bionemo-agent-toolkit

Define custom groups (Irrep subclasses), build segmented tensor products with CG coefficients, create equivariant polynomials and IrDictPolynomials, and use built-in descriptors (linear, tensor products, spherical harmonics). Use when working with cuequivariance group theory, irreps, or segmented polynomials.

Use when accelerating existing genomics workflows with NVIDIA Parabricks, improving runtime or price/performance, converting pipeline steps to GPUs, or comparing CPU and GPU workflow outputs. Adds optional GPU steps in-place with runtime toggles (default off). Do NOT use for individual pbrun command routing — use parabricks.

Write code that calls the installed nvMolKit Python API for GPU-accelerated, batched RDKit-style operations - Morgan fingerprints, Tanimoto/cosine similarity, ETKDG conformer embedding, MMFF/UFF optimization, TFD, conformer RMSD, Butina clustering, and substructure search. Use when the user is importing `nvmolkit.*`, debugging an `nvmolkit` call, choosing between nvMolKit and RDKit for a batched cheminformatics workflow, or wiring nvMolKit results into a torch/numpy pipeline. Out of scope: building nvMolKit from source.

Route NVIDIA Parabricks pbrun tools, assess GPU/runtime readiness, and provide version-aware command guidance for FASTQ/BAM processing, RNA-seq, variant calling, BAM QC, and GVCF workflows. Do NOT use for inspecting or accelerating whole pipelines — use genomics-workflow-acceleration.

Use Boltz2 NIM for biomolecular structure prediction and binding affinity. Invoke for Boltz2, protein structures, protein-ligand/DNA/RNA complexes, SMILES or CCD ligands, pIC50/IC50 affinity scoring, mmCIF output, hosted NVIDIA API calls, or local Docker deployment.

Run DiffDock molecular docking via NVIDIA NIM to predict small-molecule binding poses against protein targets. Use for DiffDock, molecular docking, ligand docking, blind docking, SMILES or SDF ligands, ranked poses, confidence scores, hosted NVIDIA API, or local Docker deployment.

Generate and analyze DNA sequences using NVIDIA's Evo 2 BioNeMo NIM microservice. Use for Evo2/Evo 2, DNA generation, genomic sequence generation, hosted generation, local Docker deployment, local forward passes, layer outputs, logits, sampled probabilities, and BioNeMo NIM workflows.

Generate novel drug-like molecules using the GenMol NIM microservice. Use for de novo generation, scaffold decoration, motif extension, lead optimization, SAFE notation, QED or LogP ranking, hosted NVIDIA API calls, or local Docker deployment. GenMol takes SAFE notation in the smiles field, not ordinary SMILES.

Run a complete computational drug discovery pipeline using NVIDIA BioNeMo NIMs: generate drug-like molecules with GenMol, dock them to a protein target with DiffDock, then predict binding affinity with Boltz2. Use this skill whenever the user wants to generate and screen small molecule drug candidates, perform hit discovery, optimize leads against a protein target, or do virtual screening combining molecule generation, docking, and affinity prediction. Triggers on: drug discovery pipeline, hit discovery, lead optimization, virtual screening, molecule generation, molecular docking, binding affinity, GenMol, DiffDock, Boltz2, SMILES, SAFE notation, NIM microservice. This is a multi-step pipeline composing three BioNeMo NIMs.

Run a complete protein structure prediction pipeline using NVIDIA BioNeMo NIMs: search for MSA alignments with MSA-Search (ColabFold), then predict the structure with OpenFold3 using the retrieved alignments. Use this skill whenever the user wants to predict a protein structure with maximum accuracy using MSA context, run the full AlphaFold3-style pipeline, generate MSA-informed structure predictions, or improve structure prediction accuracy by providing evolutionary information. Triggers on: MSA structure prediction pipeline, structure prediction pipeline, MSA-informed prediction, OpenFold3, ColabFold MSA, AlphaFold3 pipeline, protein structure, homology search, a3m alignment, UniRef30, NIM microservice. This pipeline chains MSA-Search and OpenFold3.

Use this skill for MolMIM, NVIDIA's BioNeMo NIM microservice for small-molecule latent-space generation and optimization. Invoke for MolMIM, molecular embeddings, hidden states, latent decoding, sampling around a seed SMILES, CMA-ES guided molecule generation, QED or plogP optimization, hosted NVIDIA API calls, or local Docker deployment.

