Software for experimentally driven macromolecular modelling

Every natural phenomenon, including life itself, is determined by the way simple atoms associate to form molecules. Molecules are intrinsically flexible: depending on local physical conditions they can change shape, associate into stable complexes, or vice versa dissociate. By studying a molecular atomic arrangement (i.e. structure) and its associated dynamics, direct insights into its function can be obtained. Acquiring this knowledge is an essential driver for the development of new biochemical and medical applications. A large palette of experimental techniques has been designed to provide information ranging from low resolution data about molecular shape, to the coordinates and dynamics of individual atoms.

Nevertheless, a single experimental technique can rarely capture all aspects of a system under study. Furthermore, experiments can be sometimes too expensive, complex or dangerous to perform. In this context, computational approaches represent an essential tool. By combining information from different experimental techniques and physical knowledge of atomic interactions, models describing a system can be generated. Such models can help rationalize experimental data, and produce new testable hypotheses to guide further experiments.

The overarching goal of my work is the creation of tools for interpreting and exploiting experimental data targeting the modelling of molecular systems at an atomistic level. My main focus will be an important class of biologically relevant molecules: proteins. Importantly, instead of representing them as a unique structure, I will treat them as ensembles of possible structures (a.k.a. conformations). My specific objectives are:
- assess and visualize experimental data against an ensemble of possible protein conformations. This will allow a subset of alternate conformations consistent with experimental data to be determined;
- predict the arrangement of multiple flexible proteins into a complex so that the produced model is consistent with experimental data. To reach this goal I will develop a powerful optimization engine running on high performance computing, and a new robust method to represent and assess electrostatic interactions;

These new methodologies will be implemented in software allowing complex experimental data to be interpreted with unprecedented accuracy and clarity, and guide the design of new experiments. I will strive to make my software not only easy to access, but also easy to use by creating graphical user interfaces.

The resulting methods will be applied to the study of integrins. These proteins assemble into complexes playing an essential role in cells' adhesion, migration and controlled death. Their malfunction leads to a wide range of diseases, most notably autoimmiunity and cancer. A wealth of experimental data is available. However, owing to inconsistencies between them, no agreement has been reached about integrin exact mechanism of action. Our methods will contribute to rationalizing the available data on the basis of molecular flexibility, shedding a new light on integrin mechanisms and ultimately paving the way to new therapeutic approaces.

My research will take place in Durham University's chemistry department. There I will profit from interactions with a range of experimental collaborators studying molecular conformations with a variety of different techniques, and from the excellent local high performance computing centre. My workload will be shared with a postdoctoral research assistant.

Molecular modelling allows performing in silico experiments that would be too expensive, complex or dangerous. I will make my software free, easy to access and to use. The objective is to allow non-computational scientists to operate it. I am convinced that achieving this goal will lead to considerable use of my methodology amongst a wide range of bioscientists in both academia and industry. I believe that this will greatly contribute to shedding light into important biological targets, connected to potential biotechnological and biomedical applications. This is well represented by the potential outcomes of my planned collaborations. The structural characterization of enzymes involved in Apicomplexan sphingolipid biosynthesis, for instance, has the potential to open the door for the development of new therapeutic approaches targeting illnesses such as malaria or toxoplasmosis. As another example, small Heat Shock protein mechanism has been linked to a wide range of diseases including cataract, cardiomyopathies, motor neuropathies and degenerative diseases caused by ageing. This research has therefore the potential of positively impacting national and international health.