My Citizen Scientist Engine Just Passed 478,808 Compounds – And the Hit Rate Is Exactly What I Hoped For
- mansour ansari
- Dec 5, 2025
- 4 min read
As of today, I have quietly moved 478,808 virtual Compounds through a quantum-enhanced drug discovery machine in my back office in Oklahoma.
Today I hit a nice milestone on the Citizen Scientist page of QuantumCURE Pro™ and the numbers made me smile, so I’m sharing them.
The Raw Numbers
Compounds processed (Citizen Scientist): 478,808
Compounds that made it into the Golden List: 30
Hit rate:
What does that mean?
In industry, typical virtual screening or HTS hit rates are often around 0.01–0.1% for “interesting” molecules — and that includes borderline stuff you might never seriously advance.
I’m running stricter rules:
Threshold: ΔG ≤ −8.5 kcal/mol
Toxicity: No toxicity flags allowed
Plus my usual extras: IC₅₀ estimate, entropy source tracking, CRISP-G glyph, etc.
So my 0.0063% isn’t “too low.”It means:
I’m only letting the top ~0.006% of screened compounds into the Citizen Scientist Golden List – the sharp tip of the pyramid, not the fog at the base.
Quality of the 30 Golden List Compounds
Inside those 30, the quality breakdown looks like this:
19 ultra-potent hits (≤ −9.0 kcal/mol)→ 63% of the list
11 strong hits (−8.5 to −9.0 kcal/mol)→ 37% of the list
Best affinity so far: about −15 kcal/mol
So not only is the hit rate tight, the majority of the keepers are very strong binders by docking standards. That’s exactly what you want in a funnel that will later feed:
MD refinement
consensus rescoring
and ultimately wet-lab collaboration
Scaling Up: What Happens at Millions of Compounds?
If I assume the same hit rate (~0.0063%), I can project:
5,000,000 compounds → ~300–350 Golden List hits
8,000,000 compounds → ~500 Golden List hits
And this is where it gets interesting for my next phase.
My working plan:
Batch | Total Compounds (cumulative) | Expected Golden List Size |
Batch 1 | 5M | ~300–350 |
Batch 2 | 10M | ~600–700 |
Batch 3 | 15M | ~900–1000 |
For Batch 1, my target is:
~500 Golden List compounds to feed into GANN, MD refinement, and deeper physics.
Given the current ratio, that means screening about 8 million compounds in total. That’s a lot of virtual chemistry… but it’s also doable when you have a focused engine and a quantum-aware pipeline. This will speed up when I set up an NVIDIA Spark Station and I can get up to 100% faster and that is another post....
Citizen Scientist → Factory: The Secret Tunnel
Here’s the part my inner stubbornness finds very satisfying. 😄
The Citizen Scientist page now has its own Community Golden List. Volunteers run docking jobs from their browsers, and when a compound passes the strict thresholds, it lands in that list.
Behind the scenes, I’ve built:
A private bridge from the Community Golden List into my Factory Golden List
An email alert that triggers when the Golden List crosses 300 compounds, so I know it’s time to start serious GANN and MD work.
GANN (Glyph-Aware Neural Network)An AI model that takes each Golden List compound (structure, docking scores, ADMET features, plus its CRISP-G glyph) and learns which patterns look most like real drug candidates. GANN re-scores and re-ranks the Golden List so the most promising molecules rise to the top before any expensive physics or lab work.
MD Refinement (Molecular Dynamics)A physics-based simulation where we let the protein and ligand move in explicit solvent over time. MD checks whether the docked pose is actually stable and uses those trajectories to refine the binding free energy (ΔG), giving a more realistic picture than a single static docking snapshot.
So the pipeline looks like this:
Citizen Scientist runs → Community Golden List (strict filters) →“Promote to Factory” → Private Drug Factory Golden List →GANN / MD / future wet-lab synthesis.
The community helps me explore chemical space; I curate and promote the best of the best into the factory.
Why I’m Excited About These “Tiny” Percentages
0.0063% sounds small until you remember:
Each of those 30 compounds is already filtered on:
strong binding (≤ −8.5 kcal/mol),
clean toxicity,
and richer metadata (entropy source, glyph, IC₅₀ estimate).
As I push from ~0.48M compounds toward 8M, that same selectivity should give me:
hundreds of high-quality candidates,
not thousands of noisy maybes.
In other words, I’m not trying to brag about how many hits I have. I’m quietly pleased with how hard it is to get into the Golden List at all.
If you see a molecule in there, it has already survived:
the docking gauntlet,
the toxicity filters,
the quantum forensics layer,
and my own paranoia.
And thanks to the Citizen Scientist page, a lot of those molecules will actually be discovered by other people’s browsers, not just my machines.
So yes, I’m sitting in a cluttered back office in “Timbuktu, Oklahoma,” wiring quantum entropy into cancer targets… but the numbers say the system is behaving like a serious virtual screening pipeline.
Next milestone: watching that Golden List counter hit 300, hearing my email alert go off, and kicking off the first big GANN + MD refinement batch.
“The higher-quality hit rate comes from both careful pipeline design and the way QRNG entropy explores chemical space. The QRNG doesn’t magically guarantee hits, but it seems to help avoid some of the biases and dead zones you get with pure PRNG, and our benchmarks suggest it enriches for real binders.”
— MansourQuantumLaso, LLC · QuantumCURE Pro™
Comments