NVIDIA BioNeMo Inference Runtime (BioIR): High-Throughput Biomolecular Structure Prediction
Biomolecular structure prediction has shifted from single-target runs to proteome-scale worklists. In 2026, the bottleneck is no longer whether a model can fold a protein, but how fast an entire queue of independent targets moves through parsing, featurization, GPU inference, and output writing.
NVIDIA's new technical deep dive walks through BioNeMo Inference Runtime (BioIR), a Python library that accelerates supported structure-prediction models on NVIDIA GPUs while preserving the standard PyTorch workflow. BioIR has already run at production scale: it powered the recent expansion of the AlphaFold Database, generating protein-complex structures across 4,777 proteomes, roughly 31 million candidate complexes, with 1.81 million released as high-confidence predictions.
Deployable Now
BioIR is available as an open GitHub repository with a wheel containing precompiled CUBINs. Runtime use requires Python 3.12+, a compatible NVIDIA GPU and driver, a staged model checkpoint, and per-chain A3M MSAs. It does not require nvcc, CUDA source, CMake, or the CUDA toolkit.
What Is BioIR
BioIR targets operations that general-purpose inference stacks do not fully optimize, including Pairformer and Evoformer stacks, triangle operations, pairwise attention, diffusion transformers, and atom-level modules. Models remain ordinary torch.nn.Module objects. There is no engine build, export step, or separate artifact between a checkpoint and a forward pass.
How to Use BioIR
There are two usage paths:
- End-to-end processor โ moves an
InputRequestthrough parsing, tokenization, feature generation, GPU inference, and PDB or mmCIF writing. - Direct PyTorch integration โ lets developers construct a supported model or reuse selected optimized modules inside custom code.
The tutorial demonstrates the processor path with Boltz-2 (model_source="boltz-2"). Each protein chain requires an A3M MSA. Paired or unpaired MSAs are accepted for inputs with multiple non-identical protein chains. Templates can be supplied manually because BioIR does not run HHsearch.
via MarkTechPost