CCP4 Study Weekend 2019 - AMPLE: Unconventional Molecular Replacement with ab initio models and more скачать в хорошем качестве

CCP4 Study Weekend 2019 - AMPLE: Unconventional Molecular Replacement with ab initio models and more

Molecular Replacement (MR) is the most popular route to protein crystal structure solution but requires the availability of a suitable search model. Where an existing structure in the PDB is not sufficiently similar to the unknown target for straightforward use, unmodified or lightly edited, then successful MR may require the modelling of the target ab initio (for new folds) or more sophisticated processing (in the case of distant homologues). AMPLE is an MR pipeline designed to exploit structural bioinformatics methods to extend the reach of unconventional MR. It was originally designed to work with fragment assembly ab initio models and proved to be reasonably successful with small proteins. The method uses clustering of large numbers of individual structural models to predict plausible folds. These clusters are converted into structural ensembles which are further truncated to produce smaller but potentially more accurate search model ensembles. Covariance-derived residue pair contact predictions an be used to produce ab initio models, by the fragment assembly approach or simple distant geometry methods. The additional information from contact predictions improves the modelling accuracy, thereby extending the range of proteins tractable by this approach. The contact predictions also offer an alternative to clustering to predict which ab initio models are most accurate. As covariance-directed structure modelling continues to rapidly develop there is even the possibility that crystallographers may in future simply retrieve ab initio models from databases to solve novel folds: preliminary results show this to work in some cases. Where distantly homologous structures are available in the PDB, but are not suitable for conventional MR, a number of structural bioinformatics strategies may help. AMPLE’s cluster and truncate approach turns out to be useful in many scenarios for trimming more divergent and/or less reliable portions of a structural ensemble. Where multiple distant homologues are available, such ensembles may be composed of the structures themselves or of homology models built using them as templates. Where a single distant homologue is available, structural bioinformatics methods that exploit the correlations between structural conservation and measures of rigidity or packing, can be used to produce value-added ensembles. Quick and simple normal mode calculations, for example, can extend the usefulness of single distant homologues by identifying the more rigid, better conserved structural core.