·
Okay here’s a game - you’re making nanobody therapeutics. AI can help you optimise them, but you need to prioritise order of feature importance. What is your order? Affinity Specificity Expressibility Clinical toxicity In vivo stability Purified stability Patentability Other?
/accI’d prioritize ADMET (Clinical toxicity) because it is the hardest to get with AI alone. If I had some kind of tool that could predict ADMET reliably, I’m confident you could get the other properties pretty good with existing tooling.
marinka zitnik is behind this…
what is dif between homos/lumos and fukui indices?
what are the most extreme fukui indices ever? are carboxylic acids high even tho they aren’t that toxic biologically
2:55 PM
Synthesized global versus local reactivity concepts and biological disconnect
Synthesized global versus local reactivity concepts and biological disconnect
Three distinct questions, let me take each seriously.
HOMO/LUMO vs Fukui: the actual relationship
HOMO/LUMO are global orbital descriptors. They tell you how easily the whole molecule donates or accepts an electron — the HOMO energy approximates the ionization potential, LUMO approximates electron affinity (via Janak’s theorem). They don’t tell you where.
Fukui indices are essentially the HOMO and LUMO decomposed onto individual atoms. f⁺ at atom k is roughly “what fraction of the LUMO density sits on atom k.” f⁻ is the same for HOMO.
So the relationship is:
f⁺(k) ≈ |ψ_LUMO(k)|² (projected onto atom k) f⁻(k) ≈ |ψ_HOMO(k)|² (projected onto atom k)
But it’s not exactly that, which is the interesting part. The finite-difference approach (computing charges for N, N+1, N-1 electron systems) captures orbital relaxation — when you add an electron to a molecule, the other electrons rearrange. Pure HOMO projection ignores that rearrangement. So Fukui is slightly more honest about what actually happens during a reaction.
Practically: HOMO/LUMO gap tells you global hardness/softness. Fukui tells you where to hit. You need both. A molecule can have a very high-lying HOMO (easy to oxidize) but with HOMO density spread across twenty atoms — each individual Fukui index will be small. Versus a molecule with a moderate HOMO but all density concentrated at one atom — big Fukui at that atom, that’s where chemistry happens.
Most extreme Fukui indices recorded
The condensed Fukui function is normalized so all atoms sum to 1. So the theoretical maximum for a single atom is 1.0 — all reactivity on one atom.
The most extreme cases are in small molecules and highly localized electrophiles:
- CO: the carbon f⁺ is extremely high (~0.8–0.9 depending on scheme). Almost all LUMO density on C. This is why CO is a soft nucleophile that attacks transition metals at C, not O — counterintuitive from electronegativity alone, explained by Fukui.
- BH₃: boron f⁺ approaches 0.85–0.95 in many calculations. The empty p orbital is almost entirely on B.
- Carbenes (:CH₂): the carbene carbon f⁺ and f⁰ are both close to 1.0. This is why carbenes are so indiscriminately reactive — all reactivity concentrated on one atom with essentially no electron density elsewhere to dilute it.
- Singlet oxygen (¹O₂): both oxygens split f evenly, so ~0.5 each, but the total is still extreme — very soft, very reactive.
- HF: fluorine has f⁻ close to 0.9. The lone pairs are almost entirely on F.
In large drug-like molecules you rarely see values above 0.3–0.4 on any single atom because the π system delocalization spreads it around. Ascorbate’s C3 at ~0.5–0.6 is already quite extreme for a molecule that size.
The “record” for a biologically relevant molecule is probably something like a highly reactive epoxide or aziridine intermediate, where the strained carbon can have f⁺ of 0.5+ despite being in a larger scaffold.
Carboxylic acids: high Fukui, low toxicity — the disconnect
Yes, and this is a genuinely important point about what Fukui actually measures.