Generate multiple sequence alignments (MSAs) for protein sequences using the ColabFold MSA-Search NIM. Use for homolog search, UniRef30/ColabFold env searches, A3M or FASTA alignments, paired MSA search for complexes, PDB70 structural templates, hosted NVIDIA API calls, or local Docker deployment.

Use this skill for OpenFold2, NVIDIA's BioNeMo NIM microservice for monomer protein structure prediction. Invoke whenever the user mentions OpenFold2, AlphaFold2-like monomer folding, protein sequence-to-structure prediction, A3M MSAs, mmCIF templates, hosted NVIDIA API calls, or local Docker deployment.

Use this skill for OpenFold3, NVIDIA's BioNeMo NIM microservice for biomolecular structure prediction. Invoke whenever the user mentions OpenFold3 or needs protein, protein-ligand, protein-DNA/RNA, or multi-chain complex prediction with the hosted NVIDIA API or local Docker NIM. Covers endpoint choice, auth, request payloads, output artifacts, confidence scores, and local container setup.

Run ProteinMPNN inverse folding via NVIDIA NIM to design protein sequences for a target backbone. Use for ProteinMPNN, inverse folding, sequence design, backbone redesign, fixed chains/residues, omit_AAs, sampling temperature, soluble model, hosted NVIDIA API, local Docker, PDB input, and multi-FASTA output.

Run RFDiffusion protein backbone design via NVIDIA NIM. Use for de novo protein backbones, motif scaffolding, binder design, hotspot residues, contigs syntax, diffusion steps, hosted NVIDIA API calls, local Docker deployment, and PDB backbone outputs for ProteinMPNN sequence design.

Convert a grover_base checkpoint (encoder-only or encoder + vocab heads) into a hybrid checkpoint by adding a randomly-initialized cMIM decoder + latent_dist, then continue pretraining on the user's corpus as hybrid (vocab + contrast). Effectively kermt-continue-pretrain with a one-time ckpt-conversion step prepended.

Continue pretraining from an existing KERMT checkpoint. The skill validates the user's checkpoint and pretrain CSV, prepares the data into shard/vocab/features form, then launches pretrain_ddp.py inside the kermt container (detached for long runs). Auto-dispatches `--pretrain_mode` based on the checkpoint type (grover_base vocab-only, cmim, or hybrid).

Extract per-molecule embeddings from any encoder-bearing KERMT checkpoint (grover_base / cmim / hybrid / finetuned). Writes one .npy per readout type (atom_from_atom, bond_from_atom, atom_from_bond, bond_from_bond) plus canonical_smiles.npy and validity.npy. Calls task/extract_embeddings.py (which featurizes SMILES on the fly — no pre-computed features needed).

Finetune a pretrained KERMT encoder on a labeled CSV. The skill validates the input checkpoint (must be a pretrain ckpt — grover_base / cmim / hybrid), validates the labeled CSV, prepares the data (clean + features + optional split), then launches main.py finetune inside the kermt container (detached for hours-scale runs). Hyperparameters come from agent/config/defaults_finetune.json with per-flag CLI override.

Run predictions with a finetuned KERMT checkpoint on a SMILES-only CSV. The skill validates that the input ckpt has task FFN heads (refuses pretrain ckpts with a redirect to kermt-finetune), validates the CSV, prepares the data (clean + rdkit_2d features), then launches main.py predict inside the kermt container (blocking, minutes-scale).

Check progress for a detached KERMT run (pretrain, finetune, or any kermt_run_detached invocation). Reads run.json, queries docker for container state, tails the pretrain/finetune log, and parses progress lines (epoch, step, val loss).

Pretrain a fresh KERMT model from scratch on a user-provided corpus. Builds a new vocabulary from the corpus, instantiates the model architecture from defaults, and launches pretrain_ddp.py inside the kermt container (detached for long runs). Unlike kermt-continue-pretrain, no starting checkpoint is loaded — the model is randomly initialized.

Bootstrap the KERMT agent environment — verify host docker + nvidia-container-toolkit, build the kermt:latest image from the repo's Dockerfile if it doesn't yet exist, and run a GPU smoke test inside the container. Every other kermt-* skill depends on this; invoke it first.

