[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7BK02QJ173VB00SM7VEEAY0":3,"skill-same:01M7BK02QJ173VB00SM7VEEAY0":85},{"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},"01M7BK02QJ173VB00SM7VEEAY0","diffdock",null,"Diffusion-based molecular docking. Predict protein-ligand binding poses from PDB\u002FSMILES, confidence scores, virtual screening, for structure-based drug design. Not for affinity 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\u002Fdiffdock\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: diffdock\ndescription: \"Diffusion-based molecular docking. Predict protein-ligand binding poses from PDB\u002FSMILES, confidence scores, virtual screening, for structure-based drug design. Not for affinity prediction.\"\n---\n\n# DiffDock: Molecular Docking with Diffusion Models\n\n## Overview\n\nDiffDock is a diffusion-based deep learning tool for molecular docking that predicts 3D binding poses of small molecule ligands to protein targets. It represents the state-of-the-art in computational docking, crucial for structure-based drug discovery and chemical biology.\n\n**Core Capabilities:**\n- Predict ligand binding poses with high accuracy using deep learning\n- Support protein structures (PDB files) or sequences (via ESMFold)\n- Process single complexes or batch virtual screening campaigns\n- Generate confidence scores to assess prediction reliability\n- Handle diverse ligand inputs (SMILES, SDF, MOL2)\n\n**Key Distinction:** DiffDock predicts **binding poses** (3D structure) and **confidence** (prediction certainty), NOT binding affinity (ΔG, Kd). Always combine with scoring functions (GNINA, MM\u002FGBSA) for affinity assessment.\n\n## When to Use This Skill\n\nThis skill should be used when:\n\n- \"Dock this ligand to a protein\" or \"predict binding pose\"\n- \"Run molecular docking\" or \"perform protein-ligand docking\"\n- \"Virtual screening\" or \"screen compound library\"\n- \"Where does this molecule bind?\" or \"predict binding site\"\n- Structure-based drug design or lead optimization tasks\n- Tasks involving PDB files + SMILES strings or ligand structures\n- Batch docking of multiple protein-ligand pairs\n\n## Installation and Environment Setup\n\n### Check Environment Status\n\nBefore proceeding with DiffDock tasks, verify the environment setup:\n\n```bash\n# Use the provided setup checker\npython scripts\u002Fsetup_check.py\n```\n\nThis script validates Python version, PyTorch with CUDA, PyTorch Geometric, RDKit, ESM, and other dependencies.\n\n### Installation Options\n\n**Option 1: Conda (Recommended)**\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fgcorso\u002FDiffDock.git\ncd DiffDock\nconda env create --file environment.yml\nconda activate diffdock\n```\n\n**Option 2: Docker**\n```bash\ndocker pull rbgcsail\u002Fdiffdock\ndocker run -it --gpus all --entrypoint \u002Fbin\u002Fbash rbgcsail\u002Fdiffdock\nmicromamba activate diffdock\n```\n\n**Important Notes:**\n- GPU strongly recommended (10-100x speedup vs CPU)\n- First run pre-computes SO(2)\u002FSO(3) lookup tables (~2-5 minutes)\n- Model checkpoints (~500MB) download automatically if not present\n\n## Core Workflows\n\n### Workflow 1: Single Protein-Ligand Docking\n\n**Use Case:** Dock one ligand to one protein target\n\n**Input Requirements:**\n- Protein: PDB file OR amino acid sequence\n- Ligand: SMILES string OR structure file (SDF\u002FMOL2)\n\n**Command:**\n```bash\npython -m inference \\\n  --config default_inference_args.yaml \\\n  --protein_path protein.pdb \\\n  --ligand \"CC(=O)Oc1ccccc1C(=O)O\" \\\n  --out_dir results\u002Fsingle_docking\u002F\n```\n\n**Alternative (protein sequence):**\n```bash\npython -m inference \\\n  --config default_inference_args.yaml \\\n  --protein_sequence \"MSKGEELFTGVVPILVELDGDVNGHKF...\" \\\n  --ligand ligand.sdf \\\n  --out_dir results\u002Fsequence_docking\u002F\n```\n\n**Output Structure:**\n```\nresults\u002Fsingle_docking\u002F\n├── rank_1.sdf          # Top-ranked pose\n├── rank_2.sdf          # Second-ranked pose\n├── ...\n├── rank_10.sdf         # 10th pose (default: 10 samples)\n└── confidence_scores.txt\n```\n\n### Workflow 2: Batch Processing Multiple Complexes\n\n**Use Case:** Dock multiple ligands to proteins, virtual screening campaigns\n\n**Step 1: Prepare Batch CSV**\n\nUse the provided script to create or validate batch input:\n\n```bash\n# Create template\npython scripts\u002Fprepare_batch_csv.py --create --output batch_input.csv\n\n# Validate existing CSV\npython scripts\u002Fprepare_batch_csv.py my_input.csv --validate\n```\n\n**CSV Format:**\n```csv\ncomplex_name,protein_path,ligand_description,protein_sequence\ncomplex1,protein1.pdb,CC(=O)Oc1ccccc1C(=O)O,\ncomplex2,,COc1ccc(C#N)cc1,MSKGEELFT...