L. Heo

One particle per residue is sufficient to describe all-atom protein structures

This paper shows that single beads per amino acid residue can accurately reconstruct all-atom protein structures using machine learning, with minimal information loss.

Modeling concentration-dependent phase separation processes involving peptides and RNA via residue-based coarse-graining

The paper introduces a residue-based coarse-grained model, COCOMO, to simulate concentration-dependent phase separation of peptides and RNA, showing it predicts experimental data and is computationally efficient.

Characterizing Transient Protein-Protein Interactions by Trp-Cys Quenching and Computer Simulations

This paper introduces Trp-Cys quenching experiments combined with simulations to study transient protein interactions under crowded conditions, showing they can distinguish interaction types and preferences.

Direct Generation of Protein Conformational Ensembles via Machine Learning

This paper shows that machine learning can generate protein conformational ensembles without simulation, using a transformer-based GAN trained on simulation data.

Multi-state Modeling of G-protein Coupled Receptors at Experimental Accuracy

This paper introduces a multi-state prediction protocol for GPCRs using AlphaFold2, achieving high accuracy in predicting both active and inactive states and outperforming existing methods in ligand docking and competition benchmarks.

Inhibitory proteins block substrate access by occupying the active site cleft of Bacillus subtilis intramembrane metalloprotease SpoIVFB

Modeling of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Proteins by Machine Learning and Physics-Based Refinement

The paper predicts and refines SARS-CoV-2 protein structures using a machine learning pipeline and molecular dynamics refinement, comparing results with existing models.

Direct therapeutic targeting of SWI/SNF induces epigenetic reprogramming and durable tumor regression in rhabdoid tumor

This paper shows that targeting the SWI/SNF complex with mithramycin and its analogue EC8042 induces epigenetic reprogramming and durable tumor regression in rhabdoid tumors.

High-Accuracy Protein Structures By Combining Machine-Learning With Physics-Based Refinement

This paper combines machine-learning models with physics-based refinement to improve protein structure prediction, achieving higher accuracy than existing methods.