| Details: |
Most of computational biology is predicated upon the sequence → structure → function →
phenotype paradigm. Thanks to artificial intelligence and the availability of data at various scales,
researchers have been trying to bridge gaps between the different tiers of this process, starting from the
age-old genotype–phenotype modeling to CASP and Alphafold’s sequence-structure up to recent attempts
to go from sequence to ensemble. However, physical causality is often missing in the traditional
bioinformatic models, thus far sidelining the AI-driven advances only to predictions of the forward
direction.
The lecture will introduce physical ideas to conceive generative models that backmap phenotypes down
to an ensemble of structures and sequences. For example, leveraging our work on modeling the diffusion
of charge carriers in bioenergetic membranes (Cell 2019, JACS 2024, Nature Metabolism 2024), we
computed the mechanism of chemokine binding to the Oxford Covid Vaccine (Science Adv 2021).
With AstraZenaca, we computationally redesigned the adenovirus vector to prevent potential clotting
disorders. This design strategy gave rise to the machine learning of electrostatic surfaces using Google's
inception network algorithm. We are now using this network to study disease association in patients (Cell
Systems 2024), as well as design peptide therapeutics. |