[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7BKAXYDNMD92Q4NHH4WZC2H":3,"skill-same:01M7BKAXYDNMD92Q4NHH4WZC2H":75},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":29,"ownerHtmlUrl":30,"ownerBlogUrl":6,"ownerDescription":31,"updatedAt":32,"install":33,"resources":46},"01M7BKAXYDNMD92Q4NHH4WZC2H","pytdc",null,"Therapeutics Data Commons. AI-ready drug discovery datasets (ADME, toxicity, DTI), benchmarks, scaffold splits, molecular oracles, for therapeutic ML and pharmacological prediction.","YYC-Cube\u002FYYC3-Knowledge-Base","YYC-Cube","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F235006543?v=4","01M79J2PH4SWD2X0825X6PN748","yyc3","01M79J2PHVK3F6ZBYT8K2SCZCW",1,"MIT","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-AI-应用工具\u002FYYC3-AI-应用工具-claude-code-templates-main\u002Fcli-tool\u002Fcomponents\u002Fskills\u002Fscientific\u002Fpytdc\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: pytdc\ndescription: \"Therapeutics Data Commons. AI-ready drug discovery datasets (ADME, toxicity, DTI), benchmarks, scaffold splits, molecular oracles, for therapeutic ML and pharmacological prediction.\"\n---\n\n# PyTDC (Therapeutics Data Commons)\n\n## Overview\n\nPyTDC is an open-science platform providing AI-ready datasets and benchmarks for drug discovery and development. Access curated datasets spanning the entire therapeutics pipeline with standardized evaluation metrics and meaningful data splits, organized into three categories: single-instance prediction (molecular\u002Fprotein properties), multi-instance prediction (drug-target interactions, DDI), and generation (molecule generation, retrosynthesis).\n\n## When to Use This Skill\n\nThis skill should be used when:\n- Working with drug discovery or therapeutic ML datasets\n- Benchmarking machine learning models on standardized pharmaceutical tasks\n- Predicting molecular properties (ADME, toxicity, bioactivity)\n- Predicting drug-target or drug-drug interactions\n- Generating novel molecules with desired properties\n- Accessing curated datasets with proper train\u002Ftest splits (scaffold, cold-split)\n- Using molecular oracles for property optimization\n\n## Installation & Setup\n\nInstall PyTDC using pip:\n\n```bash\nuv pip install PyTDC\n```\n\nTo upgrade to the latest version:\n\n```bash\nuv pip install PyTDC --upgrade\n```\n\nCore dependencies (automatically installed):\n- numpy, pandas, tqdm, seaborn, scikit_learn, fuzzywuzzy\n\nAdditional packages are installed automatically as needed for specific features.\n\n## Quick Start\n\nThe basic pattern for accessing any TDC dataset follows this structure:\n\n```python\nfrom tdc.\u003Cproblem> import \u003CTask>\ndata = \u003CTask>(name='\u003CDataset>')\nsplit = data.get_split(method='scaffold', seed=1, frac=[0.7, 0.1, 0.2])\ndf = data.get_data(format='df')\n```\n\nWhere:\n- `\u003Cproblem>`: One of `single_pred`, `multi_pred`, or `generation`\n- `\u003CTask>`: Specific task category (e.g., ADME, DTI, MolGen)\n- `\u003CDataset>`: Dataset name within that task\n\n**Example - Loading ADME data:**\n\n```python\nfrom tdc.single_pred import ADME\ndata = ADME(name='Caco2_Wang')\nsplit = data.get_split(method='scaffold')\n# Returns dict with 'train', 'valid', 'test' DataFrames\n```\n\n## Single-Instance Prediction Tasks\n\nSingle-instance prediction involves forecasting properties of individual biomedical entities (molecules, proteins, etc.).\n\n### Available Task Categories\n\n#### 1. ADME (Absorption, Distribution, Metabolism, Excretion)\n\nPredict pharmacokinetic properties of drug molecules.\n\n```python\nfrom tdc.single_pred import ADME\ndata = ADME(name='Caco2_Wang')  # Intestinal permeability\n# Other datasets: HIA_Hou, Bioavailability_Ma, Lipophilicity_AstraZeneca, etc.\n```\n\n**Common ADME datasets:**\n- Caco2 - Intestinal permeability\n- HIA - Human intestinal absorption\n- Bioavailability - Oral bioavailability\n- Lipophilicity - Octanol-water partition coefficient\n- Solubility - Aqueous solubility\n- BBB - Blood-brain barrier penetration\n- CYP - Cytochrome P450 metabolism\n\n#### 2. Toxicity (Tox)\n\nPredict toxicity and adverse effects of compounds.\n\n```python\nfrom tdc.single_pred import Tox\ndata = Tox(name='hERG')  # Cardiotoxicity\n# Other datasets: AMES, DILI, Carcinogens_Lagunin, etc.