Model type
Molecular docking model; this record is the paper-specific evaluated configuration.
Pipeline
DiffDock-NMDN
DiffDock-NMDN combines diffusion-based pose generation with a learned pose-selection and affinity-scoring pipeline.
Sources
Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening · DiffDock-NMDN Blind Docking and Virtual Screening Protocol (paragraph 1); Abstract (paragraph 1)How it worksEvaluated procedure (conceptual)
Conceptual input–method–output guide. Check the procedure text and linked evaluation for fitted components, additional inputs and exact settings.
Sources
Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening · Methods/Ligand Conformation Stability and Solvation Energetics (paragraph 1); Methods/Normalized Mixture Density Network Module/Training Loss for the NMDN Modules (paragraph 2)Overview
Biological inputs
Protein structure/sequence and ligand information
Sources
Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening · Methods/Interaction Module/Calculation of Protein–Ligand Pair Contributions (paragraph 8); Methods/Normalized Mixture Density Network Module/Inference (paragraph 1)Outputs
Selected binding poses and estimated binding affinities
Sources
Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening · Results and Discussion/DiffDock-NMDN Blind Docking Protocol (paragraph 4); Data Sets/Evaluation Metrics/CASF-2016 (paragraph 2)Access
Official upstream implementation and usage documentation: https://github.com/gcorso/DiffDock/blob/85c49b60d3e0b0182a59ee43a34a6d7036981284/README.md. This pinned documentation revision is not automatically the evaluated weight revision.
Sources
gcorso/DiffDock README.md · README.md; installation, model download and usage instructionslimited source coverage · Automated source review, 2026-09-16. All specifications and missing details
Evaluations and results
1 evaluation · 1 result. Different protocols are not a single leaderboard.
Filter evaluations
Applied filters: All linked evaluations
| Tested configuration | Protocol and dataset | Finding | Evidence and details |
|---|---|---|---|
| Pipeline: DiffDock-NMDN | Task: Protein–ligand virtual screening Dataset: CASF-2016 blind docked poses | 66.7 Forward-screening success rate % · unknown Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceDiffDock-NMDN: Protein–ligand virtual screening NMDN scoring on DiffDock-NMDN blind docked poses; not ligand-pose RMSD. Aggregation: Not reported Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening · Table 2, DiffDock-NMDN / NMDN row, success rate (%) column |
Source checking is not independent reproduction. Release 2026-09-29-06401fd5b220.
Use this model
How it works, versions and access
How it works
How the evaluated method works
DiffDock samples candidate poses. A normalised mixture-density network scores protein-residue/ligand-atom distances and selects a pose; an additional interaction module estimates affinity. The protein encoder uses ESM-2 650M.
Sources
Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening · Methods/Ligand Conformation Stability and Solvation Energetics (paragraph 1); Methods/Normalized Mixture Density Network Module/Training Loss for the NMDN Modules (paragraph 2)Underlying method and version boundaries
DiffDock is a molecular-docking implementation that produces ligand poses and confidence estimates. Its confidence values and predicted coordinates are different outputs from an experimentally calibrated binding-affinity measurement.
Sources
gcorso/DiffDock README.md · README.md; introduction, model description, pretrained-model and usage sections at pinned revisionWhat was evaluated
The linked evaluation record identifies DiffDock-NMDN: Protein–ligand virtual screening. Its dataset, split, adaptation and evidence origin remain attached to the reported results.
Sources
Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening · The named evaluation’s methods and comparison table; exact preserved evaluation IDs: evaluation-lit-043Strengths, limitations and unresolved questions
Strengths and limitations
Strengths and considerations
- Separates pose sampling, pose selection and affinity estimation explicitly.
Sources
Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening · Methods/Ligand Conformation Stability and Solvation Energetics (paragraph 1); Methods/Interaction Module/Calculation of Protein–Ligand Pair Contributions (paragraph 16)
Limitations and conditions
- A DiffDock confidence score is not itself binding affinity; the complete NMDN pipeline must be identified when reporting screening results.
Sources
Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening · Methods/Normalized Mixture Density Network Module/Inference (paragraph 6); Methods/Interaction Module (paragraph 1)
Profile review details
Primary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction.
