Technical Highlight - February 2010
Short description: Find structural similarities without relying on fold classification by using the MarkUs server.
Making reasonable guesses about a protein's function from its structure is often, but not always, straightforward when two proteins are very similar, but what if two proteins are similar in just a small area? The MarkUs server provides a way to exploit sequence and structural relationships, whether close or remote.
Traditional structural classification of proteins on the basis of the composition and orientation of their secondary structural elements has worked well for identifying proteins with similar properties. The popular databases SCOP and CATH use these criteria to organize proteins into discreet classes or folds, meaning that one protein domain cannot be classified under two different folds.
But geometric similarities between particular regions of proteins that have different global topologies are increasingly being recorded, suggesting that remote relationships can indicate a functional link.
Barry Honig and colleagues from PSI NESG set out to discover whether structural fragments containing as few as three secondary structural elements can be used to uncover a common function. This idea has been discussed before, but this is the first time it has been used to suggest a function and the first time software based on this approach has been available.
The software they developed, the MarkUs server, integrates several sequence- and structure-based analysis methods, such as DALI, Psi-BLAST and DelPhi, to characterize the biochemical and biophysical properties of a protein structure and to suggest structural neighbors. In particular, DelPhi is important because it calculates electrostatic potential, which is useful for inferring membrane and DNA-binding regions or the enzymatic activity. The ability to search for structural relationships without relying on classification significantly increases the number of functional predictions that can be made. MarkUs then allows the analysis of these predictions using various annotation databases such as GO, UniProt, LS-SNP and ChEBI.
Such 'new generation' integrated computational and data-retrieval tools should allow a researcher to explore sequence, structural and functional databases in such a way as to develop and validate hypthotheses.
D. Petrey, M. Fischer & B. Honig Structural relationships among proteins with different topologies and their implications for function annotation strategies.
Proc. Natl Acad. Sci. USA 106, 17377-17382 (2009). doi:10.1073/pnas.0907971106