Contact

Additional Info

  • Mail Code: 9025

All Publications

  • Unlocking hidden biomolecular conformational landscapes in diffusion models at inference time. ArXiv Richman, D. D., Karaguesian, J., Suomivuori, C. M., Dror, R. O. 2025

    Abstract

    The function of biomolecules such as proteins depends on their ability to interconvert between a wide range of structures or "conformations." Researchers have endeavored for decades to develop computational methods to predict the distribution of conformations, which is far harder to determine experimentally than a static folded structure. We present ConforMix, an inference-time algorithm that enhances sampling of conformational distributions using a combination of classifier guidance, filtering, and free energy estimation. Our approach upgrades diffusion models-whether trained for static structure prediction or conformational generation-to enable more efficient discovery of conformational variability without requiring prior knowledge of major degrees of freedom. ConforMix is orthogonal to improvements in model pretraining and would benefit even a hypothetical model that perfectly reproduced the Boltzmann distribution. Remarkably, when applied to a diffusion model trained for static structure prediction, ConforMix captures structural changes including domain motion, cryptic pocket flexibility, and transporter cycling, while avoiding unphysical states. Case studies of biologically critical proteins demonstrate the scalability, accuracy, and utility of this method.

    View details for DOI 10.1038/s41586-024-07487-w

    View details for PubMedID 41376657

    View details for PubMedCentralID PMC12687860

  • Unlocking hidden biomolecular conformational landscapes in diffusion models at inference time. ArXiv Richman, D. D., Karaguesian, J., Suomivuori, C., Dror, R. O. 2025

    Abstract

    The function of biomolecules such as proteins depends on their ability to interconvert between a wide range of structures or "conformations." Researchers have endeavored for decades to develop computational methods to predict the distribution of conformations, which is far harder to determine experimentally than a static folded structure. We present ConforMix, an inference-time algorithm that enhances sampling of conformational distributions using a combination of classifier guidance, filtering, and free energy estimation. Our approach upgrades diffusion models-whether trained for static structure prediction or conformational generation-to enable more efficient discovery of conformational variability without requiring prior knowledge of major degrees of freedom. ConforMix is orthogonal to improvements in model pretraining and would benefit even a hypothetical model that perfectly reproduced the Boltzmann distribution. Remarkably, when applied to a diffusion model trained for static structure prediction, ConforMix captures structural changes including domain motion, cryptic pocket flexibility, and transporter cycling, while avoiding unphysical states. Case studies of biologically critical proteins demonstrate the scalability, accuracy, and utility of this method.

    View details for PubMedID 41376657

  • Combinatorial effector targeting (COMET) for transcriptional modulation and locus-specific biochemistry. bioRxiv : the preprint server for biology Wilson, C. M., Pommier, G. C., Richman, D. D., Sambold, N., Hussmann, J. A., Weissman, J. S., Gilbert, L. A. 2024

    Abstract

    Understanding how human gene expression is coordinately regulated by functional units of proteins across the genome remains a major biological goal. Here, we present COMET, a high-throughput screening platform for combinatorial effector targeting for the identification of transcriptional modulators. We generate libraries of combinatorial dCas9-based fusion proteins, containing two to six effector domains, allowing us to systematically investigate more than 110,000 combinations of effector proteins at endogenous human loci for their influence on transcription. Importantly, we keep full proteins or domains intact, maintaining catalytic cores and surfaces for protein-protein interactions. We observe more than 5800 significant hits that modulate transcription, we demonstrate cell type specific transcriptional modulation, and we further investigate epistatic relationships between our effector combinations. We validate unexpected combinations as synergistic or buffering, emphasizing COMET as both a method for transcriptional effector discovery, and as a functional genomics tool for identifying novel domain interactions and directing locus-specific biochemistry.

    View details for DOI 10.1101/2024.10.28.620517

    View details for PubMedID 39554033

    View details for PubMedCentralID PMC11565746