In 2011, Fabio Pichierri proposed 'quantum proteomics': the large-scale quantum-mechanical study of the electronic structure of all proteins of an organism's proteome, using linear-scaling semiempirical calculations to obtain observables such as permanent dipole moments, residue partial charges, electrostatic potentials, and energy levels, and using these electronic fingerprints to classify proteins and connect them to biological function. The infrastructure component of the agenda has been partly realized by large ab initio fragment-molecular-orbital databases of biomacromolecules (FMODB; a 2024 dataset covering over 5,000 representative protein structures), but those store pair interaction energies for drug-discovery purposes rather than proteome-wide global electronic observables. The core scientific question remains open: whether a methodologically standardized, proteome-scale quantum-mechanical protocol can deliver reproducible electronic fingerprints, and whether such observables provide robust, statistically validated, functionally relevant protein classification across diverse families.

Background

Proteomics is the large-scale study of the proteins expressed by an organism's genome, collectively called the proteome. Protein structures are determined experimentally by X-ray crystallography, nuclear magnetic resonance (NMR) spectroscopy, or electron microscopy and deposited in the Protein Data Bank (PDB), a public archive of atomic coordinates. Classical proteomics analyses rest on sequence, structure, and interaction data; the electronic structure of proteins — the quantum-mechanical distribution of electrons that determines partial charges, electrostatic potentials, and dipole moments — is normally not computed at scale.

A permanent electric dipole moment (measured in Debye, about 3.34×10−303.34\times10^{-30} C·m) is the vector characterizing the separation of positive and negative charge in a molecule; because any folded protein has an asymmetric charge distribution, its dipole moment is nonzero. Protein dipoles have been measured experimentally for only a handful of proteins — for example, Antosiewicz and Porschke determined the experimental and calculated permanent dipole of α\alpha-chymotrypsin in 1989. On the computational side, whole-protein quantum-mechanical (QM) calculations became feasible through linear-scaling methods, whose cost grows roughly linearly with system size instead of the high polynomial scaling of conventional QM; semiempirical methods such as PM6 (a fast, parameterized approximate QM scheme) with the MOZYME localized-orbital self-consistent-field algorithm in the MOPAC program made proteins with thousands of atoms tractable. Solvent effects can be included implicitly with continuum models such as COSMO (the conductor-like screening model, which embeds the molecule in a polarizable medium characterized by a dielectric constant, here ε=78.4\varepsilon=78.4 for water). Khandogin and York had already shown in 2004 that linear-scaling semiempirical calculations yield usable 'quantum descriptors' of biomacromolecules in solution.

In this context, Fabio Pichierri's 2011 preprint 'Quantum Proteomics' proposed a new discipline: the large-scale QM study of the electronic structure of the proteins defining a proteome. As a demonstration, he computed PM6/COSMO permanent dipole moments (290–704 Debye, together with their orientations) for six proteins of the thermophilic bacterium Methanobacterium thermoautotrophicum whose NMR structures were taken from the PDB, adopting the deposited protonation states without explicit counterions (his earlier charybdotoxin study had found counterions do not significantly alter the dipole vector). He argued that, since every protein has a characteristic nonzero dipole, electronic observables could classify proteins and link electronic structure to biological function, and he envisioned a future database of dipole moments and other electronic properties of a proteome's proteins, combined with structural, kinetic, and biochemical data to illuminate biological processes. The preprint was never published in a journal and the term 'quantum proteomics' was not adopted by the community.

Meanwhile, a related but distinct line of development produced large databases of ab initio QM calculations on biomacromolecules via the fragment molecular orbital (FMO) method, which divides a macromolecule into fragments solved quantum-mechanically and then accounts for their pair interactions, giving access to inter-fragment interaction energies (IFIEs) and their energy components via pair interaction energy decomposition analysis (PIEDA). The FMODB database (Takaya et al., 2021) provides FMO-MP2 calculations for tens of thousands of PDB entries, and a 2024 Scientific Data dataset covers more than 5,000 representative protein structures across the fold families of SCOP2, a database that classifies known protein domains by their structural and evolutionary relationships, with over 200 million interaction energies. These efforts, however, target protein–ligand interactions and drug discovery; they compute pair interaction energies rather than global electronic observables of whole proteomes, and they do not test whether electronic observables classify protein function.