The carbonyl C of a carboxylic acid does have a relatively high f⁺. It’s electrophilic. But several things conspire to make that irrelevant biologically:
1. Ionization state at pH 7. Carboxylic acids (pKa ~4–5) are >99% deprotonated at physiological pH. The carboxylate anion (-COO⁻) has the negative charge delocalized across both oxygens — this massively reduces the carbonyl C electrophilicity compared to the neutral acid. The Fukui index computed on the neutral molecule doesn’t apply in vivo.
2. Hard-hard vs soft-soft mismatch. Even when reactive, the carboxylate is a hard electrophile (carbonyl). Biological nucleophiles like cysteine thiols are soft. Hard-soft mismatch = slow reaction. The electrophilic warheads that actually cause toxicity (Michael acceptors, epoxides, quinones) are softer electrophiles that match soft biological nucleophiles.
3. Reversibility. Carboxylic acids can form esters and amides, but these are often reversible under biological conditions, or enzymatically hydrolyzed. Covalent adducts with thiols from hard electrophiles tend to be unstable. Irreversible covalent modification — which is what causes toxicity — requires a match between electrophile softness and nucleophile softness.
4. Kinetics vs thermodynamics. Fukui tells you where a reaction is favorable, not how fast it goes. The activation barrier for carboxylic acid + thiol in aqueous solution is high. The reaction happens, but on timescales that allow enzymatic detox to compete.
So the short answer: Fukui index measures intrinsic orbital-level reactivity at a frozen geometry in vacuo. Biological toxicity is the product of that times ionization state times hard-soft matching times kinetics times steric access times whether the adduct is reversible times whether the enzyme gets there first. High Fukui is necessary but very far from sufficient for toxicity.
The DAQ case from earlier threads is instructive — it’s toxic precisely because it has high f⁰ at C5, it’s a soft electrophile matching soft nucleophiles (cysteines), it forms irreversible thioether adducts, and NQO1 is the enzyme that normally interrupts the cycle (which is why P187S matters).
“This is why CO is a soft nucleophile” soft and not hard??
2:56 PM
Reconciled CO’s soft nucleophile classification through HSAB theory principles
Reconciled CO’s soft nucleophile classification through HSAB theory principles
Yeah good catch, that does sound wrong at first. But it’s actually correct — CO is soft, and it’s one of the canonical examples used to teach why HSAB theory is non-obvious.
The intuition that trips people up: O is more electronegative than C, so you’d expect oxygen to be the reactive end and you’d expect the molecule to behave like a hard species. Neither is true.
What actually determines hardness/softness is polarizability, not electronegativity. Hard nucleophiles have electron density tightly held in small, contracted orbitals — F⁻, OH⁻, NH₃. Soft nucleophiles have diffuse, polarizable electron clouds that can deform to match a soft electrophile.
CO’s HOMO is a sigma lone pair localized primarily on carbon, and it sits in a relatively diffuse orbital because the C-O triple bond pulls electron density away from O toward the internuclear region, leaving C’s lone pair exposed and polarizable. It’s not tightly held. It deforms easily in an external field.
The evidence that CO is soft:
- It binds almost exclusively to low-oxidation-state transition metals — Fe(0), Ni(0), Cr(0) — which are soft acids. It has essentially no affinity for hard metal centers like Mg²⁺ or Ca²⁺.
- Hemoglobin poisoning works because Fe²⁺ in heme is a soft acid, and CO (soft base) outcompetes O₂ for it.
- The Fukui f⁺ being concentrated on C is consistent with softness — soft electrophiles and soft nucleophiles both tend to have localized frontier orbital density rather than diffuse charge.
The counterintuitive part — that the less electronegative atom is the reactive end — is exactly why Fukui functions were useful when Fukui developed the theory in the 50s. Pure electronegativity arguments predicted CO should react at O. Frontier orbital theory (and later Fukui functions) correctly predicted C, matching experiment.
So: hard nucleophile = F⁻, OH⁻, water, carboxylate. Soft nucleophile = CO, phosphines, thiolates, iodide, CO again. The dividing line is polarizability of the electron pair, not electronegativity of the atom holding it.
why isnt CO that reactive
3:05 PM
Good question and there’s a real tension here that’s worth unpacking.
