What Can ORCA Tell Us About a Drug Molecule? A Practical Quantum Chemistry Guide
ORCA reports the electrons of a ligand. A docking pose and a dynamics trajectory do not. The useful step is matching the calculation to the question: geometry, charge distribution, an orbital energy, a redox difference, or a spectrum. A number from the wrong job does not become a binding affinity.
ORCA is the electronic-structure package used for that job. Version 6.0, released in July 2024, is the rewrite described by Neese [1]. The method-choice note on this site, why B3LYP/6-31G* is not a safe default, still applies to every input below.
Start with the conformers
A flexible ligand is a population of structures. Bond rotation and intramolecular hydrogen bonds move both the energy and the electronic properties. A gap or a dipole computed on one arbitrary conformer describes that conformer.
ORCA 6 includes GOAT, a global optimization algorithm for molecules and clusters [2]. The manual documents the keyword as ! GOAT, used together with a method. The search runs many geometry optimizations, so the method has to be a fast one [3]. Structures are kept along the way, which means the result is an ensemble with relative energies, not a single optimized geometry [2], [3].
At temperature $T$, a Boltzmann weight of conformer $i$ is
$$ p_i=\frac{e^{-\Delta G_i/(RT)}}{\sum_j e^{-\Delta G_j/(RT)}}. $$The exponent takes a relative free energy. An electronic energy difference is not that free energy, and a gas-phase ranking is not the population in solution. For a flexible ligand, report the low-energy set and the weights, and say which energy was used to compute them.
What the method line is doing
A label such as B3LYP-D3(BJ)/def2-TZVP is three decisions written as one string.
- B3LYP is a hybrid functional. It approximates exchange and correlation. It is not a universal choice for barriers, metal complexes, charge-transfer states, or redox potentials.
- D3(BJ) is Grimme’s D3 dispersion correction with Becke–Johnson damping. Without a dispersion correction, noncovalent contacts and some conformer preferences sit on the wrong surface.
- def2-TZVP is a triple-zeta valence basis with polarization. It gives the orbitals more flexibility than a minimal or double-zeta basis.
A geometry optimization plus a harmonic frequency job, with the charge and multiplicity still to be set, looks like this:
! B3LYP D3BJ def2-TZVP Opt Freq TightSCF
* xyz 0 1
... atomic coordinates ...
*
The 0 1 is charge and multiplicity for a neutral closed-shell example. Both have to match the molecule in the coordinate block, and the coordinate block has to be a real structure. Harmonic frequencies then check the stationary point: a local minimum on that surface has no imaginary mode. Zero imaginary modes do not mean the structure is the global minimum. That is the conformer search.
This keyword line is a way to read an input. It is not the production default argued in the B3LYP/6-31G* note. That note’s objection to the old default is the missing dispersion and the small basis. D3(BJ) and a triple-zeta basis answer those two objections. They do not make B3LYP the right functional for every property. If the experiment was run in solution, the gas-phase number can dominate the error. ORCA 6 documents CPCM and an analytic SMD Hessian for implicit solvent [1]. The same note names SMD and COSMO-RS when the data are solution data.
Electrostatic potential is a map, not an affinity
A molecular electrostatic potential (MEP) is the potential from the nuclei and the electrons, usually drawn on an electron-density isosurface. Electron-rich regions and electron-poor regions are then visible on one picture, which is a way to talk about electrostatic complementarity with a partner.
An MEP can suggest where a hydrogen-bond donor or a cation would sit if electrostatics dominated. Solvation, desolvation, steric fit, protein motion, and entropy are still outside the picture. The map does not assign a binding affinity.
Orbital gaps are descriptors
The highest occupied and lowest unoccupied Kohn–Sham orbitals, HOMO and LUMO, are a compact picture of where the density can donate or accept an electron, and of which regions a low-lying excitation might involve. The gap drawn between them is
$$ \Delta\varepsilon=\varepsilon_{\mathrm{LUMO}}-\varepsilon_{\mathrm{HOMO}}. $$A smaller gap is sometimes read as “more reactive.” That reading skips the reaction. A Kohn–Sham gap is not, in general, an optical excitation energy, a fundamental gap, or a potency. The reaction and the experiment decide what the number means.
Conceptual DFT sometimes compresses the same idea into electronegativity $\chi$, hardness $\eta$, and electrophilicity $\omega$, estimated from an ionization energy $I$ and an electron affinity $A$:
$$ \chi\approx\frac{I+A}{2},\qquad \eta\approx\frac{I-A}{2},\qquad \omega\approx\frac{\chi^{2}}{2\eta}. $$For a number you intend to compare, take $I$ and $A$ from total-energy differences of the charge states, a $\Delta$SCF calculation, and state whether the geometry was frozen (vertical) or relaxed (adiabatic). Substituting raw orbital energies for $I$ and $A$ is a further approximation, and it should be labeled as one.
Redox is a difference of states
Oxidation and reduction matter for metabolism, electron transfer, and some toxicity paths. ORCA can compute the charge states. An oxidation free energy is the free-energy difference between the oxidized species and the reduced species. Turning that difference into a potential also needs a reference electrode and a solvation treatment that are the same for every molecule in the comparison.
A structural change can make an oxidation more or less costly. That is a well-posed ORCA question. The site of CYP-mediated metabolism is a different question. Enzyme pose, the catalytic barrier, the spin state, and the reactive oxygen species are not encoded in the ligand HOMO.
Spectra are where the calculation meets a measurement
| Property | Typical ORCA job | What the comparison can support |
|---|---|---|
| IR | Harmonic frequencies | Functional-group assignments, after the harmonic approximation is remembered |
| UV–Vis | TD-DFT excited states | Which transitions sit under a band, for that functional and that excited-state character |
| NMR | Shielding | Shifts, once a reference and, where needed, a conformer average are in place |
| Redox | Charge-state energies plus solvation | Relative ease of oxidation or reduction under the chosen reference |
Harmonic IR frequencies are commonly scaled. TD-DFT moves with the functional and fails more readily for charge-transfer and double excitations. An NMR shielding is not a chemical shift until it is referenced. None of these jobs replaces the spectrum. They give a candidate assignment.
A fluorine atom is not a gap argument
Take a matched pair in which adding fluorine changes the assay. A smaller HOMO–LUMO gap on the fluoro analogue is not an explanation of the potency. The questions that can actually be calculated are narrower.
- Did the fluorine change the conformer weights, or the intramolecular hydrogen bond?
- Did the electrostatic potential move at the contact that is supposed to matter?
- Did the dipole, the acidity, the solvation free energy, or the oxidation energy move?
- Do the docked poses and the dynamics trajectories show a different contact, and does a measurement test that contact?
Quantum chemistry describes the ligand. A stronger-binding claim needs the structure, the thermodynamics, and the experiment pointing the same way. The calculation to run is the one that distinguishes those four questions, not the full menu of descriptors followed by a search for a correlation.
References
- Neese F. Software update: the ORCA program system, version 6.0. WIREs Comput Mol Sci 2025;15:e70019. doi:10.1002/wcms.70019. ORCA 6.0 was released in July 2024.
- de Souza B. GOAT: a global optimization algorithm for molecules and atomic clusters. Angew Chem Int Ed 2025;64:e202500393. doi:10.1002/anie.202500393
- ORCA 6.0 manual, section 6.4, GOAT. faccts.de/docs/orca/6.0
