TMol brings Rosetta-style all-atom molecular energetics into differentiable PyTorch

gottapatchemall · x · 2026-11-19

Following up on the RFD4-Proteina thread, DidiKieran released the TMol paper on arXiv. TMol implements Rosetta-style all-atom energies with GPU batching and gradients in PyTorch, letting you put molecular energetics directly into your model — combining deep learning with physics.

It covers proteins, L/D residues, β/γ backbones, small molecules, modified DNA/RNA, PTMs, N-methyl peptides, cyclic peptides, and metals/cofactors, organized via PoseStack, and exposes Score, Minimize, Repack, FastRelax, and differentiable operations for protein design and molecular simulation.

Original post →

More from Research

Research channel →