- Home
- Skills
- APIs & Backend
- PHM MolCheck
PHM MolCheck
Deterministic physical-validity gate for PDB structures using grid-accelerated geometry rules.
$14.90
/moSubscription · cancel anytime
Secure checkout via Stripe
- Always the latest version
- Delivered live to your agent via MCP
- Cancel anytime, access ends at period end
Works with the AI Agents you already use
PHM MolCheck
Example session with this skill installed
Run a physical validity check on generated_binder_v4.pdb against the target_receptor.pdb reference. Summarize the clashes.
- Read your context and instructions
- Compiled the phm molcheck
The physical validity check for the provided structures is complete. The generated binder failed the geometry gate due to significant interface interference and internal bond distortion.
Molecular Validity Report
| Metric | Result | Details |
|---|---|---|
| Verdict | FAIL | 12 violations detected |
| Input Digest | sha256:e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855 | Validated |
| C1: Atomic Clashes | 4 Detected | Overlap > 0.4Å in hydrophobic core (Atoms 42, 88, 102, 115) |
| C2: Bond Lengths | 3 Deviations | > 25% variance from covalent radii in C-alpha backbone |
| C3: Cross-Clashes | 5 Detected | Steric interference at the binder-target interface |
| C4: Packing Density | 842 atoms/1000ų | Within normal range (800-950) |
Evidence Log
- Clash 01: Atom
CBResASN-12and AtomCG1ResVAL-45| Dist: 1.82Å | Overlap: 0.58Å - Bond 01: Atom
CResGLY-08and AtomNResLEU-09| Dist: 2.15Å | Deviation: 32% - Interface: Severe overlap detected at Arg-102 sidechain vs Target Glu-44.
Next steps
- Re-run the diffusion sampler with increased repulsion constraints.
- Manually inspect the interface clashes in PyMOL near residues 100-105.
- Verify the covalent geometry of the Gly-Leu peptide bond at residue index 8.
Subscription · always the latest version
- Secure checkout via Stripe
- Cancel anytime
- Delivered live via MCP
About this skill
AI structure generators often produce high-confidence molecular structures that are physically impossible. Atomic overlaps and bond length violations frequently slip through diffusion models and folding pipelines. This skill provides a deterministic physical-validity gate to verify structures against strict geometric rules before you waste compute on downstream simulations.
What it does
- Atomic clash detection flags non-bonded pairs with overlaps exceeding 0.4 Angstroms using MolProbity-style grid acceleration.
- Bond-length validation compares covalent distances against standard radii and flags deviations greater than 25%.
- Cross-clash analysis checks for physical interference between a generated binder and a target reference structure.
- Packing density audits calculates atoms per 1000 cubic Angstroms to identify structural anomalies.
- Deterministic verification provides a SHA-256 audit digest for every verdict to ensure reproducibility.
How it works
- Run the script against a single
.pdbfile to perform internal geometry checks for clashes and bond sanity. - Provide a second target
.pdbfile to trigger the interface cross-clash analysis between two structures. - Use batch mode on a directory of generated files to filter out physically invalid candidates from large datasets.
- Review the per-atom evidence report and the PASS/FAIL verdict appended with a unique audit digest.
Frameworks & tools
This skill requires Python and numpy. It functions without RDKit, GPUs, or network access, making it suitable for air-gapped or high-performance computing environments.
Why this beats prompting it yourself
LLMs cannot reliably calculate Euclidean distances or grid-based spatial overlaps. This tool replaces probabilistic guessing with deterministic geometry rules, catching errors that high-confidence scores from diffusion models often miss.
Use cases
- Filtering protein binder candidates generated by diffusion pipelines.
- Validating docked complexes for physical feasibility before MD simulation.
- Quality control for synthetic molecular datasets used in model training.
- Identifying atomic-level errors in protein folding model outputs.
Known limitations
A PASS verdict only confirms geometric validity; it does not guarantee solubility, expressibility, or actual binding affinity. Requires structures in PDB format.
PASS indicates no geometric violations but does not guarantee folding or binding. Requires PDB format and numpy. Does not replace experimental validation.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 18 days ago