Conformation-independent comparison of protein structures

The comparative analysis of protein structures is often performed in order to identify and explore similarities/dissimilarities present between target structures. Whilst many tools are available for structural comparison, the development of new tools providing different information is desirable. The work presented here concerns the development of ProSMART (Procrustes Structural Matching Alignment and Restraints Tool), a tool to aid the comparative analysis of protein structures. Primarily, the software is used for conformation-independent pairwise structural alignment, allowing identification of local similarities, and quantification of dissimilarities. Functionality also allows the identification and superposition of rigid substructures, providing output that allows visualisation of local dissimilarities by means of residue-based scoring. The ProSMART pairwise alignment method optimises the net agreement of local structures along the chain, using structural fragments. In order to maintain conformation-independence, the structure-based residue alignment does not enforce global rigidity of chains, nor domains. This feature makes the tool suited to the analysis of domain movement and other conformational changes, as well as for the identification of structural units conserved between seemingly different structures. Given an alignment, ProSMART can be used to generate external restraints on the distances between relatively close atoms, for use in the crystallographic refinement of macromolecules. Using one or more similar structures, the software generates restraints that are intended to help the target protein adopt a conformation that is more reasonable, whilst allowing global flexibility. Such restraints may stabilise refinement in some cases, especially at low resolution where experimental data alone may not be sufficient. We also present a method of Procrustes score normalisation, which allows the consideration of the significance of observed fragment scores. It is suggested that the resulting global scores for the overall pairwise agreement of protein structures may provide an interesting new way of viewing protein fold space.