruiz Echartea at the 2019 CAPRI meeting in april 2019 (http://www.capri-docking.org/events/) and at the MASIM meeting in november 2019 [28]. These results are part of her PhD Thesis that was defended on december 18, 2019 (the thesis will soon be available on HAL). A paper describing EROS-DOCK adaptation to multi-body docking in under revision in Proteins.

Protein docking

The regular participation of the Capsid team to the CAPRI challenge is acknowledged through its contribution to the review published this year on CAPRI round 46 [17].

We also contributed to an evaluation of docking software performance in protein-glycosaminoglycan systems [22].

3D modeling and virtual screening

We have built a 3D model by homology of a new class of relaxase involved in the horizontal transfer of DNA in a group of bacteria called Firmicutes [21].

We also built a 3D model of a chemosensory GPCR as a potential target to control a parasite in plants [13].

Virtual screening was applied on various targets in a re-purposing strategy and led to the discovery of small molecules active against invasive fungal disease [14], [18].