Format

Send to

Choose Destination
J Mol Biol. 2005 Sep 30;352(4):1002-15.

Assessing protein co-evolution in the context of the tree of life assists in the prediction of the interactome.

Author information

1
Structural Bioinformatics Group, Division of Molecular Biosciences, Imperial College London, London SW7 2AZ, UK. pazos@cnb.uam.es

Abstract

The identification of the whole set of protein interactions taking place in an organism is one of the main tasks in genomics, proteomics and systems biology. One of the computational techniques used by many investigators for studying and predicting protein interactions is the comparison of evolutionary histories (phylogenetic trees), under the hypothesis that interacting proteins would be subject to a similar evolutionary pressure resulting in a similar topology of the corresponding trees. Here, we present a new approach to predict protein interactions from phylogenetic trees, which incorporates information on the overall evolutionary histories of the species (i.e. the canonical "tree of life") in order to correct by the expected background similarity due to the underlying speciation events. We test the new approach in the largest set of annotated interacting proteins for Escherichia coli. This assessment of co-evolution in the context of the tree of life leads to a highly significant improvement (P(N) by sign test approximately 10E-6) in predicting interaction partners with respect to the previous technique, which does not incorporate information on the overall speciation tree. For half of the proteins we found a real interactor among the 6.4% top scores, compared with the 16.5% by the previous method. We applied the new method to the whole E.coli proteome and propose functions for some hypothetical proteins based on their predicted interactors. The new approach allows us also to detect non-canonical evolutionary events, in particular horizontal gene transfers. We also show that taking into account these non-canonical evolutionary events when assessing the similarity between evolutionary trees improves the performance of the method predicting interactions.

PMID:
16139301
DOI:
10.1016/j.jmb.2005.07.005
[Indexed for MEDLINE]

Supplemental Content

Full text links

Icon for Elsevier Science
Loading ...
Support Center