\ncomplex3,protein3.pdb,ligand3.sdf,\n```\n\n**Required Columns:**\n- `complex_name`: Unique identifier\n- `protein_path`: PDB file path (leave empty if using sequence)\n- `ligand_description`: SMILES string or ligand file path\n- `protein_sequence`: Amino acid sequence (leave empty if using PDB)\n\n**Step 2: Run Batch Docking**\n\n```bash\npython -m inference \\\n  --config default_inference_args.yaml \\\n  --protein_ligand_csv batch_input.csv \\\n  --out_dir results\u002Fbatch\u002F \\\n  --batch_size 10\n```\n\n**For Large Virtual Screening (>100 compounds):**\n\nPre-compute protein embeddings for faster processing:\n```bash\n# Pre-compute embeddings\npython datasets\u002Fesm_embedding_preparation.py \\\n  --protein_ligand_csv screening_input.csv \\\n  --out_file protein_embeddings.pt\n\n# Run with pre-computed embeddings\npython -m inference \\\n  --config default_inference_args.yaml \\\n  --protein_ligand_csv screening_input.csv \\\n  --esm_embeddings_path protein_embeddings.pt \\\n  --out_dir results\u002Fscreening\u002F\n```\n\n### Workflow 3: Analyzing Results\n\nAfter docking completes, analyze confidence scores and rank predictions:\n\n```bash\n# Analyze all results\npython scripts\u002Fanalyze_results.py results\u002Fbatch\u002F\n\n# Show top 5 per complex\npython scripts\u002Fanalyze_results.py results\u002Fbatch\u002F --top 5\n\n# Filter by confidence threshold\npython scripts\u002Fanalyze_results.py results\u002Fbatch\u002F --threshold 0.0\n\n# Export to CSV\npython scripts\u002Fanalyze_results.py results\u002Fbatch\u002F --export summary.csv\n\n# Show top 20 predictions across all complexes\npython scripts\u002Fanalyze_results.py results\u002Fbatch\u002F --best 20\n```\n\nThe analysis script:\n- Parses confidence scores from all predictions\n- Classifies as High (>0), Moderate (-1.5 to 0), or Low (\u003C-1.5)\n- Ranks predictions within and across complexes\n- Generates statistical summaries\n- Exports results to CSV for downstream analysis\n\n## Confidence Score Interpretation\n\n**Understanding Scores:**\n\n| Score Range | Confidence Level | Interpretation |\n|------------|------------------|----------------|\n| **> 0** | High | Strong prediction, likely accurate |\n| **-1.5 to 0** | Moderate | Reasonable prediction, validate carefully |\n| **\u003C -1.5** | Low | Uncertain prediction, requires validation |\n\n**Critical Notes:**\n1. **Confidence ≠ Affinity**: High confidence means model certainty about structure, NOT strong binding\n2. **Context Matters**: Adjust expectations for:\n   - Large ligands (>500 Da): Lower confidence expected\n   - Multiple protein chains: May decrease confidence\n   - Novel protein families: May underperform\n3. **Multiple Samples**: Review top 3-5 predictions, look for consensus\n\n**For detailed guidance:** Read `references\u002Fconfidence_and_limitations.md` using the Read tool\n\n## Parameter Customization\n\n### Using Custom Configuration\n\nCreate custom configuration for specific use cases:\n\n```bash\n# Copy template\ncp assets\u002Fcustom_inference_config.yaml my_config.yaml\n\n# Edit parameters (see template for presets)\n# Then run with custom config\npython -m inference \\\n  --config my_config.yaml \\\n  --protein_ligand_csv input.csv \\\n  --out_dir results\u002F\n```\n\n### Key Parameters to Adjust\n\n**Sampling Density:**\n- `samples_per_complex: 10` → Increase to 20-40 for difficult cases\n- More samples = better coverage but longer runtime\n\n**Inference Steps:**\n- `inference_steps: 20` → Increase to 25-30 for higher accuracy\n- More steps = potentially better quality but slower\n\n**Temperature Parameters (control diversity):**\n- `temp_sampling_tor: 7.04` → Increase for flexible ligands (8-10)\n- `temp_sampling_tor: 7.04` → Decrease for rigid ligands (5-6)\n- Higher temperature = more diverse poses\n\n**Presets Available in Template:**\n1. High Accuracy: More samples + steps, lower temperature\n2. Fast Screening: Fewer samples, faster\n3. Flexible Ligands: Increased torsion temperature\n4. Rigid Ligands: Decreased torsion temperature\n\n**For complete parameter reference:** Read `references\u002Fparameters_reference.md` using the Read tool\n\n## Advanced Techniques\n\n### Ensemble Docking (Protein Flexibility)\n\nFor proteins with known flexibility, dock to multiple conformations:\n\n```python\n# Create ensemble CSV\nimport pandas as