The Bioinformatics Unit of the CBMSO was founded in 2002 by Dr.Ángel Ramirez Ortiz. After the sad and premature loss of Dr.Ortiz in may 2008, the group is now directed by Dr. Ugo Bastolla. The line on drug design, directed by Dr. Antonio Morreale between 2008 and end of 2012, has now disappeared but the software that it developed is still available, although it is not anymore maintained. We also provide Bioinformatics facility to the CBMSO, especially through molecular modeling provided by the Bioinformatician David Abia.
Our group is active in the field of the computational structural biology of proteins. Our main research lines are the computational study of protein structures, stability and evolution, computational genomics, and theoretical ecology, in particular applied to bacterial communities. We summarize these lines below.
One of the main goals of our research consists in improving the methods for predicting protein folding stability that we developed some time ago. We apply these methods to the structure-aware modeling of molecular evolution for phylogenetic inference. This has also application for predicting the effect of mutations, for protein structure prediction through threading, and for predicting protein-protein interactions through the analysis of correlated mutations. We have applied these methods to characterize selective pressures favoring stability against unfolding (positive design) and misfolding (negative design), and fast folding. In this way, we found an interesting relationship between folding stability, population size and mutation bias that suggests that this interplay can explain why intracellular bacteria with reduced effective population size tend to evolve with a mutation bias favoring A+T nucleotides, hence more hydrophobic proteins. Read more
Another main goal of our research is the quantitative and, if possible, predictive understanding of the functional dynamics of the protein, in particular the dynamical couplings between residues that are key for allosteric interactions, and of the conformation changes that occur during the biological activity and the evolution of proteins. This has applications for rationalizing the mechanism of protein action, improving homology-based protein structure prediction, and improving models of protein-ligand docking. To pursue these goals, we developed a new model of normal mode analysis that predicts protein equilibrium dynamics from protein structure using torsion angles as degrees of freedom. This method allows to treat large systems much more rapidly and accurately than previous ones, producing physically realistic movements of up to several Å at once. We are currently working at a force field that allows us to predict conformation changes upon ligand binding in order to improve our drug design protocol and to refine homology models by predicting rearrangements of protein structure upon mutation. Read more
Our second goal consists in automatically classifying protein domains using their structure similarity and analyzing the structural and functional changes in their evolution. This work lead us to recognize that protein domains can be classified on a tree only for large enough similarity, whereas for lower but significant similarity their relationships must be described as a network, both due to their evolutionary origin through fragment assembly even below the domain level and because of evolutionary accelerations upon function changes. We are currently using our normal modes model to quantitatively characterize function change and function conservation in the evolution of proteins. Read more
We are also interested in the properties of the hundreds of proteins that form the Centrosome. We have found that these proteins are predicted to be much more disordered, coiled-coil and phosphorylated than control proteins of the same organism. These properties confer structural and regulatory plasticity to the Centrosome and are enhanced for organisms with a larger number of cell type, and they arose in evolution mainly through large insertions of disordered fragments that happened more frequently in evolutionary branches where the number of cell types increased significantly. We are further investigating the relationship between disorder, phosphorylation, electric charge, and the size of centrosomal proteins. Read more
Flowering plants and insects are very diverse groups, characterized by mutualistic interactions (advantageous for both partners) with nested structure, in the sense that specialist species tend to interact with generalist species. We developed and solved a mathematical model that predicts that nested mutualistic interactions decrease the effective competition of a species community and allow it to maintain a larger biodiversity. Read more
We are applying the results above for studying the interactions between bacterial taxons, which we predict from co-occurrence matrices derived from metagenomic studies. We have found that there are many more aggregations of taxons (“mutualism”) than exclusions (“competition”), and we see that such mutualistic interactions favor the cosmopolitanism of bacteria, i.e. their capacity of living in quite different environments. We have developed an algorithm for reconstructing bacterial communities that present significant aggregation starting from 16s RNA data. Read more
The research line on structure-based drug design was initiated by Ángel Ramirez Ortíz when he was in the laboratory of Prof. Federico Gago in Alcalá de Henares Universtiy. After Ángel passed away, this line was directed by Dr. Antonio Morreale, a researcher with a strong experience in the field. This line terminated when Dr. Morreale left the group at the beginning of 2013, but the software that it generated is still available, although not anymore maintained. Among this software, it has to be mentioned an integrated computational platform to perform massive Virtual Screening (VS) experiments, which allow to extract the most promising candidates able to interact with a protein of clinical interest from large databases of small molecules . This Virtual Screening Data Management on an Integrated Platform (VSDMIP) has been proved to be an useful tool for developing new methods, and it generated several patents in collaboration with experimental groups.
Another area in medical chemistry that was of interest of the research group directed by Ángel Ramirez Ortíz and subsequently by Antonio Morreale was 3D-QSAR (3-Dimensional Quantitative Structure-Activity Relationships). In this regard, the group developed gCOMBINE, a Java-written graphical user interface (GUI) for performing COMparative BINding Energy (COMBINE) analysis on a set of ligand-receptor complexes that allows predicting QSAR, with the aim to understand the molecular basis of specificity of ligand-receptor binding affinities and to guide molecular modifications that improve binding affinity and/or selectivity. Read more
Finally, Ángel Raḿírez Ortíz was very interested in molecular modeling, in particular in developing new methods and assessing them in the CASP experiment. These studies are now continuated by David Abia, responsible of the Bioinformatics facility of the CBMSO, who is applying molecular modelling techniques in collaboration with different research groups within and outside the CBMSO, which mainly involve molecular mechanics and dynamics simulations followed by quantitative free energy calculations. Read more