Loading BioAtlas...
Loading BioAtlas...
From target pocket characterization to billions of compounds screened with geometric deep learning.
Compare leading AI molecular docking tools, equivariant diffusion models, and generative chemistry platforms. Benchmark DiffDock, EquiBind, GNINA, and Chemistry42 across docking accuracy, screening throughput, and chemical validity.
The previous fixed-template workflow scorer cannot establish comparative scientific suitability. For a task-specific assessment, enter the biological objective, the actual sample/data modalities and method constraints in the offline Decision Lab. Select the reviewed method deliberately and compare it against a mandatory non-AI baseline.
Open scientific Decision LabHistorical workflow templates are examples, not validated executable pipelines, clinical decisions or ranked model recommendations.
| 工具与机构 | 许可与条款 | GPU / 硬件 | 架构 | 输入格式 | 实证等级 | 目录 |
|---|---|---|---|---|---|---|
| DiffDock MIT CSAIL / Corso et al. | MIT (Open Source) | 1x CUDA GPU (RTX 3090 / A100) | Open Source / Open Weights | PDB receptor + SMILES ligand | Peer-reviewed (ICLR 2023) | 查看工具 → |
| EquiBind Stärk et al. | MIT (Open Source) | CPU or fast GPU | Open Source | PDB receptor + SDF / Mol2 ligand | Peer-reviewed (ICML 2022) | 查看工具 → |
| GNINA University of Pittsburgh | GPL v2 (Open Source) | CPU or NVIDIA GPU (CUDA) | Open Source | PDB receptor + Ligand PDBQT | Peer-reviewed (J. Chem. Inf. Model.) | 查看工具 → |
| Chemistry42 Insilico Medicine | Commercial Enterprise SaaS | Cloud GPU Cluster (Managed) | Proprietary Software | Receptor PDB + Pharmacophore constraints | Clinically validated preclinical pipelines | 查看工具 → |
| Boltz-2 Biomolecular AI Consortium | MIT (Open Weights) | 1x A100 80GB | Open Source / Open Weights | Receptor PDB + Small-Molecule CIF/SMILES | Preprint (bioRxiv 2024) | 查看工具 → |
| Chai-1 Chai Discovery | Chai Terms | A100 GPU / Cloud API | Open Weights | Target sequence + SMILES | Preprint (bioRxiv 2024) | 查看工具 → |
选择适合您的算力基础设施、预算约束以及许可自由度的运行架构。
Ideal for academic chemistry departments and research groups screening chemical collections (ZINC20, PubChem) on university clusters.
Rapidly identify putative binding pockets without prior co-crystallized ligand guidance.
Refine binding poses and calculate 3D CNN binding affinity scores (CNNscore and CNNaffinity).
Filter docked hits for Lipinski Rule of 5, Veber parameters, PAINS alerts, and predicted synthetic accessibility.
Select top 20–50 commercially available hit compounds for microscale thermophoresis (MST) or SPR validation.
Built for biotech companies driving hit-to-lead campaigns across the Enamine REAL space (10B+ make-on-demand molecules).
Generate novel chemotypes with desired novelty, metabolic stability, and binding selectivity.
Validate induced-fit pocket adaptation and non-covalent interactions using all-atom foundation models.
Screen hit analogs across Enamine REAL 10-billion space using GPU-accelerated diffusion docking pipelines.
Automate direct PO creation with custom synthesis CROs for 4-week turnaround in vitro profiling.
在 BioAtlas 交互式对比表中查看详细的参数规模、基准得分、输入限制和许可限制。
目前没有与这些特定关键词匹配的有效职位。请查看主招聘板块了解最新职位。
查看全部 1,000+ 个已验证的 BioAI 职位 →