pd\n\nconformations = [\"conf1.pdb\", \"conf2.pdb\", \"conf3.pdb\"]\nligand = \"CC(=O)Oc1ccccc1C(=O)O\"\n\ndata = {\n    \"complex_name\": [f\"ensemble_{i}\" for i in range(len(conformations))],\n    \"protein_path\": conformations,\n    \"ligand_description\": [ligand] * len(conformations),\n    \"protein_sequence\": [\"\"] * len(conformations)\n}\n\npd.DataFrame(data).to_csv(\"ensemble_input.csv\", index=False)\n```\n\nRun docking with increased sampling:\n```bash\npython -m inference \\\n  --config default_inference_args.yaml \\\n  --protein_ligand_csv ensemble_input.csv \\\n  --samples_per_complex 20 \\\n  --out_dir results\u002Fensemble\u002F\n```\n\n### Integration with Scoring Functions\n\nDiffDock generates poses; combine with other tools for affinity:\n\n**GNINA (Fast neural network scoring):**\n```bash\nfor pose in results\u002F*.sdf; do\n    gnina -r protein.pdb -l \"$pose\" --score_only\ndone\n```\n\n**MM\u002FGBSA (More accurate, slower):**\nUse AmberTools MMPBSA.py or gmx_MMPBSA after energy minimization\n\n**Free Energy Calculations (Most accurate):**\nUse OpenMM + OpenFE or GROMACS for FEP\u002FTI calculations\n\n**Recommended Workflow:**\n1. DiffDock → Generate poses with confidence scores\n2. Visual inspection → Check structural plausibility\n3. GNINA or MM\u002FGBSA → Rescore and rank by affinity\n4. Experimental validation → Biochemical assays\n\n## Limitations and Scope\n\n**DiffDock IS Designed For:**\n- Small molecule ligands (typically 100-1000 Da)\n- Drug-like organic compounds\n- Small peptides (\u003C20 residues)\n- Single or multi-chain proteins\n\n**DiffDock IS NOT Designed For:**\n- Large biomolecules (protein-protein docking) → Use DiffDock-PP or AlphaFold-Multimer\n- Large peptides (>20 residues) → Use alternative methods\n- Covalent docking → Use specialized covalent docking tools\n- Binding affinity prediction → Combine with scoring functions\n- Membrane proteins → Not specifically trained, use with caution\n\n**For complete limitations:** Read `references\u002Fconfidence_and_limitations.md` using the Read tool\n\n## Troubleshooting\n\n### Common Issues\n\n**Issue: Low confidence scores across all predictions**\n- Cause: Large\u002Funusual ligands, unclear binding site, protein flexibility\n- Solution: Increase `samples_per_complex` (20-40), try ensemble docking, validate protein structure\n\n**Issue: Out of memory errors**\n- Cause: GPU memory insufficient for batch size\n- Solution: Reduce `--batch_size 2` or process fewer complexes at once\n\n**Issue: Slow performance**\n- Cause: Running on CPU instead of GPU\n- Solution: Verify CUDA with `python -c \"import torch; print(torch.cuda.is_available())\"`, use GPU\n\n**Issue: Unrealistic binding poses**\n- Cause: Poor protein preparation, ligand too large, wrong binding site\n- Solution: Check protein for missing residues, remove far waters, consider specifying binding site\n\n**Issue: \"Module not found\" errors**\n- Cause: Missing dependencies or wrong environment\n- Solution: Run `python scripts\u002Fsetup_check.py` to diagnose\n\n### Performance Optimization\n\n**For Best Results:**\n1. Use GPU (essential for practical use)\n2. Pre-compute ESM embeddings for repeated protein use\n3. Batch process multiple complexes together\n4. Start with default parameters, then tune if needed\n5. Validate protein structures (resolve missing residues)\n6. Use canonical SMILES for ligands\n\n## Graphical User Interface\n\nFor interactive use, launch the web interface:\n\n```bash\npython app\u002Fmain.py\n# Navigate to http:\u002F\u002Flocalhost:7860\n```\n\nOr use the online demo without installation:\n- https:\u002F\u002Fhuggingface.co\u002Fspaces\u002Freginabarzilaygroup\u002FDiffDock-Web\n\n## Resources\n\n### Helper Scripts (`scripts\u002F`)\n\n**`prepare_batch_csv.py`**: Create and validate batch input CSV files\n- Create templates with example entries\n- Validate file paths and SMILES strings\n- Check for required columns and format issues\n\n**`analyze_results.py`**: Analyze confidence scores and rank predictions\n- Parse results from single or batch runs\n- Generate statistical summaries\n- Export to CSV for downstream analysis\n- Identify top predictions across complexes\n\n**`setup_check.py`**: Verify DiffDock environment setup\n- Check Python version and dependencies\n- Verify PyTorch and CUDA availability\n- Test RDKit and PyTorch Geometric installation\n- Provide installation instructions if needed\n\n### Reference Documentation (`references\u002F`)\n\n**`parameters_reference.md`**: Complete parameter