Agent Skillssynthetic-sciences/openscience › smiles-validation

smiles-validation

GitHub

提供严格的SMILES分子结构验证、结构对比及修改确认功能,用于检测LLM生成的无效分子或结构不一致问题,支持单条校验、批量处理及变更分类。

backend/cli/skills/chemistry/smiles-validation/SKILL.md synthetic-sciences/openscience

触发场景

生成SMILES后需进行有效性校验 需要验证分子结构修改是否符合描述 批量检查生成的分子库质量

安装

npx skills add synthetic-sciences/openscience --skill smiles-validation -g -y
更多选项

非标准路径

npx skills add https://github.com/synthetic-sciences/openscience/tree/main/backend/cli/skills/chemistry/smiles-validation -g -y

不安装直接使用

npx skills use synthetic-sciences/openscience@smiles-validation

指定 Agent (Claude Code)

npx skills add synthetic-sciences/openscience --skill smiles-validation -a claude-code -g -y

安装 repo 全部 skill

npx skills add synthetic-sciences/openscience --all -g -y

预览 repo 内 skill

npx skills add synthetic-sciences/openscience --list

SKILL.md

Frontmatter
{
    "name": "smiles-validation",
    "tags": [
        "cheminformatics",
        "validation",
        "SMILES",
        "quality-control"
    ],
    "license": "MIT",
    "version": "1.0.0",
    "category": "chemistry",
    "metadata": {
        "skill-author": "Synthetic Sciences"
    },
    "description": "Strict SMILES validation, structural comparison, and modification verification. Catches invalid LLM-generated molecules.",
    "dependencies": [
        "rdkit-pypi"
    ]
}

SMILES Validation

Overview

LLMs frequently generate invalid SMILES or produce molecules that don't match their stated reasoning. This skill provides strict validation, structural comparison, and modification verification.

Key checks:

  • Parse validation: RDKit sanitization, valence checking, parenthesis/bracket balance
  • Structural comparison: Tanimoto similarity, scaffold preservation, MCS analysis
  • Modification verification: Confirm claimed structural changes exist in the actual molecule
  • Classification: Categorize changes as optimization (>0.6 similarity), significant modification (0.4-0.6), or de novo design (<0.4)

When to Use This Skill

  • After molecule generation: Validate every SMILES before reporting results
  • Optimization verification: Confirm proposed modifications match the actual structure
  • Batch validation: Check a library of generated molecules for validity
  • Quality control: Ensure reproducibility of molecular designs

Installation

pip install rdkit-pypi

Core Workflows

1. Validate a Single SMILES

python scripts/validate.py --smiles "c1ccccc1"

2. Compare Original vs Modified

python scripts/validate.py --original "c1ccccc1" --proposed "c1ccc(O)cc1" --check-modification "Added hydroxyl group"

3. Batch Validation

python scripts/validate.py --input generated_molecules.csv --output validation_report.json

Script Reference

Script Purpose Key Outputs
validate.py SMILES validation, comparison, and modification checking JSON report with validity, similarity, scaffold match, modification verification

版本历史

  • e9844a4 当前 2026-07-11 17:22

依赖关系

  • required rdkit-pypi

同 Skill 集合

.openscience/skill/bun-file-io/SKILL.md
backend/cli/skills/biology/anndata/SKILL.md
backend/cli/skills/biology/benchling-integration/SKILL.md
backend/cli/skills/biology/bioimage-analysis/SKILL.md
backend/cli/skills/biology/bioservices/SKILL.md
backend/cli/skills/biology/cancer-genomics-analysis/SKILL.md
backend/cli/skills/biology/clinical-imaging/SKILL.md
backend/cli/skills/biology/clinical-reports/SKILL.md
backend/cli/skills/biology/cobrapy/SKILL.md
backend/cli/skills/biology/curated-bio-datasets/SKILL.md
backend/cli/skills/biology/deeptools/SKILL.md
backend/cli/skills/biology/dnanexus-integration/SKILL.md
backend/cli/skills/biology/etetoolkit/SKILL.md
backend/cli/skills/biology/flow-cytometry-analysis/SKILL.md
backend/cli/skills/biology/flowio/SKILL.md
backend/cli/skills/biology/gget/SKILL.md
backend/cli/skills/biology/glycobiology/SKILL.md
backend/cli/skills/biology/histolab/SKILL.md
backend/cli/skills/biology/immunology-assays/SKILL.md
backend/cli/skills/biology/latchbio-integration/SKILL.md
backend/cli/skills/biology/microbial-dynamics/SKILL.md
backend/cli/skills/biology/molecular-cloning/SKILL.md
backend/cli/skills/biology/neurokit2/SKILL.md
backend/cli/skills/biology/neuropixels-analysis/SKILL.md
backend/cli/skills/biology/omero-integration/SKILL.md
backend/cli/skills/biology/opentrons-integration/SKILL.md
backend/cli/skills/biology/pathml/SKILL.md
backend/cli/skills/biology/pharmacology-wetlab/SKILL.md
backend/cli/skills/biology/protocolsio-integration/SKILL.md
backend/cli/skills/biology/pydeseq2/SKILL.md
backend/cli/skills/biology/pyhealth/SKILL.md
backend/cli/skills/biology/pylabrobot/SKILL.md
backend/cli/skills/biology/pysam/SKILL.md
backend/cli/skills/biology/scanpy/SKILL.md
backend/cli/skills/biology/scikit-bio/SKILL.md
backend/cli/skills/biology/scikit-survival/SKILL.md
backend/cli/skills/biology/scvi-tools/SKILL.md
backend/cli/skills/biology/synthetic-biology/SKILL.md
backend/cli/skills/biology/treatment-plans/SKILL.md
backend/cli/skills/chemistry/admet-prediction/SKILL.md
backend/cli/skills/chemistry/admet-reasoning/SKILL.md
backend/cli/skills/chemistry/binding-affinity/SKILL.md
backend/cli/skills/chemistry/datamol/SKILL.md
backend/cli/skills/chemistry/deepchem/SKILL.md
backend/cli/skills/chemistry/denovo-design/SKILL.md
backend/cli/skills/chemistry/diffdock/SKILL.md
backend/cli/skills/chemistry/drug-design/SKILL.md
backend/cli/skills/chemistry/hypogenic/SKILL.md
backend/cli/skills/chemistry/matchms/SKILL.md
backend/cli/skills/chemistry/medchem/SKILL.md

元信息

文件数
0
版本
1b7978c
Hash
351b7c62
收录时间
2026-07-11 17:22

首页 - Wiki
Copyright © 2011-2026 iteam. Current version is 2.155.2. UTC+08:00, 2026-08-04 21:47
浙ICP备14020137号-1 $访客地图$