Proton-coupled electron transfer rate prediction from first principles. Validated against 15 benchmark enzyme systems. <4% mean KIE error. SLO-1 KIE within 0.4% of experiment.
Proton-coupled electron transfer (PCET) is the mechanism behind enzyme catalysis, photosynthesis, respiration, and fuel cells. The kinetic isotope effect (KIE) — how much reaction rates change when hydrogen is replaced by deuterium — reveals the quantum mechanical tunneling contribution.
Predicting KIE accurately requires expensive DFT/QM calculations (hours per system) or empirical fitting. No existing tool provides rapid, general KIE prediction from molecular parameters alone.
Molecular Hessian matrices (force constants) are symmetric positive-definite. The V+/V− decomposition separates vibrational mode structure (shape) from coupling strength (magnitude), enabling geometric feature extraction.
This is an entirely novel application of Riemannian geometry to molecular kinetics — zero prior art in the literature. The result: <4% mean KIE error on 15 benchmark enzymes from seconds of computation.
Drug design requires understanding how enzymes accelerate reactions. KIE reveals whether proton tunneling is significant.
Fuel cells and artificial photosynthesis depend on PCET. Optimizing catalysts requires accurate rate prediction.
Mitochondrial electron transport, DNA repair, and antioxidant mechanisms all involve PCET chemistry.
15 enzyme systems with experimentally measured kinetic isotope effects.
Across 15 benchmark enzymes
Validated against experiment
398 core + 6 API tests
Kinetic isotope effect scales exponentially with donor-acceptor distance. Hover for enzyme details.
The proton sits in a double-well potential between donor and acceptor. Its quantum wavefunction tunnels through the barrier — heavier deuterium tunnels far less, producing the kinetic isotope effect.
FC overlap
FC overlap
H/D ≈ KIE driver
Reactant and product free-energy parabolas intersect at the transition state. The proton tunnels below the classical barrier — the shaded region shows the tunneling shortcut.
Activation energy
Marcus region
Crossing point
Visualize how vibrational normal modes modulate the donor-acceptor tunneling distance. Select a mode to see how different motions affect proton transfer.
C···O separation
Tunneling probability
150 cm⁻¹
The donor (C) and acceptor (O) oscillate toward and away from each other, modulating the tunneling distance. This low-frequency gating mode is the primary determinant of proton tunneling rates.
First empirical model predicting donor-acceptor gating from structural features alone. 16 enzyme systems, R² = 0.925.
Multivariate model fit
Ω vs MDA correlation
Cross-validated error
Ωgating = 808 − 104 × dDA − 27.5 × MDA
Eliminates need for expensive MD simulations. Predict gating from crystal structure + atom types.
Try the PCET Engine live. Toggle to “Live” to hit the real API.
Compute PCET rate from molecular parameters. SLO-1 wild-type example.
Screen mutations with just variant names and measured rates. No crystal structures or QM needed.
Screen directed evolution libraries computationally before wet-lab synthesis. Rank 1,000 variants in seconds, synthesize the top 50.
Saves: $50K–$500K per screening round in wet-lab costs
Predict cytochrome P450 PCET rates for drug candidates. Identify metabolic soft spots from molecular parameters, not just binding poses.
Complements: Simcyp, GastroPlus, ADMET tools
Optimize fuel cell catalysts, electrolyzer electrodes, and industrial biocatalysts. Electrochemical PCET with overpotential-dependent rates.
Markets: Hydrogen, fuel cells, water treatment
Consulting: $15K–$50K per enzyme optimization study. We calibrate, screen, and deliver ranked candidates. Learn more
Free academic tier: 100 calls/month, single-channel vibronic rates. No credit card required.
import requests
response = requests.post(
"https://pcet.omnisciences.io/v1/rate/parameters",
json={
"donor_acceptor_distance": 2.7,
"driving_force": -0.2,
"reorganization_energy": 1.1,
"coupling": 0.01,
"temperature": 300,
"proton_donor_acceptor_distance": 0.95,
"isotope": "H"
}
)
data = response.json()
print(f"Rate: {data['rate_constant']:.1f} s⁻¹")
print(f"KIE: {data['kie_factor']:.1f}")