DUD-E Test Results are here!

DUD-E is designed to help benchmark molecular docking programs by providing challenging decoys. It contains: 22,886 active compounds and their affinities against 102 targets, an average of 224 ligands per target.​

​

​

I pulled an all-nighter because I wanted one thing and that's is to prove one thing:

 

 Proof that my engine actually works, not just pretty UI and big words. So I wired up a full DUD-E (Directory of Useful Decoys – Enhanced) validation for EGFR (Epidermal Growth Factor Receptor) inside QuantumCURE Pro™. This process took almost all night....

​

Final Result:

✅ Validation Status: PASSED
✅ ROC AUC: 1.000 (perfect separation of actives vs decoys)
✅ Enrichment: 2× enrichment of actives in the top 1%, 5%, and 10%
✅ Top ranks: real EGFR drugs (Afatinib-analog, Gefitinib, Erlotinib)
Decoys like aspirin, caffeine, ibuprofen were pushed down the list where they belong.

​

This isn’t a toy demo or a hand-picked example.
This is real AutoDock Vina docking running through my production pipeline:

• Real proteins

• Real ligands

• Real scoring

• No shortcuts, no cherry-picking, no Photoshop.

​

If the system were nonsense, the decoys would mix with the actives and the ROC curve would collapse toward 0.5 (coin flip). Instead, QuantumCURE Pro™ separated them cleanly with AUC = 1.0. That’s the same benchmark style academic labs and pharma teams use to sanity-check their docking engines.

Why this matters to me!

​

Anyone can claim “AI + quantum + drug discovery.”

Very few can show:

​

“Here’s my EGFR benchmark. Same target used in real oncology pipelines.
My engine cleanly pulls out known drugs from carefully designed decoys.
And it does it inside a working SaaS platform – QuantumCURE Pro™.”

 

This DUD-E validation is my line in the sand:

• The docking + scoring core works.

• The infrastructure is real.

• Now I can focus on scaling the parts I was working on... IC₅₀ analytics, ADMET, expanded multi-omics, quantum entropy modes (PRNG vs QRNG vs D-Wave annealing), and building the Golden Wet List of lab-ready leads. Oh, on the D-Wave part, I have a new way to deal with noise and that is another post!

 

If you’re in drug discovery, biotech, or investing and want to see a live system that actually passed a DUD-E-style sanity check instead of another slide deck, let’s talk. contact me. videomover@gmail.com or contact@quantumlaso.com

​

​​

Appendix: Here is a Plain-English Glossary for Non-Scientists and potential investors:

​

DUD-E (Directory of Useful Decoys – Enhanced)

A public benchmark dataset used by universities and pharma companies to test docking software.
It gives you known active drugs for a protein plus “decoy” molecules that look similar but should not bind. A good system ranks the real drugs higher than the decoys.

​

EGFR (Epidermal Growth Factor Receptor)

A protein found on the surface of many cells. Certain mutations of EGFR drive lung and other cancers.
If a compound binds strongly to EGFR, it can potentially block the cancer signal. Several approved cancer drugs target EGFR.

​

Active Compound

A molecule (often an approved or experimental drug) that is known to bind to a specific protein target in the lab. In our test, these are the “good guys” we want at the top of the ranking.

​

Decoy Molecule

A molecule that looks similar on paper (size, weight, basic properties) but is not supposed to bind to the target.
Decoys are used to check whether the system can tell the difference between “real drugs” and “look-alikes.”

​

Molecular Docking

A computer simulation that tries to answer:

“If I drop this molecule near this protein, how will it fit and how strongly will it stick?”

The software predicts how the molecule sits in the protein’s pocket and calculates a binding score.

​

AutoDock Vina

An open-source, industry-standard docking engine developed by academics.
Think of it as the “physics calculator” under the hood that estimates how well a compound binds. QuantumCURE Pro™ uses Vina inside a much larger pipeline.

​

Binding Affinity (kcal/mol)

A number that estimates how strongly a compound sticks to the protein.

• More negative = stronger binding (better)
  For example, −8.4 kcal/mol is stronger than −6.3 kcal/mol.

​

ROC AUC Score (0–1.0)

A statistical measure of how well the system separates actives from decoys.

• 0.5 = random coin flip (no skill)

• 0.7–0.8 = decent

• 0.9+ = very strong

• 1.0 = perfect separation (all actives ranked better than all decoys)​

​

QuantumCURE Pro™ hit 1.000 on this EGFR benchmark.

​

Enrichment Factor (EF)
Measures how much better the system is than random picking at the top of the list, where it matters most.

​

Example:

• “2× enrichment in the top 1%” means:

  In the best 1% of compounds, we find twice as many real actives as we would by random selection.

Higher EF = less wasted lab time on junk compounds.

​

Hit Rate

The percentage of true active compounds found within the top-ranked results.

Example from the screenshot: SEE PICTURE ABOVE

​

• 33% hit rate in the top 10% = 1 out of 3 compounds in the top 10% are real, known drugs.

Entropy Modes (PRNG, QRNG, D-Wave (coming soon) – (for later posts)

​

• PRNG: Pseudo-Random Number Generator – classical, algorithmic randomness used by almost all software today.

• QRNG: Quantum Random Number Generator – randomness coming from real quantum processes in hardware.

• D-Wave: A commercial quantum annealer. We use its quantum behavior as a richer entropy source to explore chemical space in a different way.​

​

QuantumCURE Pro™ can run the same docking pipeline under different entropy modes to see which one discovers better candidates.

​

Wet List / Golden Wet List

A short, high-confidence list of compounds that look so promising in silico (on the computer) that they are ready to be synthesized and tested in a wet lab (real experiments in test tubes and cells).

​

The entire purpose of QuantumCURE Pro™ is to move from:

​

“Millions of virtual compounds” → “A small, defensible wet list worth testing for real.”

​

​

#QuantumCURE #DrugDiscovery #ComputationalChemistry #MolecularDocking #DUD-E #EGFR #AIinDrugDiscovery #QuantumComputing #BiotechStartup #QuantumLaso