科研插件

AI药物发现

靶点、QSAR、ADMET、筛选与分子生成。

官方收录发布者 Codar安全审核通过v0.3.025 个 Skill

README

AI药物发现:Skill 导航与方法比较

本 README 按任务而不是按来源组织 Skill。表中的入口完成相同或相近任务, 但算法假设、输入格式、计算资源和输出不同;它们是选择关系,不是待删除副本。 除非表格明确说明,描述不构成跨数据集 benchmark,实际选择仍应结合数据与版本验证。

使用方法

  1. 先定位任务族,再按“重点比较”判断数据、目标和资源约束。
  2. 打开候选 SKILL.md 核对命令、输入输出和陷阱;不要只凭工具名选择。
  3. 对模型类方法优先保留独立验证集或第二模型共识;对统计方法先核对假设。

分子对接

共同目标:围绕“分子对接”完成同一分析环节;以下入口保留不同算法、工具或工作流。

重点比较:比较盲对接/口袋对接、搜索空间、柔性处理和 pose 置信度;几何排名不等同于亲和力。

本任务族共有 2 个独立入口、2 个名称。

Skill技术路线/资源标签适用范围与主要差异
AutoDock-Vina-分子-对接R/Bioconductor、Python、CLIMolecular docking with AutoDock Vina (Python API). Receptor/ligand prep (Meeko + RDKit), grid box, docking, pose and binding energy analysis, and batch virtual screening.
DiffDock-L-预测小分子结合姿势Python、CLI使用 DiffDock-L(Corso 等,2023/2024)预测小分子结合姿势。它通过盲扩散对接,在无需预定义搜索框的情况下将配体放入蛋白质口袋,并使用学习得到的置信度模型对样本排序。适用于将 SMILES 或 SDF 对接至 PDB、为小型片段库生成排序后的三维姿势,或为后续重打分生成起始姿势。DiffDock 预测几何构象,而非亲和力。

结合亲和力预测

共同目标:围绕“结合亲和力预测”完成同一分析环节;以下入口保留不同算法、工具或工作流。

重点比较:比较经验打分、学习模型和自由能方法,关注单位、校准域、构象依赖和不确定性。

本任务族共有 2 个独立入口、2 个名称。

Skill技术路线/资源标签适用范围与主要差异
计算GPU、Python、CLIPerforms alchemical free-energy calculations including relative binding free energy (RBFE / FEP+) and absolute binding free energy (ABFE) via OpenFE, FEP+, GROMACS, AMBER pmemd, and OpenMM with explicit lambda schedulin…
Boltz-2-Boltz-2模型预测蛋白质-配体结合亲和力PythonPredict protein-ligand binding affinity using Boltz-2 model to assess molecular interactions and binding probability for drug discovery.

虚拟筛选

共同目标:围绕“虚拟筛选”完成同一分析环节;以下入口保留不同算法、工具或工作流。

重点比较:比较 docking、药效团、相似性和学习排序,关注库规模、去重、PAINS/反应性过滤及复评分。

本任务族共有 4 个独立入口、4 个名称。

Skill技术路线/资源标签适用范围与主要差异
Virtual-筛选PythonPerforms structure-based virtual screening using AutoDock Vina 1.2 for molecular docking. Prepares receptor PDBQT files, generates ligand conformers, defines binding site boxes, and ranks compounds by predicted binding…
Virtual-筛选GPU、Python、CLIPerforms structure-based virtual screening using AutoDock Vina, SMINA, GNINA (CNN scoring), and DiffDock-L hybrid workflows with explicit choice rules across rigid vs flexible docking, cross-docking vs self-docking, bin…
虚拟筛选工作流PythonVirtual Screening Pipeline - Virtual screening: search PubChem by substructure, compute similarity, filter by drug-likeness, and predict binding affinity. Use this skill for drug discovery tasks involving search pubchem…
虚拟筛选工作流脚手架读取入口说明确认依赖与输入Virtual screening skill scaffold; replace this scaffold with a tested implementation.