documentation\n- All command-line options and configuration parameters\n- Default values and acceptable ranges\n- Temperature parameters for controlling diversity\n- Model checkpoint locations and version flags\n\nRead this file when users need:\n- Detailed parameter explanations\n- Fine-tuning guidance for specific systems\n- Alternative sampling strategies\n\n**`confidence_and_limitations.md`**: Confidence score interpretation and tool limitations\n- Detailed confidence score interpretation\n- When to trust predictions\n- Scope and limitations of DiffDock\n- Integration with complementary tools\n- Troubleshooting prediction quality\n\nRead this file when users need:\n- Help interpreting confidence scores\n- Understanding when NOT to use DiffDock\n- Guidance on combining with other tools\n- Validation strategies\n\n**`workflows_examples.md`**: Comprehensive workflow examples\n- Detailed installation instructions\n- Step-by-step examples for all workflows\n- Advanced integration patterns\n- Troubleshooting common issues\n- Best practices and optimization tips\n\nRead this file when users need:\n- Complete workflow examples with code\n- Integration with GNINA, OpenMM, or other tools\n- Virtual screening workflows\n- Ensemble docking procedures\n\n### Assets (`assets\u002F`)\n\n**`batch_template.csv`**: Template for batch processing\n- Pre-formatted CSV with required columns\n- Example entries showing different input types\n- Ready to customize with actual data\n\n**`custom_inference_config.yaml`**: Configuration template\n- Annotated YAML with all parameters\n- Four preset configurations for common use cases\n- Detailed comments explaining each parameter\n- Ready to customize and use\n\n## Best Practices\n\n1. **Always verify environment** with `setup_check.py` before starting large jobs\n2. **Validate batch CSVs** with `prepare_batch_csv.py` to catch errors early\n3. **Start with defaults** then tune parameters based on system-specific needs\n4. **Generate multiple samples** (10-40) for robust predictions\n5. **Visual inspection** of top poses before downstream analysis\n6. **Combine with scoring** functions for affinity assessment\n7. **Use confidence scores** for initial ranking, not final decisions\n8. **Pre-compute embeddings** for virtual screening campaigns\n9. **Document parameters** used for reproducibility\n10. **Validate results** experimentally when possible\n\n## Citations\n\nWhen using DiffDock, cite the appropriate papers:\n\n**DiffDock-L (current default model):**\n```\nStärk et al. (2024) \"DiffDock-L: Improving Molecular Docking with Diffusion Models\"\narXiv:2402.18396\n```\n\n**Original DiffDock:**\n```\nCorso et al. (2023) \"DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking\"\nICLR 2023, arXiv:2210.01776\n```\n\n## Additional Resources\n\n- **GitHub Repository**: https:\u002F\u002Fgithub.com\u002Fgcorso\u002FDiffDock\n- **Online Demo**: https:\u002F\u002Fhuggingface.co\u002Fspaces\u002Freginabarzilaygroup\u002FDiffDock-Web\n- **DiffDock-L Paper**: https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.18396\n- **Original Paper**: https:\u002F\u002Farxiv.org\u002Fabs\u002F2210.01776\n","# DiffDock: Molecular Docking with Diffusion Models\n\n## Overview\n\nDiffDock is a diffusion-based deep learning tool for molecular docking that predicts 3D binding poses of small molecule ligands to protein targets. It represents the state-of-the-art in computational docking, crucial for structure-based drug discovery and chemical biology.\n\n**Core Capabilities:**\n- Predict ligand binding poses with high accuracy using deep learning\n- Support protein structures (PDB files) or sequences (via ESMFold)\n- Process single complexes or batch virtual screening campaigns\n- Generate confidence scores to assess prediction reliability\n- Handle diverse ligand inputs (SMILES, SDF, MOL2)\n\n**Key Distinction:** DiffDock predicts **binding poses** (3D structure) and **confidence** (prediction certainty), NOT binding affinity (ΔG, Kd). Always combine with scoring functions (GNINA, MM\u002FGBSA) for affinity assessment.\n\n## When to Use This Skill\n\nThis skill should be used when:\n\n- \"Dock this ligand to a protein\" or \"predict binding pose\"\n- \"Run molecular docking\" or \"perform protein-ligand docking\"\n- \"Virtual screening\" or \"screen compound library\"\n- \"Where does this molecule bind?