pFlexAna: Detecting Conformational Change in Remotely Related Proteins
Protein conformational changes play a critical role in vital biological functions. Describing conformational shifts or domain rearrangements usually requires two solved structures of the same protein. However, in the absence of such data, it may be possible to detect probable conformational changes for a particular protein using a structurally related protein.
We offer a web server that can detect conformational changes and domain rearrangements between two proteins without requiring any sequence similarity. Probable conformational changes are highlighted by automatic detection and visualization of common rigid domains between two structures. Alignments can also be clustered according to the similarity of their transformations.
Analyzing two structures
The input for the server is two protein structures. You can specify PDB codes or upload your own PDB files. We model noise as Gaussian, and the standard deviation sigma is a parameter that is required (see [1] for details). We have observed useful results with values between 0.2 and 0.8; a smaller value is stricter on the rigidity criteria (the lower the sigma, the more similar two fragments have to be before they're matched). You may wish to use smaller values for highly similar or homologous proteins, and higher values for others. The last input is the number of desired clusters, k, for the alignments detected. The server will attempt to cluster the alignments detected into k clusters which contain alignments with similar transformations.
The outputs of the server are matched pairs of fragments from the two proteins. These aligned fragments represent structurally conserved regions between the two structures. For each pair, the proteins are aligned on the respective fragments and a visualization is provided. The alignments are clustered according to the desired number of clusters.
Links
References
- A. Nigham and D. Hsu. (2007) RECOMB, pp. 396-411