arXiv subjects

Daniel J. Cole

Publications and source records attributed to Daniel J. Cole.

15 recordsLinked to original sources

arXiv2025

MACE-OFF: Transferable Short Range Machine Learning Force Fields for Organic Molecules

Classical empirical force fields have dominated biomolecular simulation for over 50 years. Although widely used in drug discovery, crystal structure prediction, and biomolecular dynamics, they generally lack the accuracy and transferability required for first-principles predictive modeling. In this paper, we introduce MACE-OFF, a series of short range transferable force fields for organic molecules created using state-of-the-art machine learning technology and first-principles reference data computed with a high level of quantum mechanical theory. MACE-OFF demonstrates the remarkable capabilities of short range models by accurately predicting a wide variety of gas and condensed phase properties of molecular systems. It produces accurate, easy-to-converge dihedral torsion scans of unseen molecules, as well as reliable descriptions of molecular crystals and liquids, including quantum nuclear effects. We further demonstrate the capabilities of MACE-OFF by determining free energy surfaces in explicit solvent, as well as the folding dynamics of peptides.Finally, we simulate a fully solvated small protein, observing accurate secondary structure and vibrational spectrum. These developments enable first-principles simulations of molecular systems for the broader chemistry community at high accuracy and relatively low computational cost.

physics.chem-ph↗

arXiv2024

On the design space between molecular mechanics and machine learning force fields

A force field as accurate as quantum mechanics (QM) and as fast as molecular mechanics (MM), with which one can simulate a biomolecular system efficiently enough and meaningfully enough to get quantitative insights, is among the most ardent dreams of biophysicists -- a dream, nevertheless, not to be fulfilled any time soon. Machine learning force fields (MLFFs) represent a meaningful endeavor towards this direction, where differentiable neural functions are parametrized to fit ab initio energies, and furthermore forces through automatic differentiation. We argue that, as of now, the utility of the MLFF models is no longer bottlenecked by accuracy but primarily by their speed (as well as stability and generalizability), as many recent variants, on limited chemical spaces, have long surpassed the chemical accuracy of $1$ kcal/mol -- the empirical threshold beyond which realistic chemical predictions are possible -- though still magnitudes slower than MM. Hoping to kindle explorations and designs of faster, albeit perhaps slightly less accurate MLFFs, in this review, we focus our attention on the design space (the speed-accuracy tradeoff) between MM and ML force fields. After a brief review of the building blocks of force fields of either kind, we discuss the desired properties and challenges now faced by the force field development community, survey the efforts to make MM force fields more accurate and ML force fields faster, envision what the next generation of MLFF might look like.

physics.chem-ph↗

arXiv2024

Computing solvation free energies of small molecules with experimental accuracy

Free energies play a central role in characterising the behaviour of chemical systems and are among the most important quantities that can be calculated by molecular dynamics simulations. Solvation free energies in various organic solvents, in particular, are well-studied physicochemical properties of drug-like molecules and are commonly used to assess and optimise the accuracy of nonbonded parameters in empirical forcefields, and also as a fast-to-compute surrogate of performance for protein-ligand binding free energy estimation. Machine learned potentials (MLPs) show great promise as more accurate alternatives to empirical forcefields, but are not readily decomposed into physically motivated functional forms, which has thus far rendered them incompatible with standard alchemical free energy methods that manipulate individual pairwise interaction terms. However, since the accuracy of free energy calculations is highly sensitive to the forcefield, this is a key area in which MLPs have the potential to address the shortcomings of empirical forcefields. In this work, we introduce an efficient alchemical free energy protocol that enables calculations of rigorous free energy differences in condensed phase systems modelled entirely by MLPs. Using a pretrained, transferrable, alchemically equipped MLP model, we demonstrate sub-chemical accuracy for the solvation free energies of a wide range of organic molecules.