Stable record: reported-model-6c0bc8d297cc7aSpecifications
Inputs, training, access and other details
Explanatory profile: limited source coverage · Automated source review, 2026-09-16. Review applies to the cited claims; unresolved fields are listed below. Numerical results retain their own review status.
| Property | Description and evidence |
|---|---|
| Model type | Molecular docking model; this record is the paper-specific evaluated configuration.Sourcesgcorso/DiffDock README.md · README.md model description |
| Architecture / procedure | DiffDock samples candidate poses. A normalised mixture-density network scores protein-residue/ligand-atom distances and selects a pose; an additional interaction module estimates affinity. The protein encoder uses ESM-2 650M.SourcesNormalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening · Methods/Ligand Conformation Stability and Solvation Energetics (paragraph 1); Methods/Normalized Mixture Density Network Module/Training Loss for the NMDN Modules (paragraph 2) |
| Biological inputs | Protein structure/sequence and ligand informationSourcesNormalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening · Methods/Interaction Module/Calculation of Protein–Ligand Pair Contributions (paragraph 8); Methods/Normalized Mixture Density Network Module/Inference (paragraph 1) |
| Outputs | Selected binding poses and estimated binding affinitiesSourcesNormalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening · Results and Discussion/DiffDock-NMDN Blind Docking Protocol (paragraph 4); Data Sets/Evaluation Metrics/CASF-2016 (paragraph 2) |
| Parameters | 650-million-parameter ESM-2 protein encoder; this is not the total pipeline size.SourcesNormalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening · Methods/Encoders for Protein, Ligand, and Metal (paragraph 1); Methods/Normalized Mixture Density Network Module/Inference (paragraph 1) |
| Known versions / configuration | DiffDock-NMDN is the comparison-table label; that label does not specify an immutable weight revision. · Not reported in inspected sourcesSourcesNormalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening · Model identification in the comparison table and corresponding Methods; immutable checkpoint revision is not supplied by the table label. |
| Training data / fitting | NMDN uses PDBbind 2020 with 12,554 training and 1,083 evaluation complexes from the first RTMScore split. An additional affinity-module fine-tuning stage draws weak binders from EquiVS and Papyrus: 60,000 pairs are sampled from 250,267 eligible pairs. The pretrained DiffDock pose generator is a separate component.SourcesNormalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening · Data Sets / Training Data Set Preparation / Binder Data Set and Weak-Binder Data Set |
| Context limits | A maximum input/context length for this exact evaluated configuration is not established by the inspected sources. · Not reported in inspected sourcesSources (2)Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening; gcorso/DiffDock README.md · Methods/Model Overview; Methods/Encoders for Protein, Ligand, and Metal; Methods/Normalized Mixture Density Network Module; Methods/Normalized Mixture Density Network Module/Protein–ligand NMDN Module Architecture; Methods/Normalized Mixture Density Network Module/Metal–Ligand NMDN Module; Methods/Normalized Mixture Density Network Module/Training Loss for the NMDN Modules; Methods/Normalized Mixture Density Network Module/Inference; Methods/Interaction Module; inspected for explicit maximum input length (dataset lengths and family-wide limits are not substituted); README.md at pinned repository revision |
| Access | Official upstream implementation and usage documentation: https://github.com/gcorso/DiffDock/blob/85c49b60d3e0b0182a59ee43a34a6d7036981284/README.md. This pinned documentation revision is not automatically the evaluated weight revision.Sourcesgcorso/DiffDock README.md · README.md; installation, model download and usage instructions |
| Code licence | MIT (upstream repository code at the cited revision; this does not establish every dependency or historical checkpoint licence).Sourcesgcorso/DiffDock LICENSE · LICENSE; complete licence text |
| Weights licence | The inspected model-access documentation does not explicitly identify terms for this exact evaluated checkpoint or fitted head; repository code terms are shown separately. · Not reported in inspected sourcesSourcesgcorso/DiffDock README.md · README.md; checkpoint/access documentation and licence scope |
Evidence
Source checking verifies the cited claim or transcription. It does not establish independent reproduction.
Evidence table
Inspect claims, sources and review details
Trace each statement to its source and review. A context-only reference supports the record generally; it does not verify an individual field. Source checking does not reproduce an experiment.
One row per statement and cited source. Multiple citations are not independent evaluations. Shared locators are labelled explicitly.