Problem Statement

Can quantum-mechanical electronic-structure calculations be scaled from single proteins to the proteins of an entire proteome in a methodologically standardized way, and do the resulting electronic observables provide robust, functionally relevant protein classifiers? Concretely: (i) apply a linear-scaling or fragment quantum-chemical protocol, with fixed and documented choices of semiempirical or ab initio level, solvent model and dielectric, protonation states, counterion treatment, and conformer selection, to the hundreds or thousands of proteins of a defined proteome, yielding a database of reproducible electronic fingerprints — permanent dipole moment magnitudes and orientations, residue partial charges, electrostatic potentials, and energy-level data; and (ii) integrate those fingerprints with existing structural, kinetic, and biochemical data to deliver a statistically rigorous answer — positive or negative — to whether electronic observables (beyond the permanent dipole moment alone) classify proteins by function across diverse families, over and above the information carried by the other (non-electronic) data types envisioned by the source proposal. How conformational ensembles, explicit ions, bound cofactors, and membrane environments should enter such a standardized protocol is itself part of the open methodological question rather than a required deliverable: the source demonstrated single experimental (NMR) conformers in continuum solvent, treated counterions as negligible on the basis of its charybdotoxin study, and raised coupling to molecular dynamics only as an outlook. A controlled negative result — a demonstration, under a documented protocol and with confound controls for protein size and composition, that the computed electronic observables carry no functional-classification information for the tested proteome — would answer the question in the negative for that protocol and observable set; it would not by itself establish that observables computed at other levels of theory lack classificatory value.

References

  1. Fabio Pichierri, Quantum Proteomics, arXiv:1107.5853 [q-bio.BM] (2011), https://arxiv.org/abs/1107.5853
  2. Daisuke Takaya, Chiduru Watanabe, Shunpei Nagase, Kikuko Kamisaka, Yoshio Okiyama, Hirotomo Moriwaki, Hitomi Yuki, Tomohiro Sato, Noriyuki Kurita, Yoichiro Yagi, Tatsuya Takagi, Norihito Kawashita, Kenichiro Takaba, Tomonaga Ozawa, Midori Takimoto-Kamimura, Shigenori Tanaka, Kaori Fukuzawa, Teruki Honma, FMODB: The World's First Database of Quantum Mechanical Calculations for Biomacromolecules Based on the Fragment Molecular Orbital Method, Journal of Chemical Information and Modeling 61 (2021) 777-794, DOI 10.1021/acs.jcim.0c01062, https://doi.org/10.1021/acs.jcim.0c01062
  3. D. Takaya, S. Ohno, T. Miyagishi, S. Tanaka, K. Okuwaki, C. Watanabe, K. Kato, Y. S. Tian, K. Fukuzawa, Quantum chemical calculation dataset for representative protein folds by the fragment molecular orbital method, Scientific Data 11 (2024), DOI 10.1038/s41597-024-03999-2, https://doi.org/10.1038/s41597-024-03999-2
  4. Jana Khandogin, Darrin M. York, Quantum descriptors for biological macromolecules from linear-scaling electronic structure methods, Proteins: Structure, Function, and Bioinformatics 56 (2004) 724-737, DOI 10.1002/prot.20171, https://doi.org/10.1002/prot.20171
  5. Jan Antosiewicz, Dietmar Porschke, The nature of protein dipole moments: experimental and calculated permanent dipole of alpha-chymotrypsin, Biochemistry 28 (1989) 10072-10078, DOI 10.1021/bi00452a029, https://doi.org/10.1021/bi00452a029
  6. Fabio Pichierri, The electronic structure and dipole moment of charybdotoxin, a scorpion venom peptide with K+ channel blocking activity, Computational and Theoretical Chemistry 963 (2011) 384-393, DOI 10.1016/j.comptc.2010.11.003, https://doi.org/10.1016/j.comptc.2010.11.003