\n```\n\n**Common toxicity datasets:**\n- hERG - Cardiac toxicity\n- AMES - Mutagenicity\n- DILI - Drug-induced liver injury\n- Carcinogens - Carcinogenicity\n- ClinTox - Clinical trial toxicity\n\n#### 3. HTS (High-Throughput Screening)\n\nBioactivity predictions from screening data.\n\n```python\nfrom tdc.single_pred import HTS\ndata = HTS(name='SARSCoV2_Vitro_Touret')\n```\n\n#### 4. QM (Quantum Mechanics)\n\nQuantum mechanical properties of molecules.\n\n```python\nfrom tdc.single_pred import QM\ndata = QM(name='QM7')\n```\n\n#### 5. Other Single Prediction Tasks\n\n- **Yields**: Chemical reaction yield prediction\n- **Epitope**: Epitope prediction for biologics\n- **Develop**: Development-stage predictions\n- **CRISPROutcome**: Gene editing outcome prediction\n\n### Data Format\n\nSingle prediction datasets typically return DataFrames with columns:\n- `Drug_ID` or `Compound_ID`: Unique identifier\n- `Drug` or `X`: SMILES string or molecular representation\n- `Y`: Target label (continuous or binary)\n\n## Multi-Instance Prediction Tasks\n\nMulti-instance prediction involves forecasting properties of interactions between multiple biomedical entities.\n\n### Available Task Categories\n\n#### 1. DTI (Drug-Target Interaction)\n\nPredict binding affinity between drugs and protein targets.\n\n```python\nfrom tdc.multi_pred import DTI\ndata = DTI(name='BindingDB_Kd')\nsplit = data.get_split()\n```\n\n**Available datasets:**\n- BindingDB_Kd - Dissociation constant (52,284 pairs)\n- BindingDB_IC50 - Half-maximal inhibitory concentration (991,486 pairs)\n- BindingDB_Ki - Inhibition constant (375,032 pairs)\n- DAVIS, KIBA - Kinase binding datasets\n\n**Data format:** Drug_ID, Target_ID, Drug (SMILES), Target (sequence), Y (binding affinity)\n\n#### 2. DDI (Drug-Drug Interaction)\n\nPredict interactions between drug pairs.\n\n```python\nfrom tdc.multi_pred import DDI\ndata = DDI(name='DrugBank')\nsplit = data.get_split()\n```\n\nMulti-class classification task predicting interaction types. Dataset contains 191,808 DDI pairs with 1,706 drugs.\n\n#### 3. PPI (Protein-Protein Interaction)\n\nPredict protein-protein interactions.\n\n```python\nfrom tdc.multi_pred import PPI\ndata = PPI(name='HuRI')\n```\n\n#### 4. Other Multi-Prediction Tasks\n\n- **GDA**: Gene-disease associations\n- **DrugRes**: Drug resistance prediction\n- **DrugSyn**: Drug synergy prediction\n- **PeptideMHC**: Peptide-MHC binding\n- **AntibodyAff**: Antibody affinity prediction\n- **MTI**: miRNA-target interactions\n- **Catalyst**: Catalyst prediction\n- **TrialOutcome**: Clinical trial outcome prediction\n\n## Generation Tasks\n\nGeneration tasks involve creating novel biomedical entities with desired properties.\n\n### 1. Molecular Generation (MolGen)\n\nGenerate diverse, novel molecules with desirable chemical properties.\n\n```python\nfrom tdc.generation import MolGen\ndata = MolGen(name='ChEMBL_V29')\nsplit = data.get_split()\n```\n\nUse with oracles to optimize for specific properties:\n\n```python\nfrom tdc import Oracle\noracle = Oracle(name='GSK3B')\nscore = oracle('CC(C)Cc1ccc(cc1)C(C)C(O)=O')  # Evaluate SMILES\n```\n\nSee `references\u002Foracles.md` for all available oracle functions.\n\n### 2. Retrosynthesis (RetroSyn)\n\nPredict reactants needed to synthesize a target molecule.\n\n```python\nfrom tdc.generation import RetroSyn\ndata = RetroSyn(name='USPTO')\nsplit = data.get_split()\n```\n\nDataset contains 1,939,253 reactions from USPTO database.\n\n### 3. Paired Molecule Generation\n\nGenerate molecule pairs (e.g., prodrug-drug pairs).\n\n```python\nfrom tdc.generation import PairMolGen\ndata = PairMolGen(name='Prodrug')\n```\n\nFor detailed oracle documentation and molecular generation workflows, refer to `references\u002Foracles.md` and `scripts\u002Fmolecular_generation.py`.\n\n## Benchmark Groups\n\nBenchmark groups provide curated collections of related datasets for systematic model evaluation.\n\n### ADMET Benchmark Group\n\n```python\nfrom tdc.benchmark_group import admet_group\ngroup = admet_group(path='data\u002F')\n\n# Get benchmark datasets\nbenchmark = group.get('Caco2_Wang')\npredictions = {}\n\nfor seed in [1, 2, 3, 4, 5]:\n    train, valid = benchmark['train'], benchmark['valid']\n    # Train model here\n    predictions[seed] = model.predict(benchmark['test'])\n\n# Evaluate with required 5 seeds\nresults = group.evaluate(predictions)\n```\n\n**ADMET Group includes 22 datasets** covering absorption, distribution, metabolism, excretion, and toxicity.