Protein-protein association is essential for the virtual totality of biological processes in living organisms. We want to understand the formation of specific complexes between proteins, and be able to make structural and energetic predictions about such complexes. For that, we are working on the development of new predictive methods, based on physico-chemical principles, with the help of supercomputation resources. Our group is highly multidisciplinar, and strongly motivated to contribute to this challenging scientific area.


Our research focuses on theoretical and computational approaches to protein-protein association. Two major goals of our research are: i) development of computational tools for protein docking and binding site prediction, and ii) understanding the mechanism of protein-protein association.

We aim to develop and optimize computational algorithms for characterizing and understanding protein-protein interactions, which remains one of the most important challenges in Structural Biology. Among others, we have developed software for prediction of protein-protein complex structure by docking simulations, identification of binding sites on protein surfaces, prediction of hot-spot residues from energy calculations, etc. On the most practical side, we want to provide rationales for protein interactions of biological and therapeutical interest. With this aim, we have established collaborations with different experimental laboratories, and have thus provided atomistic models for protein-protein complexes involved in different biological processes, such as signal transduction, plant immune system, drug transport in bacteria, and electron transfer during photosynthesis. Our models are being continuosly evaluated both by our in-house benchmarks, and also externally through the CAPRI competition.

The ultimate goal of our research is to understand the subtle determinants of the specificity of protein-protein association, which in turn will help us to make more accurate predictions of complexes of biological and therapeutical interest. All this knowledge will also help to understand the interaction of small molecules with protein-protein interfaces, with the goal of designing compounds capable of inhibit protein interactions of therapeutical interest.


Algorithms for protein-protein docking based on FFT and Global Energy Optimization
We are integrating the energy function that we previously developed for scoring of rigid-body docking poses, in a grid search using Fast Fourier Transform algorithms to speed up calculations.

Prediction of protein interaction sites and hot-spots
We are extending our powerful binding site predictors, such as ODA and NIP, to the identification of hot-spot residues at binding interfaces. All these tools will be applied at proteomic scale.

Understanding protein-protein association mechanism
A deep understanding of protein-protein association will be essential, not only for accurate modelling and prediction of complex structure, but also for designing small-molecule inhibitors of protein-protein complexes. We are applying physical principles to study the energy landscape of the process of association between two proteins.

Flexible algorithms for protein-protein docking
Treatment of flexibility is currently the bottleneck in protein docking. We will use the capabilities of MareNostrum to develop new algorithms based on molecular dynamics approaches.

Prediction of electron-transfer complexes
In collaboration with other laboratories, we are modelling different complexes involved in electron transfer, such as the interaction between ferredoxin-NADP+ reductase and the proteins flavodoxin and ferredoxin, or the interaction between cythochrome c oxidase and the proteins plastocianin and cyt c6.

Prediction of multi-protein systems
We are also collaborating with many experimentalist groups, on modelling of challenging multi-protein systems such as the bacteria drug pump, signal transduction interactions, plant immune system, etc.