Machine learning algorithms for autonomous multiparametric imaging and analysis of DNA molecules and interactions
Marie Skłodowska-Curie Actions doctoral network - spm 4.0
Deadline: January 2025. Fully funded position with stipend, open to international students.
Over the past 60 years, we have discovered many of the rules which determine how our genetic makeup affects our health, from Rosalind Franklin’s pioneering discovery of the helical structure of DNA, through to the human genome project. However, we still do not have the tools to measure DNA’s secondary, mechanical code, which affects nearly all interactions and therefore our wellbeing at a cellular level. This is due in part to the complexity of cellular DNA, caused by its innate flexibility, compaction in the nucleus, and manipulation by DNA-processing enzymes. These processes cause DNA to adopt a vast range of intricate structures, conformations and topologies which are hard to quantify as they occur at the nanometre length scale.
We have developed Atomic Force Microscopy (AFM) methods which enable us to routinely visualise single DNA molecules with sub-molecular resolution. This allows us to measure the twist, writhe, and topology of individual DNA molecules and quantify their mechanical and conformational structure. This project will develop new deep-learning image analysis methods to identify, segment and trace individual DNA molecules from topographic images.
Objectives:
• To train and benchmark machine learning models for AFM imaging of DNA molecules interacting with a range of proteins (collaborating with Gwyddion).
• To develop a convolutional neural network to characterise DNA and protein structures in topographic AFM images and define scanning areas around them (collaborating with IBEC)
• To contribute to integrating the neural networks developed into the post-processing software (collaborating with DATRIX)
Planned secondments
• Sorbonne University, M18 (1M) building indexed big databases.
• The Czech Metrology Institute, M24 (1M) Gwyddion programming.
• Bruker, T. Mueller, M36 (2M) multiparametric measurements.
Experimental Approach:
The project will use and develop our Python pipeline TopoStats, integrating machine learning approaches to quantitatively determine the mechanical state of individual DNA molecules.
You will be supervised by Alice. All supervisors are committed to embedding positive and inclusive research cultures in their groups. The supervisors will work together to ensure expectations on students and of supervisors are clearly defined and communicated. We welcome applicants from a diverse range of backgrounds across the physical and biological sciences and engineering with a background in programming.
We welcome applicants from a diverse range of backgrounds across the physical and biological sciences and engineering. Interested candidates are strongly encouraged to contact Alice to discuss your interest in and suitability for the project prior to submitting your application. Please refer to the SPM4.0 webpage for detailed information about the network and you can apply here.
Funding Notes
The award will fund a 3-year full-time employment contract with a gross salary of €XXX approx, depending also on the family status at the moment of recruitment, according to the MSCA-DN regulations, enrolment in a PhD programme at the University of Sheffield as well as a research grant to support costs associated with the project.