physics.chem-ph↗

arXiv2023

Riemannian Geometry and Molecular Similarity II: Kähler Quantization

Shape-similarity between molecules is a tool used by chemists for virtual screening, with the goal of reducing the cost and duration of drug discovery campaigns. This paper reports an entirely novel shape descriptor as an alternative to the previously described RGMolSA descriptors \cite{cole2022riemannian}, derived from the theory of Riemannian geometry and Kähler quantization (KQMolSA). The treatment of a molecule as a series of intersecting spheres allows us to obtain the explicit \textit{Riemannian metric} which captures the geometry of the surface, which can in turn be used to calculate a Hermitian matrix $\mathbb{M}$ as a directly comparable surface representation. The potential utility of this method is demonstrated using a series of PDE5 inhibitors considered to have similar shape. The method shows promise in its capability to handle different conformers, and compares well to existing shape similarity methods. The code and data used to produce the results are available at: \url{https://github.com/RPirie96/KQMolSA}.

math.DG↗

arXiv2022

Riemannian Geometry and Molecular Surfaces I: Spectrum of the Laplacian

Ligand-based virtual screening aims to reduce the cost and duration of drug discovery campaigns. Shape similarity can be used to screen large databases, with the goal of predicting potential new hits by comparing to molecules with known favourable properties. This paper presents the theory underpinning RGMolSA, a new alignment-free and mesh-free surface-based molecular shape descriptor derived from the mathematical theory of Riemannian geometry. The treatment of a molecule as a series of intersecting spheres allows the description of its surface geometry using the Riemannian metric, obtained by considering the spectrum of the Laplacian. This gives a simple vector descriptor constructed of the weighted surface area and eight non-zero eigenvalues, which capture the surface shape. We demonstrate the potential of our method by considering a series of PDE5 inhibitors that are known to have similar shape as an initial test case. RGMolSA displays promise when compared to existing shape descriptors and in its capability to handle different molecular conformers. The code and data used to produce the results are available via GitHub: https://github.com/RPirie96/RGMolSA.

q-bio.QM↗

arXiv2020

ONETEP + TOSCAM: uniting dynamical mean field theory and linear-scaling density functional theory

We introduce the unification of dynamical mean field theory (DMFT) and linear-scaling density functional theory (DFT), as recently implemented in ONETEP, a linear-scaling DFT package, and TOSCAM, a DMFT toolbox. This code can account for strongly correlated electronic behavior while simultaneously including the effects of the environment, making it ideally suited for studying complex and heterogeneous systems containing transition metals and lanthanides, such as metalloproteins. We systematically introduce the necessary formalism, which must account for the non-orthogonal basis set used by ONETEP. In order to demonstrate the capabilities of this code, we apply it to carbon monoxide-ligated iron porphyrin and explore the distinctly quantum-mechanical character of the iron $3d$ electrons during the process of photodissociation.

cond-mat.str-el↗

arXiv2019

Super-exchange mechanism and quantum many body excitations in the archetypal hemocyanin/tyrosinase di-Cu oxo-bridge

We perform first-principles quantum mechanical studies of dioxygen ligand binding to the hemocyanin protein. Electronic correlation effects in the functional site of hemocyanin are investigated using a state-of-the-art approach, treating the localised copper 3\emph{d} electrons with cluster dynamical mean field theory (DMFT) for the first time. This approach has enabled us to account for dynamical and multi-reference quantum mechanics, capturing valence and spin fluctuations of the 3\emph{d} electrons. Our approach explains the stabilisation of the experimentally observed di-Cu singlet for the butterflied Cu$_2$O$_2$ core, with localised charge and incoherent scattering processes across the oxo-bridge that prevent long-lived charge excitations, suggesting that the magnetic structure of hemocyanin is largely influenced by the many-body corrections. Our computational model is supported by agreement with experimental optical absorption data, and provides a revised understanding of the bonding of the peroxide to the di-Cu system \emph{in vivo}.

cond-mat.str-el↗

arXiv2018

The role of spin in the calculation of Hubbard $U$ and Hund's $J$ parameters from first principles