\n\n### Other Benchmark Groups\n\nAvailable benchmark groups include collections for:\n- ADMET properties\n- Drug-target interactions\n- Drug combination prediction\n- And more specialized therapeutic tasks\n\nFor benchmark evaluation workflows, see `scripts\u002Fbenchmark_evaluation.py`.\n\n## Data Functions\n\nTDC provides comprehensive data processing utilities organized into four categories.\n\n### 1. Dataset Splits\n\nRetrieve train\u002Fvalidation\u002Ftest partitions with various strategies:\n\n```python\n# Scaffold split (default for most tasks)\nsplit = data.get_split(method='scaffold', seed=1, frac=[0.7, 0.1, 0.2])\n\n# Random split\nsplit = data.get_split(method='random', seed=42, frac=[0.8, 0.1, 0.1])\n\n# Cold split (for DTI\u002FDDI tasks)\nsplit = data.get_split(method='cold_drug', seed=1)  # Unseen drugs in test\nsplit = data.get_split(method='cold_target', seed=1)  # Unseen targets in test\n```\n\n**Available split strategies:**\n- `random`: Random shuffling\n- `scaffold`: Scaffold-based (for chemical diversity)\n- `cold_drug`, `cold_target`, `cold_drug_target`: For DTI tasks\n- `temporal`: Time-based splits for temporal datasets\n\n### 2. Model Evaluation\n\nUse standardized metrics for evaluation:\n\n```python\nfrom tdc import Evaluator\n\n# For binary classification\nevaluator = Evaluator(name='ROC-AUC')\nscore = evaluator(y_true, y_pred)\n\n# For regression\nevaluator = Evaluator(name='RMSE')\nscore = evaluator(y_true, y_pred)\n```\n\n**Available metrics:** ROC-AUC, PR-AUC, F1, Accuracy, RMSE, MAE, R2, Spearman, Pearson, and more.\n\n### 3. Data Processing\n\nTDC provides 11 key processing utilities:\n\n```python\nfrom tdc.chem_utils import MolConvert\n\n# Molecule format conversion\nconverter = MolConvert(src='SMILES', dst='PyG')\npyg_graph = converter('CC(C)Cc1ccc(cc1)C(C)C(O)=O')\n```\n\n**Processing utilities include:**\n- Molecule format conversion (SMILES, SELFIES, PyG, DGL, ECFP, etc.)\n- Molecule filters (PAINS, drug-likeness)\n- Label binarization and unit conversion\n- Data balancing (over\u002Funder-sampling)\n- Negative sampling for pair data\n- Graph transformation\n- Entity retrieval (CID to SMILES, UniProt to sequence)\n\nFor comprehensive utilities documentation, see `references\u002Futilities.md`.\n\n### 4. Molecule Generation Oracles\n\nTDC provides 17+ oracle functions for molecular optimization:\n\n```python\nfrom tdc import Oracle\n\n# Single oracle\noracle = Oracle(name='DRD2')\nscore = oracle('CC(C)Cc1ccc(cc1)C(C)C(O)=O')\n\n# Multiple oracles\noracle = Oracle(name='JNK3')\nscores = oracle(['SMILES1', 'SMILES2', 'SMILES3'])\n```\n\nFor complete oracle documentation, see `references\u002Foracles.md`.\n\n## Advanced Features\n\n### Retrieve Available Datasets\n\n```python\nfrom tdc.utils import retrieve_dataset_names\n\n# Get all ADME datasets\nadme_datasets = retrieve_dataset_names('ADME')\n\n# Get all DTI datasets\ndti_datasets = retrieve_dataset_names('DTI')\n```\n\n### Label Transformations\n\n```python\n# Get label mapping\nlabel_map = data.get_label_map(name='DrugBank')\n\n# Convert labels\nfrom tdc.chem_utils import label_transform\ntransformed = label_transform(y, from_unit='nM', to_unit='p')\n```\n\n### Database Queries\n\n```python\nfrom tdc.utils import cid2smiles, uniprot2seq\n\n# Convert PubChem CID to SMILES\nsmiles = cid2smiles(2244)\n\n# Convert UniProt ID to amino acid sequence\nsequence = uniprot2seq('P12345')\n```\n\n## Common Workflows\n\n### Workflow 1: Train a Single Prediction Model\n\nSee `scripts\u002Fload_and_split_data.py` for a complete example:\n\n```python\nfrom tdc.single_pred import ADME\nfrom tdc import Evaluator\n\n# Load data\ndata = ADME(name='Caco2_Wang')\nsplit = data.get_split(method='scaffold', seed=42)\n\ntrain, valid, test = split['train'], split['valid'], split['test']\n\n# Train model (user implements)\n# model.fit(train['Drug'], train['Y'])\n\n# Evaluate\nevaluator = Evaluator(name='MAE')\n# score = evaluator(test['Y'], predictions)\n```\n\n### Workflow 2: Benchmark Evaluation\n\nSee `scripts\u002Fbenchmark_evaluation.py` for a complete example with multiple seeds and proper evaluation protocol.