21 evidence rows matching the loaded filters
| Property and statement | Original source and location | Review and provenance |
|---|---|---|
| Diagram caption Conceptual input–method–output guide. Check the procedure text and linked evaluation for fitted components, additional inputs and exact settings. Individual claims | Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening Methods/Ligand Conformation Stability and Solvation Energetics (paragraph 1); Methods/Normalized Mixture Density Network Module/Training Loss for the NMDN Modules (paragraph 2) Version: version of record | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
Diagram steps
| Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening Methods/Ligand Conformation Stability and Solvation Energetics (paragraph 1); Methods/Normalized Mixture Density Network Module/Training Loss for the NMDN Modules (paragraph 2) Version: version of record | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Diagram title Evaluated procedure (conceptual) Individual claims | Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening Methods/Ligand Conformation Stability and Solvation Energetics (paragraph 1); Methods/Normalized Mixture Density Network Module/Training Loss for the NMDN Modules (paragraph 2) Version: version of record | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Model type Molecular docking model; this record is the paper-specific evaluated configuration. Individual claims | gcorso/DiffDock README.md README.md model description Version: 85c49b60d3e0b0182a59ee43a34a6d7036981284 | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Architecture / procedure DiffDock samples candidate poses. A normalised mixture-density network scores protein-residue/ligand-atom distances and selects a pose; an additional interaction module estimates affinity. The protein encoder uses ESM-2 650M. Individual claims | Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening Methods/Ligand Conformation Stability and Solvation Energetics (paragraph 1); Methods/Normalized Mixture Density Network Module/Training Loss for the NMDN Modules (paragraph 2) Version: version of record | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Weights licence The inspected model-access documentation does not explicitly identify terms for this exact evaluated checkpoint or fitted head; repository code terms are shown separately. Individual claims | gcorso/DiffDock README.md README.md; checkpoint/access documentation and licence scope Version: 85c49b60d3e0b0182a59ee43a34a6d7036981284 | unreported automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Biological inputs Protein structure/sequence and ligand information Individual claims | Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening Methods/Interaction Module/Calculation of Protein–Ligand Pair Contributions (paragraph 8); Methods/Normalized Mixture Density Network Module/Inference (paragraph 1) Version: version of record | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Outputs Selected binding poses and estimated binding affinities Individual claims | Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening Results and Discussion/DiffDock-NMDN Blind Docking Protocol (paragraph 4); Data Sets/Evaluation Metrics/CASF-2016 (paragraph 2) Version: version of record | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Parameters 650-million-parameter ESM-2 protein encoder; this is not the total pipeline size. Individual claims | Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening Methods/Encoders for Protein, Ligand, and Metal (paragraph 1); Methods/Normalized Mixture Density Network Module/Inference (paragraph 1) Version: version of record | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Known versions / configuration DiffDock-NMDN is the comparison-table label; that label does not specify an immutable weight revision. Individual claims | Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening Model identification in the comparison table and corresponding Methods; immutable checkpoint revision is not supplied by the table label. Version: version of record | unreported automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
Sources and history
View linked audit checks and correction history
Release 2026-09-29-06401fd5b220 · Record review: needs review
3 source records and release history
- Normalized Protein–Ligand Distance Likelihood Score for End-to-End Blind Docking and Virtual Screening · Original source · version of record
- gcorso/DiffDock README.md · Original source · 85c49b60d3e0b0182a59ee43a34a6d7036981284
- gcorso/DiffDock LICENSE · Original source · 85c49b60d3e0b0182a59ee43a34a6d7036981284
Technical metadata and extraction receipts
Stable ID: reported-model-6c0bc8d297cc7a
- areas
- molecular-interactions
- entity level
- method
- version
- Not reported
- reported name
- DiffDock-NMDN
- historical missing metadata
- version: not_reported_in_legacy_extract; checkpoint revision: not_reported_in_legacy_extract; training data: not_reported_in_legacy_extract; licence: not_reported_in_legacy_extract
- metadata review scope
- historical_missing_metadata preserves the original discovery state. Current descriptive evidence and missingness are recorded in profile.facts; numerical-result review is separate.
- legacy kinds
- model
- entity classification
- review date: 2026-09-17; rationale: This record identifies a composed analysis workflow with separately identifiable upstream models, representations or tools and a downstream prediction/scoring procedure. Results belong to that complete composition rather than to an upstream model alone.; source ids: nmdn-2025; evidence-reported-base-diffdock-readme-md; source locator: Methods/Ligand Conformation Stability and Solvation Energetics (paragraph 1); Methods/Normalized Mixture Density Network Module/Training Loss for the NMDN Modules (paragraph 2) | README.md model description | DiffDock-NMDN Blind Docking and Virtual Screening Protocol (paragraph 1); Abstract (paragraph 1); ambiguities: This is the paper-specific pipeline identity; unspecified component checkpoints or implementation versions are not inferred from its name.