New methods for protein structure prediction

Protein structural information is crucial for an understanding of protein function and evolution. Currently, only there is only experimental data for a tiny fraction of the protein universe. Homology modelling allows structure predictions for many more proteins, but there is no structural information at all for a large number of families. Membrane protein families are particularly poorly understood due to the extra experimental difficulties they bring.
The project will entail application of new methods for proteins structure prediction, the state-of-the-art fragment assembly ab initio program Rosetta (Bradley et al., 2005) and/or the contact prediction-based EVfold method (Hopf et al., 2012). The proteins to be addressed may be of various origins, including locally produced genomes and proteomes, but are likely to include families of membrane proteins. Antimicrobial proteins of unknown structure may also be a specific target. In order to achieve the fullest possible picture, a battery of structure-based function prediction methods may be applied to models produced and those data complemented by sequence-based information. Analyses may extend to exploitation of patterns of genomic context and co-occurrence for functionally relating gene products. There may also be opportunities for the student to further the development of AMPLE, a joint Liverpool-CCP4 program for solving crystal structures using ab initio models (Bibby et al., 2012).
The project offers an exciting chance to work at the cutting edge of protein bioinformatics, leveraging information from structure predictions, complemented by a broad range of additional methods, to shed light onprotein function. The project may involve the chance to work at the A*STAR institute in Singapore for 2 years.


Training:
The project focuses on the fast-moving cutting edge of protein modelling and will provide extensive training in the current leading program Rosetta. Ab initio and contact prediction-based modelling are among the most exciting areas of contemporary bioinformatics and the student will be well placed on completion of the project to exploit the ever-increasing possibilities for its application. A full exploration of the predictive power of ab initio models for protein function requires an integrated approach to understand structure-function and their co-evolution with time. The student will therefore also be exposed to the full spectrum of protein sequence- and structure-based methods and to techniques for inferring functional relationships from genomic data.