Weitao Yang’s research while affiliated with Duke University and other places

Machine learning force fields offer the ability to simulate biomolecules with quantum mechanical accuracy while significantly reducing computational costs, attracting growing attention in biophysics. Meanwhile, leveraging the efficiency of molecular mechanics in modeling solvent molecules and long-range interactions, a hybrid machine learning/molecular mechanics (ML/MM) model offers a more realistic approach to describing complex biomolecular systems in solution. However, multiscale models with electrostatic embedding require accounting for the polarization of the ML region induced by the MM environment. To address this, we adapt the state-of-the-art NequIP architecture into a polarizable machine learning force field, NepoIP, enabling the modeling of polarization effects based on the external electrostatic potential. We found that the nanosecond MD simulations based on NepoIP/MM are stable for the periodic solvated dipeptide system and the converged sampling shows excellent agreement with the reference QM/MM level. Moreover, we show that a single NepoIP model can be transferable across different MM force fields, as well as extremely different MM environment of water and proteins, laying the foundation for developing a general machine learning biomolecular force field to be used in ML/MM with electrostatic embedding.

Despite its ubiquity in nature, some details of the animal olfactory system remain unclear. One such mystery is the mechanism by which olfactory receptors (ORs) recognize the olfactant molecules they bind to. Some evidence indicates that ORs can distinguish between molecules that differ only in isotopic composition, suggesting that olfactants' vibrational modes may play a role in their recognition. In 2023, the first experimental structure of a human olfactory receptor—OR51E2—was produced, providing computational scientists an opportunity to shed additional light on this problem. We compute the infrared spectrum of the olfactant propionate (C2H5COO–) in the OR51E2 binding site by quantum mechanics/molecular mechanics, with atomic positions taken at 25 time points over a 500 ns molecular dynamics simulation. By comparing the spectra for all 32 possible hydrogen/deuterium isotopic combinations in propionate and across the time snapshots, we estimate the relative strength of the isotope effects and thermal fluctuations in the vibrational energy of C2H5COO–. The high-frequency C–H modes are are about 800 cm^{-1} higher in energy than their deuterated counterparts, a large separation relative both to their fluctuations over time and to the thermal energy available at physiological temperature. Lower-frequency vibrations do not display such a clear isotopic separation. Thus, any vibrational component to olfactant recognition—especially one that allows distinguishing between isotopes—is likely to involve these high-frequency modes.

This Special Issue of the Journal of Chemical Physics is dedicated to the work and life of John P. Perdew. A short bio is available within the issue [J. P. Perdew, J. Chem. Phys. 160, 010402 (2024)]. Here, we briefly summarize key publications in density functional theory by Perdew and his collaborators, followed by a structured guide to the papers contributed to this Special Issue.

Density functional theory (DFT) is a powerful tool for quantum-mechanical calculations, but practical calculations suffer systematic errors like incorrect charge densities and total energies in molecular dissociation, underestimated band gaps in bulk materials, and poor energy level alignment at interfaces. These problems are due to delocalization error. The localized orbital scaling correction (LOSC) removes delocalization error in molecules effectively, but screening of the Hartree-exchange-correlation response is necessary to correct it in materials. We introduce LOSC with system-dependent linear-response screening (lrLOSC), which effectively corrects delocalization error in semiconductors and insulators. After correcting for electron-phonon effects, the band gaps of eleven test systems are predicted with a mean absolute error of 0.28 eV, comparable to self-consistent GW. This method represents a significant step forward in correcting densities and total energies across system sizes and solving the band gap and energy level alignment problems entirely within the DFT framework.

The recently developed Quasiparticle Energy (QE) scheme, based on a DFT calculation with one more (or less) electron, offers a good description of excitation energies, even with charge transfer characters. In this work, QE is further extended to calculate electron transfer (ET) couplings involving two excited states. We tested it with a donor–acceptor complex, consisting of a furan and a 1,1-dicyanoethylene (DCNE), in which two low lying charge transfer and local excitation states are involved. With generalized Mülliken-Hush and fragment charge-difference schemes, couplings from the QE approach generally agree well with those obtained from TDDFT, except that QE couplings exhibit better exponential distance dependence. Couplings from half-energy gaps with an external field are also calculated and reported. Our results show that the QE scheme is robust in calculating ET couplings with greatly reduced computational time.