小分子表征与特征

共同目标:围绕“小分子表征与特征”完成同一分析环节;以下入口保留不同算法、工具或工作流。

重点比较:比较指纹、理化描述符、图网络和 3D 构象特征,关注立体化学、标准化和适用域。

本任务族共有 39 个独立入口、39 个名称。

Skill技术路线/资源标签适用范围与主要差异
Chemical-结构-Fingerprint-比较PythonUse when when you have MS/MS spectra with known chemical structures (InChIKeys
化合物-structural-fingerprint-比较R/BioconductorUse when you have a set of query chemicals and a reference library (organized
描述符指纹特征工程读取入口说明确认依赖与输入Use when use when the workflow requires descriptor-fingerprint-feature-engineering.
分子指纹计算读取入口说明确认依赖与输入Use when when you have a set of chemical compounds (as SMILES, mol, SDF,
保留时间预测读取入口说明确认依赖与输入Use when when you have a collection of small molecules with known chemical
fingerprint-特征-格式-转换Python、CLIUse when you have a pretrained deep learning model (e.
fingerprint-生成-smiles读取入口说明确认依赖与输入Use when you have molecule IDs from PubChem or HMDB that have been converted
fingerprint-生成-morganPythonUse when you have parsed molecular structures from the SMRT dataset or
fingerprint-相似度-计算-rdkitPythonUse when when implementing a similarity-based filter for Pickaxe compound
fingerprint-vector-加载-解析PythonUse when when you have deposited or archived biosynfoni fingerprint vectors
IOKR-fingerprint-space-排序读取入口说明确认依赖与输入Use when when you have paired MS2 spectra and BGCs with structural candidates
核回归学习读取入口说明确认依赖与输入Use when when you have a training set of MS2 spectra with known chemical
脂质指纹再生神经网络读取入口说明确认依赖与输入Use when you have MS/MS spectra with initial lipid annotations from spectral
质量-光谱-fingerprint-生成PythonUse when you have unaligned MS2 spectra from one or more samples (in
分子-descriptor-fingerprint-生成读取入口说明确认依赖与输入Use when you have a collection of chemical structures (SMILES, InChI,
分析PythonMolecular Fingerprint Analysis - Fingerprint analysis: topology descriptors, structure complexity, similarity calculation, and AromaticityAnalysis. Use this skill for cheminformatics tasks involving calculate mol topolo…
分子-fingerprint-bit-频率-分析PythonUse when you have loaded a collection of molecular fingerprint vectors
分子-fingerprint-计算读取入口说明确认依赖与输入Use when you have a collection of small molecules in standardized format
分子-fingerprint-计算PythonUse when you have natural product molecules (or compounds from natural
分子指纹提取与向量化读取入口说明确认依赖与输入Use when you have annotated metabolite structures (with SMILES strings)
分子-fingerprint-生成PythonUse when when you have annotated chemical structures (SMILES or InChI
分子-fingerprint-生成-encoding读取入口说明确认依赖与输入Use when when you have paired tandem MS spectra and corresponding molecular
分子-fingerprint-生成-rdkPythonUse when you have a set of chemical compounds represented as InChI or
分子-fingerprint-解析读取入口说明确认依赖与输入Use when you have received a JSON response from the CSI:FingerID web
分子-fingerprint-预测读取入口说明确认依赖与输入Use when you have tandem MS/MS spectra paired with known molecular structures
分子-fingerprint-表征-learning读取入口说明确认依赖与输入Use when when you have labeled mass-spectrometry spectral data (precursor
当评估质谱相似度评分与实际化学结构之间的相关性时PythonUse when when evaluating how well mass spectral similarity scores correlate
机器学习的分子特征化Python、CLIMolecular featurization for ML (100+ featurizers). ECFP, MACCS, descriptors, pretrained models (ChemBERTa), convert SMILES to features, for QSAR and molecular ML.
Molfeat-机器学习的分子特征化枢纽Python、CLIMolecular featurization hub (100+ featurizers) for ML. SMILES to fingerprints (ECFP, MACCS, MAP4), descriptors (RDKit 2D, Mordred), pretrained embeddings (ChemBERTa, GIN, Graphormer), pharmacophores. Scikit-learn compat…
MS2指纹空白去除PythonUse when after aligning MS2 fingerprints across samples (generating a
ms2-fingerprint-距离-计算PythonUse when after MS2 fingerprints have been generated by counting MS2 peaks
ms2-fingerprint-矩阵-mergingPythonUse when you have two MemoMatrix objects generated from separate sample
ms2-fingerprint-vector-生成PythonUse when you have LC-MS/MS data in mzML, mzXML, or MGF format from one
RDKit-计算PythonUse when you need to (1) compare structural similarity between generated
光谱-fingerprint-vectorizationPythonUse when you have MS2 fragmentation spectra from multiple metabolomics
光谱-fingerprint-web-service-查询读取入口说明确认依赖与输入Use when you have a high-resolution LC-MS/MS spectrum or pre-computed
光谱指纹对比学习读取入口说明确认依赖与输入Use when when you have paired tandem MS spectra and corresponding molecular
计算读取入口说明确认依赖与输入Use when you have a set of compounds (e.g., novel NPS analogues in an
tanimoto指纹计算PythonUse when when preparing paired MS/MS spectra for training or validation