\" or \"predict binding site\"\n- Structure-based drug design or lead optimization tasks\n- Tasks involving PDB files + SMILES strings or ligand structures\n- Batch docking of multiple protein-ligand pairs\n\n## Installation and Environment Setup\n\n### Check Environment Status\n\nBefore proceeding with DiffDock tasks, verify the environment setup:\n\n```bash\n# Use the provided setup checker\npython scripts\u002Fsetup_check.py\n```\n\nThis script validates Python version, PyTorch with CUDA, PyTorch Geometric, RDKit, ESM, and other dependencies.\n\n### Installation Options\n\n**Option 1: Conda (Recommended)**\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fgcorso\u002FDiffDock.git\ncd DiffDock\nconda env create --file environment.yml\nconda activate diffdock\n```\n\n**Option 2: Docker**\n```bash\ndocker pull rbgcsail\u002Fdiffdock\ndocker run -it --gpus all --entrypoint \u002Fbin\u002Fbash rbgcsail\u002Fdiffdock\nmicromamba activate diffdock\n```\n\n**Important Notes:**\n- GPU strongly recommended (10-100x speedup vs CPU)\n- First run pre-computes SO(2)\u002FSO(3) lookup tables (~2-5 minutes)\n- Model checkpoints (~500MB) download automatically if not present\n\n## Core Workflows\n\n### Workflow 1: Single Protein-Ligand Docking\n\n**Use Case:** Dock one ligand to one protein target\n\n**Input Requirements:**\n- Protein: PDB file OR amino acid sequence\n- Ligand: SMILES string OR structure file (SDF\u002FMOL2)\n\n**Command:**\n```bash\npython -m inference \\\n  --config default_inference_args.yaml \\\n  --protein_path protein.pdb \\\n  --ligand \"CC(=O)Oc1ccccc1C(=O)O\" \\\n  --out_dir results\u002Fsingle_docking\u002F\n```\n\n**Alternative (protein sequence):**\n```bash\npython -m inference \\\n  --config default_inference_args.yaml \\\n  --protein_sequence \"MSKGEELFTGVVPILVELDGDVNGHKF...\" \\\n  --ligand ligand.sdf \\\n  --out_dir results\u002Fsequence_docking\u002F\n```\n\n**Output Structure:**\n```\nresults\u002Fsingle_docking\u002F\n├── rank_1.sdf          # Top-ranked pose\n├── rank_2.sdf          # Second-ranked pose\n├── ...\n├── rank_10.sdf         # 10th pose (default: 10 samples)\n└── confidence_scores.txt\n```\n\n### Workflow 2: Batch Processing Multiple Complexes\n\n**Use Case:** Dock multiple ligands to proteins, virtual screening campaigns\n\n**Step 1: Prepare Batch CSV**\n\nUse the provided script to create or validate batch input:\n\n```bash\n# Create template\npython scripts\u002Fprepare_batch_csv.py --create --output batch_input.csv\n\n# Validate existing CSV\npython scripts\u002Fprepare_batch_csv.py my_input.csv --validate\n```\n\n**CSV Format:**\n```csv\ncomplex_name,protein_path,ligand_description,protein_sequence\ncomplex1,protein1.pdb,CC(=O)Oc1ccccc1C(=O)O,\ncomplex2,,COc1ccc(C#N)cc1,MSKGEELFT...\ncomplex3,protein3.pdb,ligand3.sdf,\n```\n\n**Required Columns:**\n- `complex_name`: Unique identifier\n- `protein_path`: PDB file path (leave empty if using sequence)\n- `ligand_description`: SMILES string or ligand file path\n- `protein_sequence`: Amino acid sequence (leave empty if using PDB)\n\n**Step 2: Run Batch Docking**\n\n```bash\npython -m inference \\\n  --config default_inference_args.yaml \\\n  --protein_ligand_csv batch_input.csv \\\n  --out_dir results\u002Fbatch\u002F \\\n  --batch_size 10\n```\n\n**For Large Virtual Screening (>100 compounds):**\n\nPre-compute protein embeddings for faster processing:\n```bash\n# Pre-compute embeddings\npython datasets\u002Fesm_embedding_preparation.py \\\n  --protein_ligand_csv screening_input.csv \\\n  --out_file protein_embeddings.pt\n\n# Run with pre-computed embeddings\npython -m inference \\\n  --config default_inference_args.yaml \\\n  --protein_ligand_csv screening_input.csv \\\n  --esm_embeddings_path protein_embeddings.pt \\\n  --out_dir results\u002Fscreening\u002F\n```\n\n### Workflow 3: Analyzing Results\n\nAfter docking completes, analyze confidence scores and rank predictions:\n\n```bash\n# Analyze all results\npython scripts\u002Fanalyze_results.py results\u002Fbatch\u002F\n\n# Show top 5 per complex\npython scripts\u002Fanalyze_results.py results\u002Fbatch\u002F --top 5\n\n# Filter by confidence threshold\npython scripts\u002Fanalyze_results.py results\u002Fbatch\u002F --threshold 0.0\n\n# Export to CSV\npython scripts\u002Fanalyze_results.py results\u002Fbatch\u002F --export summary.csv\n\n# Show top 20 predictions across all complexes\npython scripts\u002Fanalyze_results.py results\u002Fbatch\u002F --best 20\n```\n\nThe analysis script:\n- Parses confidence scores from all predictions\n- Classifies as High (>0), Moderate (-1.5 to 0), or Low (\u003C-1.5)\n- Ranks predictions within and across complexes\n- Generates statistical summaries\n- Exports results to