A high-performance, GPU-accelerated library for key computational chemistry tasks, such as molecular similarity, conformer generation, and geometry relaxation. - NVIDIA-Digital-Bio/nvMolKit

A high-performance, GPU-accelerated library for key computational chemistry tasks, such as molecular similarity, conformer generation, and geometry relaxation. - NVIDIA-Digital-Bio/nvMolKit

NVIDIA (NVDA) expanded its BioNeMo open development platform at the J.P. Morgan Healthcare Conference to accelerate AI-driven biology and drug discovery. Key additions include RNAPro for RNA structure prediction, ReaSyn v2 for synthetically feasible designs, BioNeMo Recipes, and nvMolKit GPU-accelerated cheminformatics libraries. NVIDIA announced collaborations with Lilly to create a co-innovation AI lab $1 billion investment over five years) and with Thermo Fisher to build autonomous, edge-to-cloud lab infrastructure and agentic workflows. Multiple biotech partners are adopting BioNeMo for model training, deployment and lab-in-the-loop workflows.

Most of readers know that power of GPU change the way of cheminformatics. I introduced nvMolkit which is developed by NVIDIA for cheminformatics tasks. Now we can handle huge amount of data in short time 🙂 nvMolKit can calculate compound similarity and conformation rapidly with GPU but GPU assisted alignment is not implemented. Ligand based approach is common way of drug discovery evenif alpha fold3 is availabe in these days.

Unfortunately I could not participate RDKit UGM 2025 in this year.... I would like to join the meeting at next year By the way, recently we can use GPUs for acceralate chem/bio informatics calculations such as Deep Learning application or clustering tasks. Nvidia's rapids is one of the famous package for GPU based data science. But there are no package which accelarate RDKit function directry.

A guest post from NVIDIA

SyntheticGestaltがNVIDIAのnvMolKitを用いた4D分子基盤モデルの学習を発表し、AIエージェントの開発計画も明らかに。新たな時代のAI技術を探る。

nvMolKit