Developing reliable ab-inito software for the interpretation of protein structure from BioSaxs data.

Biological small-angle X-ray scattering (BioSaxs) is an increasingly important method for determining the structure of proteins in solution. Data interpretation in this field is challenging due to random motions in solution and requires an initial guess (forward modelling) of the protein's shape to make a predictions from the data.

The project will build on a forward modelling techniques pioneered by the primary supervisor to develop a suite of techniques for consistent and accurate interpretation of protein structures. Example target proteins include one which interacts with meiotic telomeric DNA and one which modifies chromosomes during meiosis.

Research Project

Background: The forward model developed by Prior, utilises a parsimonious discrete curve model to interpret BioSaxs data. It satisfies local structural constraints imposed by the stereochemisty of amino acid chains and its efficacy has been demonstrated on both globular and coiled-coil proteins. This will be paired with the pioneering scattering data analysis techniques of Rambo, to develop the first reliable and widely available abinito tertiary structure prediction software for BioSaxs.

Target molecules (Davies lab): Both play a crucial role in meiotic cell division and are crucial for fertilitzation. First multiple mutants of MAJIN-TERB2 which lead to conformational change from its solved crystal structure in a manner that is predicted to affect its interaction with meiotic telomeric DNA. Its individual disruption in mice leads to meiotic arrest with failure of telomere attachments, chromosomal movement, and impaired synapsis. The second a novel helical dimeric structure of a region of MeiP22, which has an emerging role in regulating SUMO/ubiquitin modification of chromosomes during meiosis. The Mei-P22 structure does not crystallise whilst the MAJIN structure yields incompatible information in solvent by comparison to its crystal structure, so both are ideal targets for our methods.

In addition to providing fundamental science, this analysis will provide a template for how the software developed can be used in future ab-initio studies.

Research plan: The student will complete training courses in experimental SAXS techniques at Diamond. In doing so they will collect the required data for the target protein as well as training data for the theoretical tools.

In order to significantly expand the reliability of the ab-initio interpretation technique, as well as train the student in crucial theoretical soft matter tool- sets, we have constructed three theoretical work plans.

• Modelling experimental noise. This comes from experimental error and random protein motion/polymerisation. The student will learn to develop and apply statistical models for these sources and apply them to data using Python, and C++ (guided by Rambo and Prior).

• Developing improved search algorithms Sophisticated Bayesian search methods are required to ensure this the hard to navigate the tertiary fold space is explored comprehensively and parsimoniously. The student will learn to apply Bayesian sampling techniques in an optimization problem (C++, code guided by Prior).

• Development of automated post-search structural assessment. We anticipate a high number of model candidates to be yielded by the method. The student will develop methods to automatically rate the quality of these predictions. Firstly, guided by Davies, the student will learn how to translate the model predictions into the Rosetta computational protein modelling suite to generate assessable protein models from the model of [P1]. Secondly, guided by Prior, they will learn to apply topological metrics for classifying and comparing predictions for fold similarity (Python and C++).

Strategic vision: Diamond is a BioSaxs world leader, providing numerous software packages for interpreting the data obtained at its various beamlines. This gives it a unique (in the UK) ability to publicize its methods to a plethora of scientific communities. This reach will ensure the software developed has the potential for significant impact.

Training & Skills

Skills developed by the student during their time at the various institutions will include

Durham:
• Training in the use of statistical techniques such as Bayesian learning.
• Knowledge of cutting edge Topological and global geometrical metrics used to classify and compare known and new protein structures.
• Optimization and mathematical programming (C++ and Python).

Newcastle:
• The use of crucial protein modelling software such as the Rosetta protein modelling suite

Diamond:
• World-class training in state-of-the-art synchrotron measurements using BioSaxs.
• Implementation of a web-accessible algorithms for general scientific community.
• Participation in BioSaxs training workshops at Diamond.
• Publishing in high impact journals.

Further Information

Dr Christopher Prior
[Email Address Removed]
+441913343112

How to Apply

To apply for this project please visit the Durham University application portal to be found at: https://www.dur.ac.uk/study/pg/apply/

Please select the course code F1A201 for a PhD in Molecular Sciences for Medicine and indicate the reference MoSMed20-11 in the ‘Field of Study’ section of the application form.

Should you have any queries regarding the application process at Durham University please contact the Durham MoSMed CDT Manager, Emma Worden at: [Email Address Removed]