ADMET 与毒性预测

共同目标:围绕“ADMET 与毒性预测”完成同一分析环节;以下入口保留不同算法、工具或工作流。

重点比较:比较单终点/多任务、规则/ML 和体内外终点,关注数据泄漏、适用域和不确定性。

本任务族共有 7 个独立入口、7 个名称。

Skill技术路线/资源标签适用范围与主要差异
ADMET-药物性质报告PythonADMET & Drug-Likeness Report - Generate comprehensive ADMET and drug-likeness report: molecular properties, H-bond analysis, hydrophobicity, topology, and ADMET prediction. Use this skill for medicinal chemistry tasks i…
ADMET-导向遗传分子优化工作流Python、CLIADMET-guided genetic molecule optimization workflow from seed SMILES; use when the agent needs to build or run an RDKit/SA-Score/ADMET-AI GA pipeline for molecule optimization, enforce molecule lineage logs, render opti…
ADMET-预测PythonPredicts ADMET properties using ADMETlab 3.0 (119 platform features, including 77 prediction models with modeled-endpoint uncertainty), ADMET-AI, DeepChem MolNet, and chemprop D-MPNN with explicit handling of OECD QSAR…
ADMET-预测读取入口说明确认依赖与输入ADMET prediction skill scaffold; replace this scaffold with a tested implementation.
ADMET-预测Python、网络/APIPredicts ADMET properties using ADMETlab 3.0 API or DeepChem models. Estimates bioavailability, CYP inhibition, hERG liability, and 119 toxicity endpoints with uncertainty quantification. Filters for PAINS and other str…
预测输入分子的ADMETPythonPredict the ADMET (absorption, distribution, metabolism, excretion, and toxicity) properties of the input molecules.
ADMET-预测GPU、CPU、Python、CLIComprehensive ADMET (Absorption, Distribution, Metabolism, Excretion, Toxicity) profiling for drug candidates. Integrates ADMET-AI predictions, SwissADME drug-likeness, PubChemTox experimental toxicity, ChEMBL clinical…

名称不同但语义近似的补充候选

以下分组由名称与描述中的稀有任务词识别,未自动合并。它们可能是别名、 包含关系或相邻步骤;使用前应以各自入口的输入输出边界为准。

候选组 1

  • 分析:Bioassay Data Analysis - Analyze bioassay data: PubChem assay summary, ChEMBL activity search, compound properties, and target info. Use this skill for bioassay science tasks invo…
  • 分析:Cell Line Assay Analysis - Analyze cell line assays: ChEMBL cell line info, assay search, activity data, and target info. Use this skill for cell biology tasks involving get cell…

示例提示词

围绕《你的疾病或候选靶点》梳理遗传、功能和药理证据,比较支持与反对理由,提出值得进一步验证的问题。

包含的科研 Skills25 个

常见问题

AI药物发现能做什么?

靶点、QSAR、ADMET、筛选与分子生成。

如何安装AI药物发现?

在科搭 Codar 左侧点击「插件」,在「插件」页签搜索「AI药物发现」并安装;也可以把本页的安装提示词发给 Claude Code 或 Codex。

AI药物发现需要哪些权限?

该科研插件没有声明额外权限;执行具体操作时仍由 Agent 按自身规则请求确认。

AI药物发现经过官方审核吗?

是。SpeedSkills 收录的每个版本都会记录内容摘要与权限声明,当前版本已通过安全审核。