#1 · Structural bio & protein design
MD simulations link conformational states to membrane binding and organization
The common thread is molecular dynamics used to connect protein conformational states and intermolecular contacts to membrane engagement. For the DspS histidine kinase, full-length dimer simulations combining AlphaFold-Multimer, accelerated MD, and Markov state modeling map CDA-dependent transmembrane helix dynamics. All-atom MD (57.0 μs total) of hBD-3 monomers and dimers, wild-type and disulfide-broken analogs, in model membranes is analyzed by contact-matrix PCA/LDA. For the SARS-CoV-2 M protein, MD and continuum modeling indicate that self-interactions alone drive membrane clustering, with direct oligomerization dominating. Septin Cdc12 amphipathic helix pairs adopt a bent membrane-bound geometry with antiparallel, salt-bridged cooperative binding. What is new is the explicit linkage of trajectory-derived states to membrane organization and binding cooperativity across these systems.
Published papers
FragDockRL: A Reinforcement Learning Method for Fragment-Based Ligand Design via Building-Block Assembly and Tethered Docking.
FragDock: building-block virtual synthesis with tethered docking; FragDockRL, a DQN-based RL search, benchmarked against random, beam, and MCTS search on…
Preprints
D2ML: Models for Direct-to-Biology Small Molecule Screening
Assay-aware ML models for direct-to-biology small molecule screening predict activity and selectivity from crude and purified data, guiding ROS1 selectivity…
Paired activity learning supports systematic identification of mutant-selective EGFR inhibitors
Paired-activity regression plus SVM screening of 12 million molecules finds mutant-selective EGFR triple-mutant inhibitors; two crystal structures reveal…
When Does Active Learning Improve ADMET Prediction? A Multi-Endpoint Benchmark
Protocol-matched benchmark of seven active-learning acquisition strategies and two uncertainty methods across eight ADMET regression endpoints, using a strict…
Task-Dependent Performance of GNNs and Pretrained Models in ADMET Prediction
Matched-budget, scaffold-split benchmark of GNNs, frozen pretrained models (ChemBERTa, MolCLR) and fingerprints on six TDC ADMET endpoints; no family…
Bioactivity foundation model for hit identification and hit-to-lead optimization from sequence data
PSICHIC+, a bioactivity foundation model trained on ~82M protein-ligand records, predicts binding affinity distributions from sequence and SMILES for virtual…
Topic history
| Date | Rank |
|---|---|
| #1 | |
| #1 | |
| #1 | |
| #4 | |
| #4 | |
| #5 | |
| #5 | |
| #4 | |
| #5 | |
| #4 | |
| #16 | |
| #19 | |
| #7 | |
| #9 | |
| #2 | |
| #14 | |
| #15 | |
| #18 |