What GOBSMACKED does
GOBSMACKED folds, docks, relaxes and annotates a protein-ligand complex, then checks the result against the experimental structure. The heavy compute happens on your own GPU, in a bundle this server writes; the annotation and the verification happen here.
Why the superposition is on pocket Ca atoms
A predicted model can be excellent at the binding site and several angstroms out at a disordered terminus. Superposing whole chains spreads that error into the pocket and inflates the ligand RMSD, which is the one number this app exists to report honestly. Everything geometric here is measured after fitting the residues within 8 A of the reference ligand, and the whole-chain TM-score is reported separately as context.
What the grades mean
| Metric | A | B | C | D | F |
|---|---|---|---|---|---|
| Ligand RMSD, best of pose 1 and MD-final | ≤ 1.0 A | ≤ 2.0 | ≤ 3.0 | ≤ 4.0 | > 4.0 |
| Pocket Ca RMSD, MD-final | ≤ 0.8 | ≤ 1.2 | ≤ 1.8 | ≤ 2.5 | > 2.5 |
| chi1 agreement, MD-final | ≥ 0.85 | ≥ 0.70 | ≥ 0.55 | ≥ 0.40 | < 0.40 |
| PLIP Jaccard, best of pose 1 and MD-final | ≥ 0.75 | ≥ 0.55 | ≥ 0.40 | ≥ 0.25 | < 0.25 |
| MD stability: ligand drift, last 200 ps minus first 200 ps | ≤ 0.5 A | ≤ 1.0 | ≤ 1.5 | ≤ 2.5 | > 2.5 |
| Rescue: pocket Ca RMSD before MD minus after | ≥ +0.5 A | ≥ +0.2 | ≥ 0 | ≥ −0.3 | < −0.3 |
| Pose validity (clashes, bond lengths, chirality, box) | pass | fail | |||
The composite GOBSMACK score is the weighted mean of those grades (ligand RMSD 30, PLIP overlap 20, pocket Ca 15, chi1 10, stability 10, rescue 10, validity 5). A metric that could not be measured drops out and the remaining weights are renormalised, which is stated on the card. A validity failure caps the composite at 40. A run with no reference structure gets no composite at all: there is nothing to verify it against.
Software and references
Prepare
| Tool | What it does here | Licence | Links |
|---|---|---|---|
| UniProt | Canonical sequence, features, protein naming | CC BY 4.0 | repository · The UniProt Consortium 2023, Nucleic Acids Research |
| InterPro / Pfam | Domain assignment, and the family router that follows from it | CC0 | repository · Paysan-Lafosse et al. 2023, Nucleic Acids Research |
| KLIFS | 85-residue kinase pocket numbering, DFG and alphaC states | open, attribution | repository · Kanev et al. 2021, Nucleic Acids Research |
| GPCRdb | Ballesteros-Weinstein generic numbering, segments, microswitches | Apache-2.0 | repository · Pandy-Szekeres et al. 2023, Nucleic Acids Research |
| RCSB PDB | Reference structure search, entry metadata, coordinates | CC0 | repository · Burley et al. 2023, Nucleic Acids Research |
| AlphaFold DB | Predicted starting structures and their pLDDT | CC BY 4.0 | repository · Varadi et al. 2024, Nucleic Acids Research |
| ESM Metagenomic Atlas | Folding short sequences without a GPU | MIT | repository · Lin et al. 2023, Science |
| RDKit | SMILES validation, depiction, fingerprints, symmetry-aware RMSD | BSD-3-Clause | repository · rdkit.org |
Run
| Tool | What it does here | Licence | Links |
|---|---|---|---|
| ESMFold (esm) | Folding the sequence when no model exists | MIT | repository · Lin et al. 2023, Science |
| OpenFold | ESMFold's structure module | Apache-2.0 | repository · Ahdritz et al. 2024, Nature Methods |
| PDBFixer | Missing atoms, protonation, cleaning before MD | MIT | repository · The OpenMM project |
| PandaDock | Pose search and SE(3) GNN rescoring | MIT | repository · Panda 2026, bioRxiv |
| OpenMM | Minimisation, equilibration and production MD | MIT | repository · Eastman et al. 2024, Journal of Physical Chemistry B (OpenMM 8) |
| openmmforcefields / OpenFF | Ligand parameters (OpenFF 2.1.0) alongside Amber14 | MIT | repository · Boothroyd et al. 2023, Journal of Chemical Theory and Computation |
| MDTraj | Trajectory analysis, in the bundle and on the server | LGPL-2.1 | repository · McGibbon et al. 2015, Biophysical Journal |
Analyze
| Tool | What it does here | Licence | Links |
|---|---|---|---|
| PLIP | Interaction fingerprints, run as a subprocess only | GPL-2.0 | repository · Adasme et al. 2021, Nucleic Acids Research |
| PandaMap | 2D interaction maps and an empirical binding-energy estimate | MIT | repository · Panda 2025, PandaMap |
| biotite | Sequence alignment behind every residue-number mapping | BSD-3-Clause | repository · Kunzmann and Hamacher 2018, BMC Bioinformatics |
| tmtools / TM-align | Whole-chain TM-score, reported as context for the pocket fit | MIT | repository · Zhang and Skolnick 2005, Nucleic Acids Research |
| PoseBusters (criteria) | The pose-validity checks, reimplemented on this app's own geometry | BSD-3-Clause | repository · Buttenschoen et al. 2024, Chemical Science |
| gemmi | Structure input and output, mmCIF and PDB | MPL-2.0 | repository |
Browser
| Tool | What it does here | Licence | Links |
|---|---|---|---|
| Mol* | Every 3D view, including the three-way overlay | MIT | repository · Sehnal et al. 2021, Nucleic Acids Research |
| Plotly.js | Trajectory panels | MIT | repository · plotly.com |
Hosting
| Tool | What it does here | Licence | Links |
|---|---|---|---|
| Flask, gunicorn, nginx, SQLite | The web application and its store | BSD / MIT / public domain | repository |
| pixi and uv | The bundle's environment, and the server's | BSD-3-Clause / Apache-2.0 | repository |
Credits and licence
GOBSMACKED 1.0.0 was written by Marc C. Deller, D.Phil. (marc@marcdeller.com, marcdeller.com). Source and issues: github.com/bellcheddar/GOBSMACKED.
The application is MIT licensed. To cite it, name the repository and the version.
PLIP is GPL-2.0. GOBSMACKED runs it as a subprocess and never imports it, so no GPL code is linked into this application, and PLIP is deliberately absent from the run bundle. Every other component's licence is in the table above.