Predict biomolecular structures from proteins, DNA, RNA, and ligands on GPU.
npx skills add https://github.com/yjcyxky/biominer-skills --skill openfold3
Please help me install this Agent Skill. Skill: openfold3 Source: https://github.com/yjcyxky/biominer-skills/tree/main/openfold3 Command: npx skills add https://github.com/yjcyxky/biominer-skills --skill openfold3
OpenFold3 provides a GPU-accelerated, open-source path to predict biomolecular structures, including proteins, DNA, RNA, and ligands, enabling researchers to generate high-confidence models from sequence data quickly.
Use the OpenFold3 Docker image to run help and basic prediction commands. For example:
docker run --rm --gpus all --ipc=host --shm-size=64g
ghcr.io/open-prophetdb/openfold3:arm64-blackwell
run_openfold --help
docker run --rm --gpus all --ipc=host --shm-size=64g
ghcr.io/open-prophetdb/openfold3:arm64-blackwell
run_openfold predict --help
High-intent search queries and answers about installing and using this skill.
To predict biomolecular complex structures, you can deploy the OpenFold3 Docker image with GPU access. It requires passing `--gpus all` and configuring shared memory using `--ipc=host --shm-size=64g` to execute the `run_openfold predict` command.
You can predict proteins, DNA, RNA, and ligands, including complex biomolecular assemblies. The tool supports multi-format inputs to generate high-confidence structural models from sequence data.
Yes, OpenFold3 targets NVIDIA DGX Spark (Grace Blackwell) on ARM64. It provides a specific Docker image with pre-compiled Triton kernels for fast startup and GPU-accelerated structure prediction.
Yes, running OpenFold3 in Docker requires configuring shared memory with `--ipc=host --shm-size=64g`. This configuration is necessary for the Docker container to operate correctly during GPU-accelerated predictions.
You can run batch predictions for protein complexes to compare binding interfaces using the provided Docker image. This allows generating multiple models and complex assemblies within a single workflow.
An open-source AlphaFold3 implementation provides a GPU-accelerated path to predict biomolecular structures quickly. It enables research teams to generate high-confidence models from sequence data without proprietary constraints.
NVIDIA-BioNeMoPredict 3D biomolecular structures for proteins, DNA, RNA, and ligands via OpenFold3 NIM.
felixboehmRetrieve AlphaFold-predicted 3D protein structures by UniProt ID.
plurigridGenerate deterministic colors across CPU and GPU with SplitMix64 PRNG.
thc1006Deploy Kubernetes clusters on DGX Spark ARM64 with Kubespray.