Chaperones are a large family of proteins crucial for maintaining cellular protein homeostasis. One such chaperone is the 70 kDa heat shock protein (Hsp70), which plays a crucial role in protein (re)folding, stability, functionality, and translocation. While the key events in the Hsp70 chaperone cycle are well established, a relatively small number of distinct substrates were repetitively investigated. This is despite Hsp70 engaging with a plethora of cellular proteins of various structural properties and folding pathways. Here we analyzed novel Hsp70 substrates, based on tandem repeats of NanoLuc (Nluc), a small and highly bioluminescent protein with unique structural characteristics. In previous mechanical unfolding and refolding studies, we have identified interesting misfolding propensities of these Nluc‐based tandem repeats. In this study, we further investigate these properties through in vitro bulk experiments. Similar to monomeric Nluc, engineered Nluc dyads and triads proved to be highly bioluminescent. Using the bioluminescence signal as the proxy for their structural integrity, we determined that heat‐denatured Nluc dyads and triads can be efficiently refolded by the E. coli Hsp70 chaperone system, which comprises DnaK, DnaJ, and GrpE. In contrast to previous studies with other substrates, we observed that Nluc repeats can be efficiently refolded by DnaK and DnaJ, even in the absence of GrpE co‐chaperone. Taken together, our study offers a new powerful substrate for chaperone research and raises intriguing questions about the Hsp70 mechanisms, particularly in the context of structurally diverse proteins.

As one of the founders of quantum chemistry, Robert G. Parr (RGP) was an influential theoretical chemist of the last few decades. He made enormous scientific contributions to both wave function theory and density functional theory. He was also the founding father of conceptual density functional theory. RGP was both a devoted scientist and a family man. RGP met the love of his life, Jane Bolstad (who passed away on January 26, 2020, at the age of 97), at the University of Minnesota when they were graduate students. They successfully navigated 72 years of marriage with three children, Steven Parr of Tokyo, Jeanne Lemkau and Carol Lachenman, both of Chapel Hill.

We develop an efficient method to compute the electrostatic component of the solvation free energy via the two-point Padé approximation. The Padé approximant uses four parameters to describe the electrostatic free energy change of the solvation process, which could be readily determined from four thermodynamic properties obtained in two simulations, namely, the first- and second-order free energy gradients of any two states. Therefore, instead of sampling at multiple intermediate states, only two states, e.g., electrostatically fully solvated and desolvated, are needed to determine the Padé approximant and compute the corresponding free energy contribution. Applications to several model systems, including both neutral and charged species, show that the method can accurately produce electrostatic solvation free energy. The method would be very useful to save computational cost in applications in which accurate but expensive energy functions like quantum mechanics are used.

Molecular dynamics (MD) is an extremely powerful, highly effective, and widely used approach to understanding the nature of chemical processes in atomic details for proteins. The accuracy of results from MD simulations is highly dependent on force fields. Currently, molecular mechanical (MM) force fields are mainly utilized in MD simulations because of their low computational cost. Quantum mechanical (QM) calculation has high accuracy, but it is exceedingly time consuming for protein simulations. Machine learning (ML) provides the capability for generating accurate potential at the QM level without increasing much computational effort for specific systems that can be studied at the QM level. However, the construction of general machine learned force fields, needed for broad applications and large and complex systems, is still challenging. Here, general and transferable neural network (NN) force fields based on CHARMM force fields, named CHARMM-NN, are constructed for proteins by training NN models on 27 fragments partitioned from the residue-based systematic molecular fragmentation (rSMF) method. The NN for each fragment is based on atom types and uses new input features that are similar to MM inputs, including bonds, angles, dihedrals, and non-bonded terms, which enhance the compatibility of CHARMM-NN to MM MD and enable the implementation of CHARMM-NN force fields in different MD programs. While the main part of the energy of the protein is based on rSMF and NN, the nonbonded interactions between the fragments and with water are taken from the CHARMM force field through mechanical embedding. The validations of the method for dipeptides on geometric data, relative potential energies, and structural reorganization energies demonstrate that the CHARMM-NN local minima on the potential energy surface are very accurate approximations to QM, showing the success of CHARMM-NN for bonded interactions. However, the MD simulations on peptides and proteins indicate that more accurate methods to represent protein–water interactions in fragments and non-bonded interactions between fragments should be considered in the future improvement of CHARMM-NN, which can increase the accuracy of approximation beyond the current mechanical embedding QM/MM level.

We employed the chemical potential equalization principle to demonstrate that fractional electrons are involved in the electro-inductive effect as well as the vibrational Stark effect. By the chemical potential model, we were able to deduce that the frontier molecular orbitals of immobilized molecules can provide valuable insight into these effects. To further understand and quantify these findings, we introduced fractional charge density functional theory (FC-DFT), a canonical ensemble approach for open systems. This method allows for the calculation of electronic energies, nuclear gradients, and the Hessian matrix of fractional electronic systems. To correct the spurious delocalization error commonly found in approximate density functionals for small systems, we imposed the Perdew-Parr-Levy-Balduz (PPLB) condition through linear interpolation of two adjacent integer points (LI-FC-DFT). Although this approach is relatively simple in terms of molecular modeling, the results obtained through LI-FC-DFT calculations predict the same trend seen in experimental reactivity and the frequency change of immobilized molecules.