Alex Torbey

PhD

Scientist

All Crystallography software Drug screening Drug Simulations Drug-Target Interaction Studies Fragment-based drug design G protein receptor activity profiling Large language model implementation membrane protein design Molecular Dynamics Simulations Molecular Modeling Protein Design Protein Engineering Protein/Drug Modeling Software Engineering Structure-Based Drug Design

Preprint(s) 1st author preprint(s) Peer-reviewed publication(s) 1st author peer-reviewed publication(s) Conference Poster(s) Patent(s) Reviewed Publication(s)

1) VecPRS (Recursion Pharmaceuticals): Built VecPRS, a physics-based method that analyzes protein conformational dynamics from AI-generated structures and molecular dynamics simulations to predict whether an unknown ligand will functionally act as an activator, inhibitor, or intermediate signal modulator, including allosteric effects. The method achieves ~68% accuracy on unseen binders and is covered by a provisional patent filed by Recursion

2) End-to-end computational drug-discovery pipelines (independent research): Designed and implemented large-scale pipelines combining molecular simulations, statistical inference, and ML-based scoring to featurize protein–ligand interactions and visualize them in a cleaner novel representation. This work is currently under peer review and will be published.

3) Scientific data engines for AI: Built data representations and processing pipelines that convert heterogeneous structural biology, chemistry, and simulation outputs into model-ready datasets suitable for learning, optimization, and iterative design loops used in Valence Labs to increase inductive-bias dataset size and diversity. I also developed custom statistical tooling to transform time-correlated data into reproducible, decision-ready metrics, including uncertainty-aware feature extraction, effective sample size control, and state-specific mechanistic deconvolution