\n\n### Workflow 3: Molecular Generation with Oracles\n\nSee `scripts\u002Fmolecular_generation.py` for an example of goal-directed generation using oracle functions.\n\n## Resources\n\nThis skill includes bundled resources for common TDC workflows:\n\n### scripts\u002F\n\n- `load_and_split_data.py`: Template for loading and splitting TDC datasets with various strategies\n- `benchmark_evaluation.py`: Template for running benchmark group evaluations with proper 5-seed protocol\n- `molecular_generation.py`: Template for molecular generation using oracle functions\n\n### references\u002F\n\n- `datasets.md`: Comprehensive catalog of all available datasets organized by task type\n- `oracles.md`: Complete documentation of all 17+ molecule generation oracles\n- `utilities.md`: Detailed guide to data processing, splitting, and evaluation utilities\n\n## Additional Resources\n\n- **Official Website**: https:\u002F\u002Ftdcommons.ai\n- **Documentation**: https:\u002F\u002Ftdc.readthedocs.io\n- **GitHub**: https:\u002F\u002Fgithub.com\u002Fmims-harvard\u002FTDC\n- **Paper**: NeurIPS 2021 - \"Therapeutics Data Commons: Machine Learning Datasets and Tasks for Drug Discovery and Development\"\n","# PyTDC (Therapeutics Data Commons)\n\n## Overview\n\nPyTDC is an open-science platform providing AI-ready datasets and benchmarks for drug discovery and development. Access curated datasets spanning the entire therapeutics pipeline with standardized evaluation metrics and meaningful data splits, organized into three categories: single-instance prediction (molecular\u002Fprotein properties), multi-instance prediction (drug-target interactions, DDI), and generation (molecule generation, retrosynthesis).\n\n## When to Use This Skill\n\nThis skill should be used when:\n- Working with drug discovery or therapeutic ML datasets\n- Benchmarking machine learning models on standardized pharmaceutical tasks\n- Predicting molecular properties (ADME, toxicity, bioactivity)\n- Predicting drug-target or drug-drug interactions\n- Generating novel molecules with desired properties\n- Accessing curated datasets with proper train\u002Ftest splits (scaffold, cold-split)\n- Using molecular oracles for property optimization\n\n## Installation & Setup\n\nInstall PyTDC using pip:\n\n```bash\nuv pip install PyTDC\n```\n\nTo upgrade to the latest version:\n\n```bash\nuv pip install PyTDC --upgrade\n```\n\nCore dependencies (automatically installed):\n- numpy, pandas, tqdm, seaborn, scikit_learn, fuzzywuzzy\n\nAdditional packages are installed automatically as needed for specific features.\n\n## Quick Start\n\nThe basic pattern for accessing any TDC dataset follows this structure:\n\n```python\nfrom tdc.\u003Cproblem> import \u003CTask>\ndata = \u003CTask>(name='\u003CDataset>')\nsplit = data.get_split(method='scaffold', seed=1, frac=[0.7, 0.1, 0.2])\ndf = data.get_data(format='df')\n```\n\nWhere:\n- `\u003Cproblem>`: One of `single_pred`, `multi_pred`, or `generation`\n- `\u003CTask>`: Specific task category (e.g., ADME, DTI, MolGen)\n- `\u003CDataset>`: Dataset name within that task\n\n**Example - Loading ADME data:**\n\n```python\nfrom tdc.single_pred import ADME\ndata = ADME(name='Caco2_Wang')\nsplit = data.get_split(method='scaffold')\n# Returns dict with 'train', 'valid', 'test' DataFrames\n```\n\n## Single-Instance Prediction Tasks\n\nSingle-instance prediction involves forecasting properties of individual biomedical entities (molecules, proteins, etc.).\n\n### Available Task Categories\n\n#### 1. ADME (Absorption, Distribution, Metabolism, Excretion)\n\nPredict pharmacokinetic properties of drug molecules.\n\n```python\nfrom tdc.single_pred import ADME\ndata = ADME(name='Caco2_Wang')  # Intestinal permeability\n# Other datasets: HIA_Hou, Bioavailability_Ma, Lipophilicity_AstraZeneca, etc.\n```\n\n**Common ADME datasets:**\n- Caco2 - Intestinal permeability\n- HIA - Human intestinal absorption\n- Bioavailability - Oral bioavailability\n- Lipophilicity - Octanol-water partition coefficient\n- Solubility - Aqueous solubility\n- BBB - Blood-brain barrier penetration\n- CYP - Cytochrome P450 metabolism\n\n#### 2. Toxicity (Tox)\n\nPredict toxicity and adverse effects of compounds.\n\n```python\nfrom tdc.single_pred import Tox\ndata = Tox(name='hERG')  # Cardiotoxicity\n# Other datasets: AMES, DILI, Carcinogens_Lagunin, etc.