End-to-end Proteina-Complexa design pipeline driver. Reach for this skill whenever the user wants to "design a binder", "design binders for X", "run complexa design", "de novo binder", "PDL1 binder", "TrkA binder", "design proteins for target", "protein binder design", "ligand binder", "design a small-molecule binder", "ATP-binding protein", "AME motif scaffolding", "scaffold a motif near a ligand", "motif + ligand design", "enzyme scaffolding", "flow matching protein design", "beam-search binder", "FK steering", "MCTS protein design", "refold with AF2", "refold with RF3", "refold with ESMFold", or wants success rates, interface pAE, scRMSD, or FoldSeek diversity from a single command. This is the scientific anchor of the skill set: it drives `complexa design <pipeline>` from target picking to manifest emission and tells the user how many designs passed.

Standalone evaluation of an existing PDB directory with Proteina-Complexa. Use this skill whenever the user wants to "evaluate PDB files", "re-fold these designs", "compute interface pAE", "compute i_pLDDT for a folder", "run AF2 / RF3 / ESMFold on my designs", "score binder candidates", "designability of this folder", "scRMSD for designs", "motif RMSD for these PDBs", "complexa analysis", "complexa evaluate from a PDB directory", "evaluate from pdb dir", or score third-party outputs (BindCraft, AlphaProteo, RFdiffusion, hand-curated decoys). It picks the correct `evaluate_*.yaml` config, wires `++dataset.pdb_dir` and the folding backend, runs `complexa analysis` (the evaluate → analyze chain), parses the result CSV, reports pass-rates against the right `result_type` thresholds, and emits a replayable `eval_manifest.json`. Reach for this skill before hand-rolling refolding scripts.

First-time setup, environment configuration, and model-weight installation for Proteina-Complexa. Reach for this skill whenever the user says "set up complexa", "install complexa", "configure my .env", "first-time setup", "what models do I have installed", "what's in my .env", "download model weights", "download Complexa / AF2 / RF3 / ProteinMPNN / LigandMPNN / ESM2 / ESMFold checkpoints", "preflight my GPU", "verify environment", "complexa init", "complexa download", "complexa download --status", "complexa validate env", or any time a fresh checkout needs to be made runnable. This is the first skill to run on a new clone — it drives `complexa init`, `complexa download`, and `complexa validate env` end-to-end, edits the required `.env` keys, picks the right runtime (UV vs Docker), and emits a replayable setup artifact.

Launch Proteina-Complexa pipelines on a remote SLURM cluster — binder search, LaProteina monomer design, or multi-node distributed training. Reach for this skill whenever the user says "launch on SLURM", "submit to the cluster", "submit binder search to SLURM", "kick off training on the cluster", "multi-node training", "cluster job", "sbatch", "remote GPU run", "complexa slurm", "launch_protein_binder_search.sh", "launch_laproteina_train.sh", "launch_laproteina_design_pipeline.sh", "launch on grizzly / polar", "--on-cluster", "run distributed training", "sweep on SLURM", "run all targets on the cluster", "kick off a multi-target binder search", "rsync to cluster", "submit a singleton requeue chain", or whenever a Hydra config / sweep needs to escape a single workstation. This skill drives the launcher scripts under `slurm_utils/`, **always previews with `--dry-run` first**, then submits, captures SLURM job IDs, and emits a replayable manifest. SLURM submission costs cluster time and is hard to reverse, so the

Use this skill whenever the user wants to run a parameter sweep over a Proteina-Complexa design pipeline — cartesian-product hyperparameter scans, Pareto search over generation/reward/evaluation knobs, or any "compare configurations" workflow. Trigger phrases include "sweep beam width", "sweep nsteps", "hyperparameter sweep", "parameter scan", "scan beam_width and temperature", "compare configurations", "find the best generation params", "what's the optimal nsteps", "Pareto search for binder quality vs wall-clock", "complexa sweep", "tune Complexa", "ablate the reward weights", "configs/sweeps", "--sweeper", "run beam_width.yaml". This is the only skill that owns sweeper YAML authoring, cartesian-product expansion, and per-config result ranking. For cluster submission mechanics see the `complexa-slurm` skill.

