Molecular dynamics (MD) has matured into a powerful tool for studying dynamics and function of proteins with atomistic and femtosecond resolutions. Typical MD simulations (~microsecond timescale) produce large trajectories of high-dimensional data. A great challenge is to extract useful information about important protein properties from such large (terra- to petabytes) datasets.
Tremendous progress made in machine learning (ML) technologies has been utilized to automate, accelerate and improve accuracy of analysis of MD data . This has been demonstrated e.g. through identification of global conformational states of proteins, discovery of novel ligand binding sites and allosteric mechanisms for a wide range of proteins .
The project aims to develop and apply a novel machine learning based methodology that combines molecular dynamics with neural networks for analysis of large biomolecular simulation datasets. The tools will be tested on and applied for analysis of extensive all-atom MD simulations of the mRNA capping enzymes, namely RNMT, Capping Enzyme, and CAPAM . These enzymes are responsible for construction of the cap structure at the 5’ end of the pre-mRNA transcripts, which is required for efficient gene expression and regulation in eukaryotic cells.
The project will proceed under an established collaboration  with Professor Vicky Cowling from the School of Life Sciences (2nd supervisor of the project), an expert in capping enzymes. This will allow direct biochemical testing and validation of computational predictions.
A particular focus will be placed on understanding mechanisms of allosteric regulation of the capping enzymes. It is known that the human Capping Enzyme is regulated by the C-terminal domain of RNA Polymerase-II, whereas human RNMT – by a mini-protein RAM. However, fine details of such long-range allosteric communication are not available. In addition to understanding the mechanisms of enzymes’ recruitment to the transcription site and gene expression regulation, this will offer opportunities for developing allosteric modulators, a new strategy in drug discovery. The impact and potential of the capping enzymes as promising drug targets was recently evidenced by a spinout project RNACapRx led by Vicky Cowling and the Drug Discovery Unit in Dundee.
The developed ML framework will be made available to the community and will enhance analysis of MD data for a wide range of proteins.
1. F. Noé et al. Machine Learning for Molecular Simulation, Annual Review of Physical Chemistry (2020) 71:361-390.
2. O. Fleetwood et al. Molecular insights from conformational ensembles via machine learning. Biophysical Journal (2020) 118: 765-780.
3. a) M. Bage et al. A novel RNA pol II CTD interaction site on the mRNA capping enzyme is essential for its allosteric activation. Nucleic Acids Research (2021) gkab130; b) J. Bueren-Calabuig et al. Mechanism of allosteric activation of human mRNA cap methyltransferase (RNMT) by RAM: insights from accelerated molecular dynamics simulations. Nucleic Acids Research (2019) 47:8675–8692.