Research group Structural bioinformatics

About

We focus on bridging data-driven artificial intelligence (AI) and physics-based 3D modelling to better understand the molecular basis of health and diseases and to rationalize drug design.

AI on protein structures

Our recent focus is TCRpMHC (T-cell receptor and peptide-major histocompatibility complex) protein complexes, which play a key role in various cancer immunotherapies, such as cancer vaccine and TCR therapies.

Despite the promising clinical successes witnessed in cancer immunotherapies for some advanced-stage patients, the translation of these advancements into a universally applicable therapy has encountered challenges. These treatments are effective only on a small subset of patients and are very expensive (sipuleucel-T is priced at $93,000 for three shots). To address this issue, our efforts are concentrated on two key fronts:

1. Leveraging 3D Models for Improved AI Predictions:
   By harnessing 3D models of protein structures, we aim to enhance the accuracy of artificial intelligence in predicting cancer vaccine candidates. This approach integrates structural insights, allowing AI algorithms to discern potential targets more precisely.
2. AI-driven Enhancement of 3D Modeling of protein complexes:
   Our second focus involves employing generative AI (like in DALL-E) to model the 3D modeling of pMHC and TCRpMHC complexes. This initiative aims to provide a comprehensive 3D interpretation of MHC restriction and TCR specificity, elucidating the TCR's ability to selectively recognize and respond to specific pMHC complexes.

The confluence of breakthroughs in AI, exemplified by innovations like AlphaFold, along with strides in image generation, positions us at an opportune moment. We aim to leverage these advances in structural biology and AI to pioneer rational cancer vaccine design and enhance TCR engineering. Our commitment is to capitalize on these advancements to drive progress in the field and contribute to the development of more effective and broadly applicable cancer immunotherapies.

3D protein modelling & health

3D-Protein modelling for health & disease
We are involved in several collaborative projects that evolve around the use and interpretation of 3D protein structures. We collaborate with many other departments to study the molecular effects of genetic diversity. When necessary, we can build, visualize and interpret homology models as well. Information obtained from these structures and models can be used for, for example, variant analysis, drug docking, and intelligent experimental design.

3D-boosted detection of exon skipping targets
This line of research focusses on the improvement of exon-skip target detection by using 3D-protein structures. We use a combination of 3D-predictions and AI to learn which parts of proteins can be skipped and are therefore potential targets for antisense nucleotide therapy. Current predictions are focussed on genes causing Usher-syndrome.