The density functional theory (DFT)+$U$ method is a pragmatic and effective approach for calculating the ground-state properties of strongly-correlated systems, and linear response calculations are widely used to determine the requisite Hubbard parameters from first principles. We provide a detailed treatment of spin within this linear response approach, demonstrating that the conventional Hubbard $U$ formula, unlike the conventional DFT+$U$ corrective functional, incorporates interactions that are off-diagonal in the spin indices and places greater weight on one spin channel over the other. We construct alternative definitions for Hubbard and Hund's parameters that are consistent with the contemporary DFT+$U$ functional, expanding upon the minimum-tracking linear response method. This approach allows Hund's $J$ and spin-dependent $U$ parameters to be calculated with the same ease as for the standard Hubbard $U$. Our methods accurately reproduce the experimental band gap, local magnetic moments, and the valence band edge character of manganese oxide, a canonical strongly-correlated system. We also apply our approach to a complete series of transition-metal complexes [M(H$_2$O)$_6$]$^{n+}$ (for M = Ti to Zn), showing that Hubbard corrections on oxygen atoms are necessary for preserving bond lengths, and demonstrating that our methods are numerically well-behaved even for near-filled subspaces such as in zinc. However, spectroscopic properties appear beyond the reach of the standard DFT+$U$ approach. Collectively, these results shed new light on the role of spin in the calculation of the corrective parameters $U$ and $J$, and point the way towards avenues for further development of DFT+$U$-type methods.

cond-mat.str-el↗

arXiv2014

Constrained geometric dynamics of the Fenna-Matthews-Olson complex: The role of correlated motion in reducing uncertainty in excitation energy transfer

The Fenna Mathews Olson (FMO) complex of green sulphur bacteria is an example of a photosynthetic pigment protein complex, in which the electronic properties of the pigments are modified by the protein environment to promote efficient excitonic energy transfer from antenna complexes to the reaction centres. Many of the electronic properties of the FMO complex can be extracted from knowledge of the static crystal structure. However, the recent observation and analysis of long lasting quantum dynamics in the FMO complex point to protein dynamics as a key factor in protecting and generating quantum coherence under laboratory conditions. While fast inter and intra molecular vibrations have been investigated extensively, the slow dynamics which effectively determine the optical inhomogeneous broadening of experimental ensembles has received less attention. Our study employs constrained geometric dynamics to study the flexibility in the protein network by efficiently generating the accessible conformational states from the published crystal structure. Statistical and principle component analysis reveal highly correlated low frequency motions between functionally relevant elements, including strong correlations between pigments that are excitonically coupled. Our analysis reveals a hierarchy of structural interactions which enforce these correlated motions, from the level of monomer monomer interfaces right down to the alpha helices, beta sheets and pigments. In addition to inducing strong spatial correlations across the conformational ensemble, we find that the overall rigidity of the FMO complex is exceptionally high. We suggest that these observations support the idea of highly correlated inhomogeneous disorder of the electronic excited states, which is further supported by the remarkably low variance of the excitonic couplings of the conformational ensemble.

q-bio.BM↗

arXiv2014

Renormalization of myoglobin-ligand binding energetics by quantum many-body effects

We carry out a first-principles atomistic study of the electronic mechanisms of ligand binding and discrimination in the myoglobin protein. Electronic correlation effects are taken into account using one of the most advanced methods currently available, namely a linear-scaling density functional theory (DFT) approach wherein the treatment of localized iron 3d electrons is further refined using dynamical mean-field theory (DMFT). This combination of methods explicitly accounts for dynamical and multi-reference quantum physics, such as valence and spin fluctuations, of the 3d electrons, whilst treating a significant proportion of the protein (more than 1000 atoms) with density functional theory. The computed electronic structure of the myoglobin complexes and the nature of the Fe-O2 bonding are validated against experimental spectroscopic observables. We elucidate and solve a long standing problem related to the quantum-mechanical description of the respiration process, namely that DFT calculations predict a strong imbalance between O2 and CO binding, favoring the latter to an unphysically large extent. We show that the explicit inclusion of many body-effects induced by the Hund's coupling mechanism results in the correct prediction of similar binding energies for oxy- and carbonmonoxymyoglobin.