\n```\n\n**Common toxicity datasets:**\n- hERG - Cardiac toxicity\n- AMES - Mutagenicity\n- DILI - Drug-induced liver injury\n- Carcinogens - Carcinogenicity\n- ClinTox - Clinical trial toxicity\n\n#### 3. HTS (High-Throughput Screening)\n\nBioactivity predictions from screening data.\n\n```python\nfrom tdc.single_pred import HTS\ndata = HTS(name='SARSCoV2_Vitro_Touret')\n```\n\n#### 4. QM (Quantum Mechanics)\n\nQuantum mechanical properties of molecules.\n\n```python\nfrom tdc.single_pred import QM\ndata = QM(name='QM7')\n```\n\n#### 5. Other Single Prediction Tasks\n\n- **Yields**: Chemical reaction yield prediction\n- **Epitope**: Epitope prediction for biologics\n- **Develop**: Development-stage predictions\n- **CRISPROutcome**: Gene editing outcome prediction\n\n### Data Format\n\nSingle prediction datasets typically return DataFrames with columns:\n- `Drug_ID` or `Compound_ID`: Unique identifier\n- `Drug` or `X`: SMILES string or molecular representation\n- `Y`: Target label (continuous or binary)\n\n## Multi-Instance Prediction Tasks\n\nMulti-instance prediction involves forecasting properties of interactions between multiple biomedical entities.\n\n### Available Task Categories\n\n#### 1. DTI (Drug-Target Interaction)\n\nPredict binding affinity between drugs and protein targets.\n\n```python\nfrom tdc.multi_pred import DTI\ndata = DTI(name='BindingDB_Kd')\nsplit = data.get_split()\n```\n\n**Available datasets:**\n- BindingDB_Kd - Dissociation constant (52,284 pairs)\n- BindingDB_IC50 - Half-maximal inhibitory concentration (991,486 pairs)\n- BindingDB_Ki - Inhibition constant (375,032 pairs)\n- DAVIS, KIBA - Kinase binding datasets\n\n**Data format:** Drug_ID, Target_ID, Drug (SMILES), Target (sequence), Y (binding affinity)\n\n#### 2. DDI (Drug-Drug Interaction)\n\nPredict interactions between drug pairs.\n\n```python\nfrom tdc.multi_pred import DDI\ndata = DDI(name='DrugBank')\nsplit = data.get_split()\n```\n\nMulti-class classification task predicting interaction types. Dataset contains 191,808 DDI pairs with 1,706 drugs.\n\n#### 3. PPI (Protein-Protein Interaction)\n\nPredict protein-protein interactions.\n\n```python\nfrom tdc.multi_pred import PPI\ndata = PPI(name='HuRI')\n```\n\n#### 4. Other Multi-Prediction Tasks\n\n- **GDA**: Gene-disease associations\n- **DrugRes**: Drug resistance prediction\n- **DrugSyn**: Drug synergy prediction\n- **PeptideMHC**: Peptide-MHC binding\n- **AntibodyAff**: Antibody affinity prediction\n- **MTI**: miRNA-target interactions\n- **Catalyst**: Catalyst prediction\n- **TrialOutcome**: Clinical trial outcome prediction\n\n## Generation Tasks\n\nGeneration tasks involve creating novel biomedical entities with desired properties.\n\n### 1. Molecular Generation (MolGen)\n\nGenerate diverse, novel molecules with desirable chemical properties.\n\n```python\nfrom tdc.generation import MolGen\ndata = MolGen(name='ChEMBL_V29')\nsplit = data.get_split()\n```\n\nUse with oracles to optimize for specific properties:\n\n```python\nfrom tdc import Oracle\noracle = Oracle(name='GSK3B')\nscore = oracle('CC(C)Cc1ccc(cc1)C(C)C(O)=O')  # Evaluate SMILES\n```\n\nSee `references\u002Foracles.md` for all available oracle functions.\n\n### 2. Retrosynthesis (RetroSyn)\n\nPredict reactants needed to synthesize a target molecule.\n\n```python\nfrom tdc.generation import RetroSyn\ndata = RetroSyn(name='USPTO')\nsplit = data.get_split()\n```\n\nDataset contains 1,939,253 reactions from USPTO database.\n\n### 3. Paired Molecule Generation\n\nGenerate molecule pairs (e.g., prodrug-drug pairs).\n\n```python\nfrom tdc.generation import PairMolGen\ndata = PairMolGen(name='Prodrug')\n```\n\nFor detailed oracle documentation and molecular generation workflows, refer to `references\u002Foracles.md` and `scripts\u002Fmolecular_generation.py`.\n\n## Benchmark Groups\n\nBenchmark groups provide curated collections of related datasets for systematic model evaluation.\n\n### ADMET Benchmark Group\n\n```python\nfrom tdc.benchmark_group import admet_group\ngroup = admet_group(path='data\u002F')\n\n# Get benchmark datasets\nbenchmark = group.get('Caco2_Wang')\npredictions = {}\n\nfor seed in [1, 2, 3, 4, 5]:\n    train, valid = benchmark['train'], benchmark['valid']\n    # Train model here\n    predictions[seed] = model.predict(benchmark['test'])\n\n# Evaluate with required 5 seeds\nresults = group.evaluate(predictions)\n```\n\n**ADMET Group includes 22 datasets** covering absorption, distribution, metabolism, excretion, and toxicity.