Against the completed rescue/support Boltz runs in this workspace, the ranking was roughly compound 38 > compound 30 > carnosic acid > EGCG > quercetin > honokiol > TH10785 > sulforaphane > curcumin > eugenol > ferulic acid > rosmarinic acid > gallic acid. So the best tradeoffs are:
Best cheap screen: quercetin
Best serious buy-now OGG1 activator: TH10785
Best custom quote targets for something like Onepot Flow: compound 30, compound 38, and probably compound F from the older rescue paper (2018 rescue paper); for compound F I could not confirm a current catalog source, but it looks simpler than 30/38, so custom synthesis is likely more straightforward
and for defective asian NQO1*2
automated chemical synthesis might be moving fast! [tho the most general form is allegedly harder than bacterial synthesis of peptides]
this is an example of a workflow/fix for defective Asian OGG1 (can you can follow up with tool-orchestra) and other chemicals that might be synthesizeable…
Against the two medicinal rescue-like benchmarks in your OGG1 panels:
compound 38: binding_confidence 0.649, complex_ipde 0.275, lesion distance 3.29 A
compound 30: 0.624, 0.286, lesion distance 3.46 A
Your three new leaders compare like this:
harmine: 0.795, 0.271, lesion distance 4.46 A
Best raw scorer, and its pocket-shell overlap with compound 38/30 is high (0.783 / 0.857).
Interpretation: strongest affinity-style hit, but less lesion-proximal than the best rescue-like poses.
cryptotanshinone: 0.622, 0.294, lesion distance 2.53 A
Nearly tied compound 30 on score, but sits much closer to the 8OG lesion and keeps strong shell overlap (0.826 / 0.818).
Interpretation: best overall “rescue-like” lead of the new set.
baicalein: 0.602, 0.298, lesion distance 3.98 A
Slightly below compound 30, but clearly above the old plant leaders like quercetin, honokiol, and EGCG.
Interpretation: practical plant lead, probably cleaner to advance than harmine.
If I had to rank them for next wet-lab work: cryptotanshinone first, baicalein second, harmine third. harmine wins the scoreboard, but cryptotanshinone looks more like the known rescue-pocket geometry.
Sigma 233630 exists, but one US page said “not currently sold in your country”
Practical first order:
cryptotanshinone: 10–25 mg
baicalein: 100–500 mg
harmine: 200–500 mg
One caution: vendor-reported solubility for cryptotanshinone is inconsistent. Cayman reports very poor DMSO solubility, while Selleck reports roughly 8–10 mg/mL. Treat it as a solubility-risk compound and pilot stocks first.
How To Test Next
Best next step is a direct biochemical rescue assay, not another modeling round.
Build a purified-enzyme assay with OGG1 WT and your S326C mutant on a defined 8-oxoG:C duplex.
Use a denaturing gel or capillary readout, not a simple fluorescence-only plate assay.
Reason: harmine, baicalein, and cryptotanshinone are all plausible optical-interference compounds.
Run 0.1–100 µM dose-response curves with fixed DMSO <= 1%.
Include controls:
compound 38 and compound 30 if you can get them
otherwise resveratrol, naringenin, and carnosic acid
vehicle only
no-enzyme control
APE1-only control if you use a coupled assay
Add counterscreens:
DNA binding/intercalation without enzyme
detergent sensitivity for aggregation
fluorescence/absorbance interference
redox-cycling check
For substrate, IDT currently lists internal 8-oxo-dG oligo chemistry here: IDT Int 8-oxo dG. Their page showed 100 nmol DNA oligo pricing at $299 and says HPLC purification is required.
For protein, off-the-shelf OGG1 I found was WT-centric:
So for the real experiment, the clean path is: express/purify your exact S326C protein, use WT as benchmark, and test cryptotanshinone, baicalein, and harmine side by side.
The main new insight is that cryptotanshinone looks like the best new rescue-like lead, not harmine. cryptotanshinone is much closer to the published OGG1 activator contact logic around Gly42 and Phe319, and in your model it is more lesion-proximal than compound 38 or compound 30 while keeping strong shell overlap. harmine still wins on raw binding_confidence, but it now looks more like a high-scoring planar binder with a serious DNA-intercalation and off-target problem than a clean rescue template. baicalein is the best indirect or hybrid follow-up because it has decent pocket geometry and there is already literature linking it to higher OGG1 expression and lower oxidative DNA damage.
My practical ranking for next work is:
cryptotanshinone as the primary direct-rescue family
baicalein as the primary indirect/hybrid-rescue family
harmine as a mechanistic control, not the main translational lead
emodin as a useful false-positive control because its geometry is better than expected but its topo II / genotoxic liability is too strong