# Deepchem Circular Featurization

> Use this skill to compute compact DeepChem circular fingerprints from a small set of SMILES strings with the repo-managed DeepChem prefix. Do not use it for model training, docking, or large library screening.

- **Type:** Skill
- **Install:** `agentstack add skill-ma-compbio-lab-skillfoundry-deepchem-circular-featurization`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [ma-compbio-lab](https://agentstack.voostack.com/s/ma-compbio-lab)
- **Installs:** 0
- **Category:** [AI & ML](https://agentstack.voostack.com/c/ai-and-ml)
- **Latest version:** 0.1.0
- **License:** Apache-2.0
- **Upstream author:** [ma-compbio-lab](https://github.com/ma-compbio-lab)
- **Source:** https://github.com/ma-compbio-lab/SkillFoundry/tree/main/skills/drug-discovery-and-cheminformatics/deepchem-circular-featurization
- **Website:** https://ma-compbio-lab.github.io/SkillFoundry/

## Install

```sh
agentstack add skill-ma-compbio-lab-skillfoundry-deepchem-circular-featurization
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

## Purpose
Turn one or more SMILES strings into deterministic DeepChem `CircularFingerprint` summaries without requiring TensorFlow or PyTorch.

## When to use
- You need a lightweight DeepChem-backed fingerprinting step before downstream molecular ML work.
- You want a compact JSON payload with canonical SMILES, dense bit vectors, and active bit indices.

## When not to use
- You need graph featurizers, model training, or dataset download workflows.
- You need batch-scale featurization for very large libraries.

## Inputs
- Repeated `--smiles` arguments, or no arguments to use the bundled aspirin/caffeine example
- Optional `--size`, `--radius`, and `--out`

## Outputs
- JSON summary with `canonical_smiles`, `bit_vector`, `on_bits`, and `on_bit_count` for each molecule

## Requirements
- `slurm/envs/deepchem`
- DeepChem 2.8.0 and RDKit installed in that prefix

## Procedure
1. Run `slurm/envs/deepchem/bin/python skills/drug-discovery-and-cheminformatics/deepchem-circular-featurization/scripts/compute_circular_fingerprints.py --out skills/drug-discovery-and-cheminformatics/deepchem-circular-featurization/assets/aspirin_caffeine_fingerprints.json`.
2. Inspect `size`, `radius`, and each molecule's `canonical_smiles`, `bit_vector`, and `on_bits`.
3. Reuse the compact JSON as a deterministic preprocessing artifact for later experiments.

## Validation
- The command exits successfully under `slurm/envs/deepchem/bin/python`.
- Each molecule gets a non-empty canonical SMILES and a bit vector of the requested length.
- Repeated runs with the same inputs produce the same fingerprint payload.

## Failure modes and fixes
- Missing DeepChem runtime: run the script with `slurm/envs/deepchem/bin/python`.
- Invalid SMILES: correct the input string before featurization.
- Optional backend warnings: TensorFlow and PyTorch are not required for this fingerprint-only skill.

## Safety and limits
- Local featurization only.
- No activity prediction, medicinal-chemistry recommendation, or safety interpretation is implied.

## Provenance
- DeepChem documentation: https://deepchem.readthedocs.io/en/latest/
- RDKit documentation: https://www.rdkit.org/docs/index.html

## Related skills
- `rdkit-molecular-descriptors`

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [ma-compbio-lab](https://github.com/ma-compbio-lab)
- **Source:** [ma-compbio-lab/SkillFoundry](https://github.com/ma-compbio-lab/SkillFoundry)
- **License:** Apache-2.0
- **Homepage:** https://ma-compbio-lab.github.io/SkillFoundry/

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-ma-compbio-lab-skillfoundry-deepchem-circular-featurization
- Seller: https://agentstack.voostack.com/s/ma-compbio-lab
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.