\n\n### Other Benchmark Groups\n\nAvailable benchmark groups include collections for:\n- ADMET properties\n- Drug-target interactions\n- Drug combination prediction\n- And more specialized therapeutic tasks\n\nFor benchmark evaluation workflows, see `scripts\u002Fbenchmark_evaluation.py`.\n\n## Data Functions\n\nTDC provides comprehensive data processing utilities organized into four categories.\n\n### 1. Dataset Splits\n\nRetrieve train\u002Fvalidation\u002Ftest partitions with various strategies:\n\n```python\n# Scaffold split (default for most tasks)\nsplit = data.get_split(method='scaffold', seed=1, frac=[0.7, 0.1, 0.2])\n\n# Random split\nsplit = data.get_split(method='random', seed=42, frac=[0.8, 0.1, 0.1])\n\n# Cold split (for DTI\u002FDDI tasks)\nsplit = data.get_split(method='cold_drug', seed=1)  # Unseen drugs in test\nsplit = data.get_split(method='cold_target', seed=1)  # Unseen targets in test\n```\n\n**Available split strategies:**\n- `random`: Random shuffling\n- `scaffold`: Scaffold-based (for chemical diversity)\n- `cold_drug`, `cold_target`, `cold_drug_target`: For DTI tasks\n- `temporal`: Time-based splits for temporal datasets\n\n### 2. Model Evaluation\n\nUse standardized metrics for evaluation:\n\n```python\nfrom tdc import Evaluator\n\n# For binary classification\nevaluator = Evaluator(name='ROC-AUC')\nscore = evaluator(y_true, y_pred)\n\n# For regression\nevaluator = Evaluator(name='RMSE')\nscore = evaluator(y_true, y_pred)\n```\n\n**Available metrics:** ROC-AUC, PR-AUC, F1, Accuracy, RMSE, MAE, R2, Spearman, Pearson, and more.\n\n### 3. Data Processing\n\nTDC provides 11 key processing utilities:\n\n```python\nfrom tdc.chem_utils import MolConvert\n\n# Molecule format conversion\nconverter = MolConvert(src='SMILES', dst='PyG')\npyg_graph = converter('CC(C)Cc1ccc(cc1)C(C)C(O)=O')\n```\n\n**Processing utilities include:**\n- Molecule format conversion (SMILES, SELFIES, PyG, DGL, ECFP, etc.)\n- Molecule filters (PAINS, drug-likeness)\n- Label binarization and unit conversion\n- Data balancing (over\u002Funder-sampling)\n- Negative sampling for pair data\n- Graph transformation\n- Entity retrieval (CID to SMILES, UniProt to sequence)\n\nFor comprehensive utilities documentation, see `references\u002Futilities.md`.\n\n### 4. Molecule Generation Oracles\n\nTDC provides 17+ oracle functions for molecular optimization:\n\n```python\nfrom tdc import Oracle\n\n# Single oracle\noracle = Oracle(name='DRD2')\nscore = oracle('CC(C)Cc1ccc(cc1)C(C)C(O)=O')\n\n# Multiple oracles\noracle = Oracle(name='JNK3')\nscores = oracle(['SMILES1', 'SMILES2', 'SMILES3'])\n```\n\nFor complete oracle documentation, see `references\u002Foracles.md`.\n\n## Advanced Features\n\n### Retrieve Available Datasets\n\n```python\nfrom tdc.utils import retrieve_dataset_names\n\n# Get all ADME datasets\nadme_datasets = retrieve_dataset_names('ADME')\n\n# Get all DTI datasets\ndti_datasets = retrieve_dataset_names('DTI')\n```\n\n### Label Transformations\n\n```python\n# Get label mapping\nlabel_map = data.get_label_map(name='DrugBank')\n\n# Convert labels\nfrom tdc.chem_utils import label_transform\ntransformed = label_transform(y, from_unit='nM', to_unit='p')\n```\n\n### Database Queries\n\n```python\nfrom tdc.utils import cid2smiles, uniprot2seq\n\n# Convert PubChem CID to SMILES\nsmiles = cid2smiles(2244)\n\n# Convert UniProt ID to amino acid sequence\nsequence = uniprot2seq('P12345')\n```\n\n## Common Workflows\n\n### Workflow 1: Train a Single Prediction Model\n\nSee `scripts\u002Fload_and_split_data.py` for a complete example:\n\n```python\nfrom tdc.single_pred import ADME\nfrom tdc import Evaluator\n\n# Load data\ndata = ADME(name='Caco2_Wang')\nsplit = data.get_split(method='scaffold', seed=42)\n\ntrain, valid, test = split['train'], split['valid'], split['test']\n\n# Train model (user implements)\n# model.fit(train['Drug'], train['Y'])\n\n# Evaluate\nevaluator = Evaluator(name='MAE')\n# score = evaluator(test['Y'], predictions)\n```\n\n### Workflow 2: Benchmark Evaluation\n\nSee `scripts\u002Fbenchmark_evaluation.py` for a complete example with multiple seeds and proper evaluation protocol.