openfold3-nim

You can predict 3D biomolecular structures for proteins, DNA, RNA, and ligands using the OpenFold3 NIM microservice, which automates co-folding without manual HPC configuration. It provides confidence metrics like pLDDT and pTM scores to assess biological plausibility.

Yes, you can run structure prediction locally using Docker-based NIM containers for high-throughput workflows. The OpenFold3 skill supports flexible deployment, allowing you to switch between local Docker containers and hosted NVIDIA API endpoints.

To start predicting protein structures, you need valid NGC API credentials for the hosted API or a local Docker setup. You must also provide input sequences and MSA data adhering to specific schema constraints required by the OpenFold3 NIM microservice.

Yes, OpenFold3 supports multi-entity prediction, allowing you to co-fold proteins, nucleic acids like DNA and RNA, and small-molecule ligands together. This capability enables comprehensive structural biology workflows for complex biomolecular interactions.

When predicting biomolecular structures, the OpenFold3 NIM microservice provides scientific validation through confidence metrics like pLDDT and pTM scores. These metrics help you assess the biological plausibility and accuracy of the generated 3D structural models.

Hosted NVIDIA API endpoints for structure prediction require valid NGC API credentials and strict adherence to input schema constraints for sequence and MSA data. For high-throughput workflows, you might encounter limitations best handled by switching to local Docker-based NIM inference.