TSEMA: interactive prediction of protein pairings between interacting families

Abstract

An entire family of methodologies for predicting protein interactions is based on the observed fact that families of interacting proteins tend to have similar phylogenetic trees due to co-evolution. One application of this concept is the prediction of the mapping between the members of two interacting protein families (which protein within one family interacts with which protein within the other). The idea is that the real mapping would be the one maximizing the similarity between the trees. Since the exhaustive exploration of all possible mappings is not feasible for large families, current approaches use heuristic techniques which do not ensure the best solution to be found. This is why it is important to check the results proposed by heuristic techniques and to manually explore other solutions. Here we present TSEMA, the server for efficient mapping assessment. This system calculates an initial mapping between two families of proteins based on a Monte Carlo approach and allows the user to interactively modify it based on performance figures and/or specific biological knowledge. All the explored mappings are graphically shown over a representation of the phylogenetic trees. The system is freely available at . Standalone versions of the software behind the interface are available upon request from the authors.

Chapter 4: Protein Interactions and Disease

“…The computational techniques based on experimental data use the relative frequency of interacting domains [19], maximum likelihood estimation of domain interaction probability [20], [21], co-expression [22], or network properties [23]–[27] to predict protein and domain interactions. The main disadvantage of empirical computations is that, by relying on an existing protein network to infer new nodes, they propagate the inaccuracies of the experimental methods.…”

Computational Approaches to Elucidating Transient Protein-Protein Interactions, Predicting Receptor-Ligand Pairings

“…Another, relatively inexpensive, way to predict protein-protein interactions does not include wet lab analysis, using instead a variety of computational approaches. These approaches can complement experimental wet lab techniques and are often supported by either the hypothesis of protein co-evolution [Tan et al 2004, Tillier et al 2006, Izarzugaza et al 2006], structural similarities [Gong et al 2005, Ogmen et al 2005 or amino-acids sequence conservation [Pitre et al 2006].…”

“…Three of them, supported by the protein co-evolution hypothesis, are: TSEMA [Izarzugaza et al 2006], ADVICE [Tan et al 2004], Codep [Tillier et al 2006]. The other three, supported by datasets of verified interactions, are: PIPE [Pitre et al 2006], PSIbase [Gong et al 2005], and PRISM [Ogmen et al 2005].…”

Integrating Bioinformatics and Computational Biology: Perspectives and Possibilities for In Silico Network Reconstruction in Molecular Systems Biology

“…By using sequence information and evolutionary analysis there are methods that predict, for example, co-evolution of residues in different proteins or gene fusion events in different genomes. Predictions of co-evolution or detection of gene fusion events are interpreted as indicating that there is a physical interaction between two proteins (see [299][300][301][302] for a review). On-line resources where information from HTP interaction experiments is deposited include the BIND database [303] the PRIME database [304], the MIPS database [305], the DIP database [306] and the MINT database [307].…”