Pattern Recognition

See allHide authors and affiliations

Science  18 Nov 2005:
Vol. 310, Issue 5751, pp. 1091-1093
DOI: 10.1126/science.310.5751.1091e

Protein-protein interaction space is gradually becoming less nebulous as predictions from global two-hybrid screens of model organisms are confirmed or refuted on the basis of direct experimental trials or structure and sequence-based bioinformatic analyses. On the other hand, lower-affinity interactions between smaller, peptide-sized linear motifs and their protein partners have been more difficult to catalog, in part because they are more likely to be found in the disordered (nonhelical, nonsheet) regions of protein structures and because they may be less conserved across species.

Neduva et al. propose a bioinformatic approach for identifying these short stretches of amino acids and apply it first to a curated set of eukaryotic linear motifs and then to the two-hybrid data set from Drosophila. From the sequences of a group of predicted partner proteins, they remove well-defined structural elements and homologous regions. By assessing the nonrandom appearance of peptide motifs in what remains, they obtain rankings of candidates; two of their predictions, tested in binding assays, are peptides with affinities of 20 and 40 μM, suggesting that it might now be possible to look at weak or transient interactions in a systematic fashion (see also Ruotolo et al., Reports, Science Express, 17 November 2005). — GJC

PloS Biol. 3, e405 (2005).

Navigate This Article