\n\n### Workflow 3: Molecular Generation with Oracles\n\nSee `scripts\u002Fmolecular_generation.py` for an example of goal-directed generation using oracle functions.\n\n## Resources\n\nThis skill includes bundled resources for common TDC workflows:\n\n### scripts\u002F\n\n- `load_and_split_data.py`: Template for loading and splitting TDC datasets with various strategies\n- `benchmark_evaluation.py`: Template for running benchmark group evaluations with proper 5-seed protocol\n- `molecular_generation.py`: Template for molecular generation using oracle functions\n\n### references\u002F\n\n- `datasets.md`: Comprehensive catalog of all available datasets organized by task type\n- `oracles.md`: Complete documentation of all 17+ molecule generation oracles\n- `utilities.md`: Detailed guide to data processing, splitting, and evaluation utilities\n\n## Additional Resources\n\n- **Official Website**: https:\u002F\u002Ftdcommons.ai\n- **Documentation**: https:\u002F\u002Ftdc.readthedocs.io\n- **GitHub**: https:\u002F\u002Fgithub.com\u002Fmims-harvard\u002FTDC\n- **Paper**: NeurIPS 2021 - \"Therapeutics Data Commons: Machine Learning Datasets and Tasks for Drug Discovery and Development\"\n",{"name":5,"description":7},"main","87afe6ce96220accf50a345dea46784c8c9f210e","YYC3-AI-应用工具\u002FYYC3-AI-应用工具-claude-code-templates-main\u002Fcli-tool\u002Fcomponents\u002Fskills\u002Fscientific\u002Fpytdc",0,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base","https:\u002F\u002Fkb.yyc3.top\u002F","https:\u002F\u002Fgithub.com\u002FYYC-Cube","万象归元于云枢 | 深栈智启新纪元","2026-11-10 10:37:32",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add YYC-Cube\u002FYYC3-Knowledge-Base --skill pytdc",{"tool":39,"command":40},"bunx","bunx skills add YYC-Cube\u002FYYC3-Knowledge-Base --skill pytdc",{"tool":42,"command":43},"pnpm","pnpm dlx skills add YYC-Cube\u002FYYC3-Knowledge-Base --skill pytdc","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-AI-应用工具\u002FYYC3-AI-应用工具-claude-code-templates-main\u002Fcli-tool\u002Fcomponents\u002Fskills\u002Fscientific\u002Fpytdc\u002FSKILL.md",true,[47,53,57,61,67,71],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"references\u002Fdatasets.md","REFERENCE","参考文档",8792,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-AI-应用工具\u002FYYC3-AI-应用工具-claude-code-templates-main\u002Fcli-tool\u002Fcomponents\u002Fskills\u002Fscientific\u002Fpytdc\u002Freferences\u002Fdatasets.md",{"path":54,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":55,"sourceHtmlUrl":56},"references\u002Foracles.md",10051,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-AI-应用工具\u002FYYC3-AI-应用工具-claude-code-templates-main\u002Fcli-tool\u002Fcomponents\u002Fskills\u002Fscientific\u002Fpytdc\u002Freferences\u002Foracles.md",{"path":58,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":59,"sourceHtmlUrl":60},"references\u002Futilities.md",16045,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-AI-应用工具\u002FYYC3-AI-应用工具-claude-code-templates-main\u002Fcli-tool\u002Fcomponents\u002Fskills\u002Fscientific\u002Fpytdc\u002Freferences\u002Futilities.md",{"path":62,"resourceType":63,"resourceTypeDesc":64,"sizeBytes":65,"sourceHtmlUrl":66},"scripts\u002Fbenchmark_evaluation.py","SCRIPT","脚本",9238,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-AI-应用工具\u002FYYC3-AI-应用工具-claude-code-templates-main\u002Fcli-tool\u002Fcomponents\u002Fskills\u002Fscientific\u002Fpytdc\u002Fscripts\u002Fbenchmark_evaluation.py",{"path":68,"resourceType":63,"resourceTypeDesc":64,"sizeBytes":69,"sourceHtmlUrl":70},"scripts\u002Fload_and_split_data.py",6350,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-AI-应用工具\u002FYYC3-AI-应用工具-claude-code-templates-main\u002Fcli-tool\u002Fcomponents\u002Fskills\u002Fscientific\u002Fpytdc\u002Fscripts\u002Fload_and_split_data.py",{"path":72,"resourceType":63,"resourceTypeDesc":64,"sizeBytes":73,"sourceHtmlUrl":74},"scripts\u002Fmolecular_generation.py",11335,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-AI-应用工具\u002FYYC3-AI-应用工具-claude-code-templates-main\u002Fcli-tool\u002Fcomponents\u002Fskills\u002Fscientific\u002Fpytdc\u002Fscripts\u002Fmolecular_generation.py",[76,83,90,97,104,111],{"skillNo":77,"name":78,"nameZh":6,"description":79,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":80,"categories":81,"keywords":82,"publishedAt":19},"01M7CTDWHERCXPQCP0K03A86AS","vercel-sandbox","Run agent-browser + Chrome inside Vercel Sandbox microVMs for browser automation from any Vercel-deployed app. Use when the user needs browser automation in a Vercel app (Next.js, SvelteKit, Nuxt, Remix, Astro, etc.), wants to run headless Chrome without binary size limits, needs persistent browser sessions across commands, or wants ephemeral isolated browser environments. Triggers include \"Vercel Sandbox browser\", \"microVM Chrome\", \"agent-browser in sandbox\", \"browser automation on Vercel\", or any task requiring Chrome in a Vercel Sandbox.