Start here — in plain words
The strongest one is ranked #1 under Binding energy. Lower number = sticks stronger. Right now that is flufenoxuron. It is a candidate to test — not a proven killer.
Tap the blue button, bottom right. Ask normal questions like “Is it safe for bees?” It answers only from Affinix backend data — if it does not know, it says so.
Pick a chemical in Lab protocol below. That is the 7-step test plan a lab runs for you. Their result goes back into the backend and the ranking learns.
One liquid. Two jobs.
One bottle for the Oriental fruit fly (Bactrocera dorsalis). It should kill the flies that are here now, and stop the next ones from growing up. Same liquid underneath — for the farm, for the home, and for trash, where flies breed.
Live from /dual-action: NO — no single proven molecule does both fast adult kill and slow growth inhibition for this fly. Our slow molecules — the benzoylureas, like flufenoxuron — only stop young flies from growing. They do not kill adult flies fast. So one molecule is not enough. The plan is a blend of two, in one liquid.
Lead blend to test — one bottle, two molecules inside
Hits chitin — the stuff a young fly needs to make its skin. The young fly cannot grow up. Docking −10.694 against BdCHS2 (rank 5, best pest-vs-bee selectivity of the top 10: 1.82). ChEMBL CHEMBL1869031. This is the long-term job: no next wave of flies.
Hits the fly's nerves — the nAChR alpha 6 target (M4H3L1). Literature, not computed: LC50 at 5 hours = 9.03 ppm, LT50 under 2 hours at 100 ppm (McQuate 2005, B. dorsalis, protein bait). Its docking into the chitin pocket BdCHS2 is only −7.717 — expected, because its real target is the nerve, not chitin. ChEMBL CHEMBL4297065.
Keeps novaluron for the slow job and swaps in indoxacarb for the fast job — it blocks the voltage-gated sodium channel. Docking into BdCHS2 −8.770 (again, its real target is the nerve). This pair is a proven formulation for caterpillars (ADAMA Plethora, novaluron 5.25% + indoxacarb 4.5%) — not proven for B. dorsalis. Also docked, not in a blend: spinetoram −6.488 — worse than the sucrose negative control (−6.864), so no chitin-pocket fit at all.
Blend protocol — live from /protocol-blend
Mixing, in plain words: both molecules go in the same liquid, each at the same dose it gets when tested alone — that is how the lab can tell if the blend truly beats each single. There is no fixed ratio in the backend: the lab picks the 5-dose range and the route (treated diet / protein bait, or contact spray) and records it. Spinosad's fast numbers come from a protein-bait test, so bait liquid is the lead route to test first.
Blend protocol — lead (novaluron + spinosad)
- 4 arms, same doses: A slow alone, B fast alone, C blend, D solvent control — 30 flies per group, 3 groups per dose
- Count dead at 1, 2, 5 and 24 hours, then daily to day 7
- Pass fast: at least 50% dead by 24 hours in B and C • Pass slow: at least 50% more egg-hatch / moult failure vs control by day 7 in A and C
- Run honeybees in parallel at the same doses
Farm
The problem: A female fly lays eggs inside fruit. The fruit rots from inside and cannot be sold. One bad shipment can also be turned away.
How the liquid helps: Put out early, in bait stations in the orchard. Adults die fast, so they stop laying eggs. Young flies do not grow up, so the flies do not come back week after week.
What matters most here: It must keep working for weeks, and bees must be protected — bait stations, not open spraying while bees are feeding.
Home
The problem: Flies around bananas, the kitchen, and the bin. You see a fly now, and you want it gone now — not next week.
How the liquid helps: A small bait liquid in the kitchen. The fast part kills the flies you can see. The slow part stops eggs already laid from turning into a new wave.
What matters most here: It has to feel like it works on day one, and it must be kept away from food, kids, and pets, in a closed bait station.
Trash
The problem: Bins and waste areas are where flies breed. Kill only the adults you see, and more hatch tomorrow.
How the liquid helps: Bait liquid in and around the bin area. Adults feeding there die fast. Young flies in the waste do not grow up, so the bin stops making new flies.
What matters most here: It has to be cheap, simple, and topped up easily — this is the everyday, repeat use.
Dual-action lab test plan
One test, four groups. This is how a lab finds out if the blend really does both jobs — or if one part is doing all the work. Same flies, same doses, same days for every group. Count dead adults in the first hours and days, then count how many young flies grow up over the full test.
Should stop young flies from growing up, but should not kill adults fast. If adults die fast here, something is wrong with the test.
Should kill adults in hours. Watch whether flies come back later — with no slow part, the next wave should still grow up.
The one that must do both: adults dead fast, and no next wave after. This group has to beat groups 1 and 2 together, not just one of them.
Plain liquid, nothing in it. This is the honest baseline. Every other group is judged against this one — no control, no proof.
Pass rules (from /dual-action): Fast half passes if groups 2 and 3 kill at least 50% of adults by 24 hours. Slow half passes if groups 1 and 3 show at least 50% more egg-hatch / moult failure than control by day 7. Dual action is proven only if the blend passes both and beats each single group on its weak half. Run bees alongside, in parallel — same doses on honeybee workers; bee-safe only if bee deaths stay within 10% of control at the fruit-fly dose. Next step: send this four-group plan to a lab with the blend. Until that test runs, this folder is a plan to test — not a proven product.
Binding energy
Live from /dockTop candidates
Ranked by docking into real BdCHS2
2. Bistrifluron — −10.881
3. Fluazuron — −10.842
Top 10 — full ranking
Live from /rank| # | Molecule | Docking | Status |
|---|---|---|---|
| 1 | Flufenoxuron | −11.093 | CANDIDATE |
| 2 | Bistrifluron | −10.881 | CANDIDATE |
| 3 | Fluazuron | −10.842 | CANDIDATE |
The two scores, in plain words
Why one is trusted more than the otherDocking — the fit test PASSED
The computer takes the real shape of the fly's chitin protein (BdCHS2) and tries to fit each molecule into it, like a key into a lock.
Lower number = sticks stronger. Measured in kcal/mol. Flufenoxuron is −11.093, the strongest fit.
It passed its honesty check: real killers scored far stronger than sugar controls — a gap of 4.063. So docking is the score we rank by.
PLAPT — the guess test FAILED
A different computer model guesses how strongly a molecule binds, just from the molecule and protein written out as text — no real shape.
Higher number = stronger guess — the opposite direction from docking, which is why the two columns look confusing side by side.
It failed its honesty check: glucose — plain sugar, a known non-killer — scored highest of all, and the gap between killers and controls was only 0.115. So PLAPT scores are shown for honesty only. They do not pick winners.
Uncertainty — how unsure we are
A number from 0 to 1. 1.00 means the two tests completely disagree — glucose is 1.00 because PLAPT loved it and docking did not. The benzoylureas sit at 0.80.
High uncertainty is not bad news — it marks the molecules where one real lab result teaches us the most. That is why the “Test these first” list is ranked by uncertainty, not by score.
What “trustworthy” would take: a second test that passes its honesty check on this protein, agreeing with docking — plus a real lab result. Until then: TRUSTWORTHY: NONE.
Screening frequency
Last 28 daysTest these first — most to learn
Live from /learnTop binding pockets
Bee safety & lab protocol
Lab protocol — live insect test
- 30 flies per group, 5 doses, plus control
- Count dead every 24 hours for 7 days
- Pass if mortality is 50% higher than control and dose response is clear
- Run the same doses on honeybees in parallel
- Post the result to /results so the ranking learns
Lab results
Live from /resultsSafety ranges
FAIL
FAIL
FAIL