CO is actually quite reactive — with the right partners
It’s not universally unreactive. It’s just selective. With soft low-valent transition metals it’s extremely reactive — carbonyl complexes like Ni(CO)₄ form spontaneously at room temperature, which is why nickel carbonyl is so acutely toxic. With organolithium and Grignard reagents it inserts readily. In the Fischer-Tropsch process it reacts under industrial conditions.
The biological toxicity at ppm concentrations is also pretty good evidence it’s not sluggish — it just needs the right target (Fe²⁺ in heme).
Why it seems unreactive in most contexts
Three separate reasons, which are worth distinguishing:
1. Kinetic barrier, not thermodynamic stability
The HOMO lone pair on C is the nucleophilic end, but it’s sigma symmetry — it points away from the molecule linearly. To donate into a metal d orbital requires precise geometric approach. Collisions at wrong angles don’t react. This is an orientation/steric problem, not an intrinsic electronic weakness.
With small organic electrophiles like protons or carbocations, the sigma lone pair on C actually has poor orbital overlap compared to a nitrogen or oxygen lone pair, which are better geometrically disposed to interact in typical organic reaction geometries.
2. The triple bond creates kinetic stability even when thermodynamics is favorable
CO has a very strong triple bond (bond dissociation energy ~1072 kJ/mol, one of the strongest bonds known). Even when a reaction with CO is thermodynamically downhill, you sometimes have to partially distort or activate the molecule first. This shows up in industrial catalysis — CO reactions often need metal catalysts precisely because the metal activates CO by backdonation into the CO π* orbital, weakening the triple bond and making the carbon more electrophilic.
Without a metal doing that backdonation, CO is much less reactive as an electrophile. So paradoxically: CO is a good nucleophile (at C) but a mediocre electrophile until activated.
3. HOMO/LUMO mismatch with most organic reaction partners
In typical organic chemistry the electrophiles you encounter are hard (carbonyl carbons, protons, hard Lewis acids). CO is soft. Soft nucleophile + hard electrophile = poor orbital overlap, slow reaction, high activation barrier. This is the HSAB kinetic argument — mismatched pairs react slowly even when thermodynamics is fine.
The Janusian thing worth noticing
CO is simultaneously one of the strongest ligands in inorganic/organometallic chemistry and nearly inert in everyday organic chemistry. Same molecule, wildly different reactivity depending on what you put it with. That’s not a contradiction — it’s the HSAB principle working correctly. The selectivity IS the property.
It’s also worth flagging that CO’s apparent unreactivity in air, water, and most organic contexts is partly why it’s so dangerous biologically — it doesn’t react with the aqueous environment, doesn’t get metabolized quickly, diffuses freely across membranes, and then hits exactly the soft Fe²⁺ target it was “designed” (evolutionarily speaking, in the sense of orbital matching) to bind.
oh, CO has high activation energy with most substrates?
3:06 PM
Recognized straightforward chemistry inquiry requiring direct knowledge application
Recognized straightforward chemistry inquiry requiring direct knowledge application
Sort of, but I want to be more precise because “high activation energy” isn’t quite the right frame for all cases.
It’s more that the activation energy is context-dependent in an unusual way
For CO reacting with most things you’d encounter in organic chemistry or aqueous biology — protons, carbonyls, alcohols, amines — yeah, the barrier is high enough that the reaction just doesn’t happen at ambient conditions. But the reason isn’t one thing:
With hard electrophiles (protons, hard Lewis acids): it’s genuinely a barrier issue. The orbital symmetry mismatch means poor overlap in the transition state. The lone pair on C doesn’t stabilize the transition state well.
With radical species: CO actually inserts into radicals with moderate barriers — this matters in industrial chemistry and combustion. Not that slow actually.
With transition metals at low oxidation state: essentially no barrier in many cases. Ni(CO)₄ formation from Ni(0) and CO is fast at room temperature — the barrier is negligible.