CSV for downstream analysis\n\n## Confidence Score Interpretation\n\n**Understanding Scores:**\n\n| Score Range | Confidence Level | Interpretation |\n|------------|------------------|----------------|\n| **> 0** | High | Strong prediction, likely accurate |\n| **-1.5 to 0** | Moderate | Reasonable prediction, validate carefully |\n| **\u003C -1.5** | Low | Uncertain prediction, requires validation |\n\n**Critical Notes:**\n1. **Confidence ≠ Affinity**: High confidence means model certainty about structure, NOT strong binding\n2. **Context Matters**: Adjust expectations for:\n   - Large ligands (>500 Da): Lower confidence expected\n   - Multiple protein chains: May decrease confidence\n   - Novel protein families: May underperform\n3. **Multiple Samples**: Review top 3-5 predictions, look for consensus\n\n**For detailed guidance:** Read `references\u002Fconfidence_and_limitations.md` using the Read tool\n\n## Parameter Customization\n\n### Using Custom Configuration\n\nCreate custom configuration for specific use cases:\n\n```bash\n# Copy template\ncp assets\u002Fcustom_inference_config.yaml my_config.yaml\n\n# Edit parameters (see template for presets)\n# Then run with custom config\npython -m inference \\\n  --config my_config.yaml \\\n  --protein_ligand_csv input.csv \\\n  --out_dir results\u002F\n```\n\n### Key Parameters to Adjust\n\n**Sampling Density:**\n- `samples_per_complex: 10` → Increase to 20-40 for difficult cases\n- More samples = better coverage but longer runtime\n\n**Inference Steps:**\n- `inference_steps: 20` → Increase to 25-30 for higher accuracy\n- More steps = potentially better quality but slower\n\n**Temperature Parameters (control diversity):**\n- `temp_sampling_tor: 7.04` → Increase for flexible ligands (8-10)\n- `temp_sampling_tor: 7.04` → Decrease for rigid ligands (5-6)\n- Higher temperature = more diverse poses\n\n**Presets Available in Template:**\n1. High Accuracy: More samples + steps, lower temperature\n2. Fast Screening: Fewer samples, faster\n3. Flexible Ligands: Increased torsion temperature\n4. Rigid Ligands: Decreased torsion temperature\n\n**For complete parameter reference:** Read `references\u002Fparameters_reference.md` using the Read tool\n\n## Advanced Techniques\n\n### Ensemble Docking (Protein Flexibility)\n\nFor proteins with known flexibility, dock to multiple conformations:\n\n```python\n# Create ensemble CSV\nimport pandas as pd\n\nconformations = [\"conf1.pdb\", \"conf2.pdb\", \"conf3.pdb\"]\nligand = \"CC(=O)Oc1ccccc1C(=O)O\"\n\ndata = {\n    \"complex_name\": [f\"ensemble_{i}\" for i in range(len(conformations))],\n    \"protein_path\": conformations,\n    \"ligand_description\": [ligand] * len(conformations),\n    \"protein_sequence\": [\"\"] * len(conformations)\n}\n\npd.DataFrame(data).to_csv(\"ensemble_input.csv\", index=False)\n```\n\nRun docking with increased sampling:\n```bash\npython -m inference \\\n  --config default_inference_args.yaml \\\n  --protein_ligand_csv ensemble_input.csv \\\n  --samples_per_complex 20 \\\n  --out_dir results\u002Fensemble\u002F\n```\n\n### Integration with Scoring Functions\n\nDiffDock generates poses; combine with other tools for affinity:\n\n**GNINA (Fast neural network scoring):**\n```bash\nfor pose in results\u002F*.sdf; do\n    gnina -r protein.pdb -l \"$pose\" --score_only\ndone\n```\n\n**MM\u002FGBSA (More accurate, slower):**\nUse AmberTools MMPBSA.py or gmx_MMPBSA after energy minimization\n\n**Free Energy Calculations (Most accurate):**\nUse OpenMM + OpenFE or GROMACS for FEP\u002FTI calculations\n\n**Recommended Workflow:**\n1. DiffDock → Generate poses with confidence scores\n2. Visual inspection → Check structural plausibility\n3. GNINA or MM\u002FGBSA → Rescore and rank by affinity\n4. Experimental validation → Biochemical assays\n\n## Limitations and Scope\n\n**DiffDock IS Designed For:**\n- Small molecule ligands (typically 100-1000 Da)\n- Drug-like organic compounds\n- Small peptides (\u003C20 residues)\n- Single or multi-chain proteins\n\n**DiffDock IS NOT Designed For:**\n- Large biomolecules (protein-protein docking) → Use DiffDock-PP or AlphaFold-Multimer\n- Large peptides (>20 residues) → Use alternative methods\n- Covalent docking → Use specialized covalent docking tools\n- Binding affinity prediction → Combine with scoring functions\n- Membrane proteins → Not specifically trained, use with caution\n\n**For complete limitations:** Read `references\u002Fconfidence_and_limitations.md` using the Read tool\n\n## Troubleshooting\n\n### Common Issues\n\n**Issue: Low confidence