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Fvercel-sandbox\u002FSKILL.md",[],[],{"skillNo":84,"name":85,"nameZh":6,"description":86,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":87,"categories":88,"keywords":89,"publishedAt":19},"01M7CTD8D2BNJY54S65M47HEJA","slack","Interact with Slack workspaces using browser automation. Use when the user needs to check unread channels, navigate Slack, send messages, extract data, find information, search conversations, or automate any Slack task. Triggers include \"check my Slack\", \"what channels have unreads\", \"send a message to\", \"search Slack for\", \"extract from Slack\", \"find who said\", or any task requiring programmatic Slack interaction.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Fslack\u002FSKILL.md",[],[],{"skillNo":91,"name":92,"nameZh":6,"description":93,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":94,"categories":95,"keywords":96,"publishedAt":19},"01M7CTCSSB8ZP3Q58C3YZDZK1W","electron","Automate Electron desktop apps (VS Code, Slack, Discord, Figma, Notion, Spotify, etc.) using agent-browser via Chrome DevTools Protocol. Use when the user needs to interact with an Electron app, automate a desktop app, connect to a running app, control a native app, or test an Electron application. Triggers include \"automate Slack app\", \"control VS Code\", \"interact with Discord app\", \"test this Electron app\", \"connect to desktop app\", or any task requiring automation of a native Electron application.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Felectron\u002FSKILL.md",[],[],{"skillNo":98,"name":99,"nameZh":6,"description":100,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":101,"categories":102,"keywords":103,"publishedAt":19},"01M7CTCBJY76688V4DCZGWFHGQ","dogfood","Systematically explore and test a web application to find bugs, UX issues, and other problems. Use when asked to \"dogfood\", \"QA\", \"exploratory test\", \"find issues\", \"bug hunt\", \"test this app\u002Fsite\u002Fplatform\", or review the quality of a web application. Produces a structured report with full reproduction evidence -- step-by-step screenshots, repro videos, and detailed repro steps for every issue -- so findings can be handed directly to the responsible teams.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Fdogfood\u002FSKILL.md",[],[],{"skillNo":105,"name":106,"nameZh":6,"description":107,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":108,"categories":109,"keywords":110,"publishedAt":19},"01M7CTC4H0FFGVEAQR9PA4GA1A","agent-browser","Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to \"open a website\", \"fill out a form\", \"click a button\", \"take a screenshot\", \"scrape data from a page\", \"test this web app\", \"login to a site\", \"automate browser actions\", or any task requiring programmatic web interaction.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Fagent-browser\u002FSKILL.md",[],[],{"skillNo":112,"name":113,"nameZh":6,"description":114,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":115,"categories":116,"keywords":117,"publishedAt":19},"01M7CTBKVFYBG78FDSABSDJTTQ","shadcn","Manages shadcn components and projects — adding, searching, fixing, debugging, styling, and composing UI, including chat interfaces. Provides project context, component docs, and usage examples. Applies when working with shadcn\u002Fui, component registries, presets, --preset codes, or any project with a components.json file. Also triggers for \"shadcn init\", \"create an app with --preset\", \"switch to --preset\", and for building a design system or applying a DESIGN.md or brand spec to shadcn\u002Fui.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-UI-界面组件\u002FYYC3-UI-界面组件-ui\u002Fskills\u002Fshadcn\u002FSKILL.md",[],[]]