New DDSA Fellow will improve the predictions of protein variants using machine learning methods

Protein engineering has a wide range of applications from biotechnology to drug discovery. In essence, it is a search problem, where the goal is to find mutations, or variants, that improve a particular trait of a protein, for instance its stability or function. In his project 'Principled Bayesian Optimization and scientific ML for Protein Engineering', Richard will investigate and develop new methods to improve the predictions of protein variant candidates with respect to a desired function.

The space of possible proteins is vast, and experimentally probing all relevant candidates is intractable. The design of proteins with the intended properties entails a vast discrete search-space and while there are computational representations and experimental observations available, there is a lack of methods to adequately combine all available information. Richard will use state-of-the-art probabilistic optimization and propose a novel machine learning method – Principled Bayesian Optimization on latent representations - with the aim of finding better candidates using fewer experimental resources.

Joins DIKU’s Machine Learning section

The project is a collaboration between the Machine Learning in Biology group at the Department of Computer Science, University of Copenhagen (DIKU), and the Department of Chemistry under joint supervision at the University of Copenhagen. Richard will therefore be an affiliate of the Machine Learning Section at DIKU and will be supervised by Associate Professor Wouter Boomsma.

– I am very happy that we managed to get one of the first of these fellowships to DIKU and look forward to working with Richard on his project, says Wouter Boomsma, supervisor and Associate Professor in the Machine Learning section at DIKU.

One of the first to receive DDSA fellowship

The Danish Data Science Academy is a new national initiative that supports data science activities in Denmark, including travel grants, funding for inviting academic visitors and organizing scientific meetings. They also provide funding for postdocs and 10 annual Ph.D. fellowships. Richard is one of the first ten PhD students to receive the fellowship.

– I am thrilled about the opportunity that the DDSA provides and look greatly forward to research at the intersection of Machine Learning, probabilistic optimization, and its integration into experimental design together with our collaborators, concludes PhD Fellow Richard Michael. 

Richard holds a bachelor’s degree in Cognitive Science from the University of Osnabrück in Germany and a master’s degree in Bioinformatics in Computer Science from the University of Copenhagen.

Learn more about his project in the video below.