Use this skill whenever the user wants to add, register, edit, list, show, or validate a Proteina-Complexa design target for any pipeline — protein binder (default), ligand binder, or AME / enzyme scaffolding. Triggers include "add a target", "define a new target for binder design", "register a hotspot", "set up a PDL1 binder target", "ligand binder pocket", "SMILES target", "AME task", "enzyme motif", "M0024_1nzy", "M0096_1chm", "complexa target add", "complexa target show", "configure target X", "what targets are available", "where do hotspots live", "what does target_input mean", "chain-spec syntax", "binder length range", "contig_atoms", or any question about `configs/targets/{,ligand_}targets_dict.yaml` and `configs/design_tasks/ame_dict_v2.yaml`. Also covers `complexa validate target`. This is the only skill that touches the three targets dict files.

Use Boltz2 NIM for biomolecular structure prediction and binding affinity. Invoke for Boltz2, protein structures, protein-ligand/DNA/RNA complexes, SMILES or CCD ligands, pIC50/IC50 affinity scoring, mmCIF output, hosted NVIDIA API calls, or local Docker deployment.

Define custom groups (Irrep subclasses), build segmented tensor products with CG coefficients, create equivariant polynomials and IrDictPolynomials, and use built-in descriptors (linear, tensor products, spherical harmonics). Use when working with cuequivariance group theory, irreps, or segmented polynomials.

Run DiffDock molecular docking via NVIDIA NIM to predict small-molecule binding poses against protein targets. Use for DiffDock, molecular docking, ligand docking, blind docking, SMILES or SDF ligands, ranked poses, confidence scores, hosted NVIDIA API, or local Docker deployment.

Run a complete computational drug discovery pipeline using NVIDIA BioNeMo NIMs: generate drug-like molecules with GenMol, dock them to a protein target with DiffDock, then predict binding affinity with Boltz2. Use this skill whenever the user wants to generate and screen small molecule drug candidates, perform hit discovery, optimize leads against a protein target, or do virtual screening combining molecule generation, docking, and affinity prediction. Triggers on: drug discovery pipeline, hit discovery, lead optimization, virtual screening, molecule generation, molecular docking, binding affinity, GenMol, DiffDock, Boltz2, SMILES, SAFE notation, NIM microservice. This is a multi-step pipeline composing three BioNeMo NIMs.

Generate and analyze DNA sequences using NVIDIA's Evo 2 BioNeMo NIM microservice. Use for Evo2/Evo 2, DNA generation, genomic sequence generation, hosted generation, local Docker deployment, local forward passes, layer outputs, logits, sampled probabilities, and BioNeMo NIM workflows.

Generate novel drug-like molecules using the GenMol NIM microservice. Use for de novo generation, scaffold decoration, motif extension, lead optimization, SAFE notation, QED or LogP ranking, hosted NVIDIA API calls, or local Docker deployment. GenMol takes SAFE notation in the smiles field, not ordinary SMILES.

Use when accelerating existing genomics workflows with NVIDIA Parabricks, improving runtime or price/performance, converting pipeline steps to GPUs, or comparing CPU and GPU workflow outputs. Adds optional GPU steps in-place with runtime toggles (default off). Do NOT use for individual pbrun command routing — use parabricks.

Use this skill for MolMIM, NVIDIA's BioNeMo NIM microservice for small-molecule latent-space generation and optimization. Invoke for MolMIM, molecular embeddings, hidden states, latent decoding, sampling around a seed SMILES, CMA-ES guided molecule generation, QED or plogP optimization, hosted NVIDIA API calls, or local Docker deployment.

Generate multiple sequence alignments (MSAs) for protein sequences using the ColabFold MSA-Search NIM. Use for homolog search, UniRef30/ColabFold env searches, A3M or FASTA alignments, paired MSA search for complexes, PDB70 structural templates, hosted NVIDIA API calls, or local Docker deployment.

Run a complete protein structure prediction pipeline using NVIDIA BioNeMo NIMs: search for MSA alignments with MSA-Search (ColabFold), then predict the structure with OpenFold3 using the retrieved alignments. Use this skill whenever the user wants to predict a protein structure with maximum accuracy using MSA context, run the full AlphaFold3-style pipeline, generate MSA-informed structure predictions, or improve structure prediction accuracy by providing evolutionary information. Triggers on: MSA structure prediction pipeline, structure prediction pipeline, MSA-informed prediction, OpenFold3, ColabFold MSA, AlphaFold3 pipeline, protein structure, homology search, a3m alignment, UniRef30, NIM microservice. This pipeline chains MSA-Search and OpenFold3.