The more interesting point is barrier shape, not just barrier height
CO approaching most organic electrophiles has a barrier that’s largely entropy-driven rather than purely electronic. The required geometry is very specific — linear approach, correct orbital alignment. So ΔS‡ is very negative (you have to give up a lot of translational and rotational freedom to hit the right geometry), which contributes to a high ΔG‡ even when the electronic barrier is moderate.
This is actually different from, say, a carbanion reacting with a carbonyl — those reactions also need specific geometry but the electrostatic attraction helps funnel the approach. CO has no strong electrostatic handle to guide it in.
Where I’d flag uncertainty
I’m fairly confident on the qualitative picture. The specific activation energies for CO with various substrates I’m less certain about numerically — I’m reasoning from principles rather than from a table of measured Ea values. If you wanted actual numbers you’d want computational or kinetic data per reaction class. The statement “high activation energy with most substrates” is roughly right as a generalization but elides a lot of variation.
from fukui pov, why is guanine the most easily damaged base
3:12 PM
Reconciled ionization potential and Fukui orbital density to explain guanine’s reactivity
Reconciled ionization potential and Fukui orbital density to explain guanine’s reactivity
Three converging reasons from a Fukui perspective, building on what we already drew:
1. Highest HOMO energy of the four bases
Guanine’s ionization potential is the lowest of all four DNA bases — roughly 7.7 eV vs ~8.8 for adenine, ~9.0 for cytosine, ~9.5 for thymine. Since f⁻ is essentially HOMO density projected onto atoms, a higher-lying HOMO means electrons are less tightly held, so f⁻ values are large. The molecule is more willing to donate an electron to whatever oxidant hits it.
The reason guanine’s HOMO is highest: the amino group at C2 and the carbonyl at C6 together push electron density into the bicyclic purine ring. Two fused rings give more π delocalization than the pyrimidine bases (cytosine, thymine) which only have one ring. More delocalization = HOMO rises in energy.
2. HOMO density concentrated on few atoms
A high HOMO energy would matter less if the density were spread across twenty atoms — each individual f⁻ would be small. But in guanine the HOMO is concentrated primarily on N7, C8, and the C4-C5 bond region. So those atoms have large individual f⁻ and f⁰ values even though the molecule isn’t tiny.
This is exactly the CO lesson in reverse — localized frontier orbital density means high Fukui at specific atoms, which means high local reactivity regardless of the global picture.
3. N7 sits in the major groove, unprotected
This is where Fukui meets geometry. N7 has high f⁻ — good electron donor, easy electrophilic attack site. And it’s physically exposed in the major groove, not involved in Watson-Crick hydrogen bonding, not stacked against a partner. Electrophiles like DAQ, alkylating agents, and cisplatin can reach it without displacing base pairing.
Cytosine also has nucleophilic nitrogens but they’re involved in H-bonding with guanine on the opposite strand — geometrically protected. Thymine’s most nucleophilic sites face inward. Guanine’s N7 is both electronically activated (high f⁻) and sterically exposed simultaneously.
Why 8-oxoG is then more susceptible than G
Which connects to the diagram we drew — once you form 8-oxoG, you’ve added a carbonyl at C8 that further raises electron density in the ring, pushing the HOMO even higher. The condensed Fukui functions can be directly linked to antioxidative activity because a molecule with a low HOMO eigenvalue is likely to be a poor electron donor, whereas a high HOMO makes it a good electron acceptor target for oxidants. ACS Publications 8-oxoG’s HOMO is higher than G’s — it’s a better electron donor, meaning oxidants hit it preferentially over unmodified guanine. That’s why oxidative damage clusters at 8-oxoG sites and why the lesion is self-amplifying if not repaired.
So the Fukui summary: guanine wins on all three axes simultaneously — highest HOMO energy (large f⁻ globally), concentrated HOMO density (large f⁻ locally at N7/C8), and geometric exposure in the major groove. The other bases are missing at least one of those three.