scores across all predictions**\n- Cause: Large\u002Funusual ligands, unclear binding site, protein flexibility\n- Solution: Increase `samples_per_complex` (20-40), try ensemble docking, validate protein structure\n\n**Issue: Out of memory errors**\n- Cause: GPU memory insufficient for batch size\n- Solution: Reduce `--batch_size 2` or process fewer complexes at once\n\n**Issue: Slow performance**\n- Cause: Running on CPU instead of GPU\n- Solution: Verify CUDA with `python -c \"import torch; print(torch.cuda.is_available())\"`, use GPU\n\n**Issue: Unrealistic binding poses**\n- Cause: Poor protein preparation, ligand too large, wrong binding site\n- Solution: Check protein for missing residues, remove far waters, consider specifying binding site\n\n**Issue: \"Module not found\" errors**\n- Cause: Missing dependencies or wrong environment\n- Solution: Run `python scripts\u002Fsetup_check.py` to diagnose\n\n### Performance Optimization\n\n**For Best Results:**\n1. Use GPU (essential for practical use)\n2. Pre-compute ESM embeddings for repeated protein use\n3. Batch process multiple complexes together\n4. Start with default parameters, then tune if needed\n5. Validate protein structures (resolve missing residues)\n6. Use canonical SMILES for ligands\n\n## Graphical User Interface\n\nFor interactive use, launch the web interface:\n\n```bash\npython app\u002Fmain.py\n# Navigate to http:\u002F\u002Flocalhost:7860\n```\n\nOr use the online demo without installation:\n- https:\u002F\u002Fhuggingface.co\u002Fspaces\u002Freginabarzilaygroup\u002FDiffDock-Web\n\n## Resources\n\n### Helper Scripts (`scripts\u002F`)\n\n**`prepare_batch_csv.py`**: Create and validate batch input CSV files\n- Create templates with example entries\n- Validate file paths and SMILES strings\n- Check for required columns and format issues\n\n**`analyze_results.py`**: Analyze confidence scores and rank predictions\n- Parse results from single or batch runs\n- Generate statistical summaries\n- Export to CSV for downstream analysis\n- Identify top predictions across complexes\n\n**`setup_check.py`**: Verify DiffDock environment setup\n- Check Python version and dependencies\n- Verify PyTorch and CUDA availability\n- Test RDKit and PyTorch Geometric installation\n- Provide installation instructions if needed\n\n### Reference Documentation (`references\u002F`)\n\n**`parameters_reference.md`**: Complete parameter documentation\n- All command-line options and configuration parameters\n- Default values and acceptable ranges\n- Temperature parameters for controlling diversity\n- Model checkpoint locations and version flags\n\nRead this file when users need:\n- Detailed parameter explanations\n- Fine-tuning guidance for specific systems\n- Alternative sampling strategies\n\n**`confidence_and_limitations.md`**: Confidence score interpretation and tool limitations\n- Detailed confidence score interpretation\n- When to trust predictions\n- Scope and limitations of DiffDock\n- Integration with complementary tools\n- Troubleshooting prediction quality\n\nRead this file when users need:\n- Help interpreting confidence scores\n- Understanding when NOT to use DiffDock\n- Guidance on combining with other tools\n- Validation strategies\n\n**`workflows_examples.md`**: Comprehensive workflow examples\n- Detailed installation instructions\n- Step-by-step examples for all workflows\n- Advanced integration patterns\n- Troubleshooting common issues\n- Best practices and optimization tips\n\nRead this file when users need:\n- Complete workflow examples with code\n- Integration with GNINA, OpenMM, or other tools\n- Virtual screening workflows\n- Ensemble docking procedures\n\n### Assets (`assets\u002F`)\n\n**`batch_template.csv`**: Template for batch processing\n- Pre-formatted CSV with required columns\n- Example entries showing different input types\n- Ready to customize with actual data\n\n**`custom_inference_config.yaml`**: Configuration template\n- Annotated YAML with all parameters\n- Four preset configurations for common use cases\n- Detailed comments explaining each parameter\n- Ready to customize and use\n\n## Best Practices\n\n1. **Always verify environment** with `setup_check.py` before starting large jobs\n2. **Validate batch CSVs** with `prepare_batch_csv.py` to catch errors early\n3. **Start with defaults** then tune parameters based on system-specific needs\n4. **Generate multiple samples** (10-40) for robust predictions\n5. **Visual inspection** of top poses before downstream analysis\n6. **Combine with scoring** functions for affinity assessment\n7. **Use confidence scores** for initial ranking, not final decisions\n8. **Pre-compute embeddings** for virtual screening campaigns\n9. **Document parameters** used for reproducibility\n10. **Validate results** experimentally when possible\n\n## Citations\n\nWhen using DiffDock, cite the appropriate papers:\n\n**DiffDock-L (current default model):**\n```\nStärk et al. (2024) \"DiffDock-L: Improving Molecular Docking with Diffusion Models\"\narXiv:2402.18396\n```\n\n**Original DiffDock:**\n```\nCorso et al. (2023) \"DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking\"\nICLR 2023, arXiv:2210.01776\n```\n\n## Additional Resources\n\n- **GitHub Repository**: https:\u002F\u002Fgithub.com\u002Fgcorso\u002FDiffDock\n- **Online Demo**: https:\u002F\u002Fhuggingface.co\u002Fspaces\u002Freginabarzilaygroup\u002FDiffDock-Web\n- **DiffDock-L Paper**: https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.18396\n- **Original Paper**: https:\u002F\u002Farxiv.org\u002Fabs\u002F2210.01776\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\u002Fdiffdock",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 diffdock",{"tool":39,"command":40},"bunx","bunx skills add YYC-Cube\u002FYYC3-Knowledge-Base --skill diffdock",{"tool":42,"command":43},"pnpm","pnpm dlx skills add YYC-Cube\u002FYYC3-Knowledge-Base --skill diffdock","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\u002Fdiffdock\u002FSKILL.md",true,[47,53,57,63,67,71,77,81],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"assets\u002Fbatch_template.csv","ASSET","资产文件",410,"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\u002Fdiffdock\u002Fassets\u002Fbatch_template.csv",{"path":54,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":55,"sourceHtmlUrl":56},"assets\u002Fcustom_inference_config.yaml",2846,"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\u002Fdiffdock\u002Fassets\u002Fcustom_inference_config.yaml",{"path":58,"resourceType":59,"resourceTypeDesc":60,"sizeBytes":61,"sourceHtmlUrl":62},"references\u002Fconfidence_and_limitations.md","REFERENCE","参考文档",6900,"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\u002Fdiffdock\u002Freferences\u002Fconfidence_and_limitations.md",{"path":64,"resourceType":59,"resourceTypeDesc":60,"sizeBytes":65,"sourceHtmlUrl":66},"references\u002Fparameters_reference.md",5072,"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\u002Fdiffdock\u002Freferences\u002Fparameters_reference.md",{"path":68,"resourceType":59,"resourceTypeDesc":60,"sizeBytes":69,"sourceHtmlUrl":70},"references\u002Fworkflows_examples.md",10520,"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\u002Fdiffdock\u002Freferences\u002Fworkflows_examples.md",{"path":72,"resourceType":73,"resourceTypeDesc":74,"sizeBytes":75,"sourceHtmlUrl":76},"scripts\u002Fanalyze_results.py","SCRIPT","脚本",11036,"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\u002Fdiffdock\u002Fscripts\u002Fanalyze_results.py",{"path":78,"resourceType":73,"resourceTypeDesc":74,"sizeBytes":79,"sourceHtmlUrl":80},"scripts\u002Fprepare_batch_csv.py",8730,"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\u002Fdiffdock\u002Fscripts\u002Fprepare_batch_csv.py",{"path":82,"resourceType":73,"resourceTypeDesc":74,"sizeBytes":83,"sourceHtmlUrl":84},"scripts\u002Fsetup_check.py",8016,"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\u002Fdiffdock\u002Fscripts\u002Fsetup_check.py",[86,93,100,107,114,121],{"skillNo":87,"name":88,"nameZh":6,"description":89,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"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":94,"name":95,"nameZh":6,"description":96,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":97,"categories":98,"keywords":99,"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":101,"name":102,"nameZh":6,"description":103,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":104,"categories":105,"keywords":106,"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":108,"name":109,"nameZh":6,"description":110,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":111,"categories":112,"keywords":113,"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":115,"name":116,"nameZh":6,"description":117,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":118,"categories":119,"keywords":120,"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":122,"name":123,"nameZh":6,"description":124,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":125,"categories":126,"keywords":127,"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",[],[]]