physics.chem-ph↗

arXiv2013

Ligand Discrimination in Myoglobin from Linear-Scaling DFT+U

Myoglobin modulates the binding of diatomic molecules to its heme group via hydrogen-bonding and steric interactions with neighboring residues, and is an important benchmark for computational studies of biomolecules. We have performed calculations on the heme binding site and a significant proportion of the protein environment (more than 1000 atoms) using linear-scaling density functional theory and the DFT+U method to correct for self-interaction errors associated with localized 3d states. We confirm both the hydrogen-bonding nature of the discrimination effect (3.6 kcal/mol) and assumptions that the relative strain energy stored in the protein is low (less than 1 kcal/mol). Our calculations significantly widen the scope for tackling problems in drug design and enzymology, especially in cases where electron localization, allostery or long-ranged polarization influence ligand binding and reaction.

physics.chem-ph↗

arXiv2009

Water structuring and collagen adsorption at hydrophilic and hydrophobic silicon surfaces

The adsorption of a collagen fragment on both a hydrophobic, hydrogen-terminated and a hydrophilic, natively oxidised Si surface is investigated using all-atom molecular dynamics. While favourable direct protein-surface interactions via localised contact points characterise adhesion to the hydrophilic surface, evenly spread surface/molecule contacts and stabilisation of the helical structure occurs upon adsorption on the hydrophobic surface. In the latter case, we find that adhesion is accompanied by a mutual fit between the hydrophilic/hydrophobic pattern within the protein and the layered water structure at the solid/liquid interface, which may provide an additional driving force to the classic hydrophobic effect.

cond-mat.soft↗

arXiv2009

Stress-driven oxidation chemistry of wet silicon surfaces

The formation of a hydroxylated native oxide layer on Si(001) under wet conditions is studied by means of first principles molecular dynamics simulations. Water molecules are found to adsorb and dissociate on the oxidised surface leading to rupture of Si-O bonds and producing reactive sites for attack by dissolved dioxygen or hydrogen peroxide molecules. Tensile strain is found to enhance the driving force for the dissociative adsorption of water, suggesting that similar reactions could be responsible for environmentally-driven sub-critical crack propagation in silicon.

cond-mat.mtrl-sci↗

arXiv2008

Stress Development and Impurity Segregation during Oxidation of the Si(100) Surface

We have studied the segregation of P and B impurities during oxidation of the Si(100) surface by means of combined static and dynamical first-principles simulations based on density functional theory. In the bare surface, dopants segregate to chemically stable surface sites or to locally compressed subsurface sites. Surface oxidation is accompanied by development of tensile surface stress up to 2.9 N/m at a coverage of 1.5 monolayers of oxygen and by formation of oxidised Si species with charges increasing approximately linearly with the number of neighbouring oxygen atoms. Substitutional P and B defects are energetically unstable within the native oxide layer, and are preferentially located at or beneath the Si/SiOx interface. Consistently, first-principles molecular dynamics simulations of native oxide formation on doped surfaces reveal that dopants avoid the formation of P-O and B-O bonds, suggesting a surface oxidation mechanism whereby impurities remain trapped at the Si/SiOx interface. This seems to preclude a direct influence of impurities on the surface electrostatics and, hence, on the interactions with an external environment.

cond-mat.mtrl-sci↗

arXiv2008

Development of a Classical Force Field for the Oxidised Si Surface: Application to Hydrophilic Wafer Bonding

We have developed a classical two- and three-body interaction potential to simulate the hydroxylated, natively oxidised Si surface in contact with water solutions, based on the combination and extension of the Stillinger-Weber potential and of a potential originally developed to simulate SiO2 polymorphs. The potential parameters are chosen to reproduce the structure, charge distribution, tensile surface stress and interactions with single water molecules of a natively oxidised Si surface model previously obtained by means of accurate density functional theory simulations. We have applied the potential to the case of hydrophilic silicon wafer bonding at room temperature, revealing maximum room temperature work of adhesion values for natively oxidised and amorphous silica surfaces of 97 mJ/m2 and 90mJ/m2, respectively, at a water adsorption coverage of approximately 1 monolayer. The difference arises from the stronger interaction of the natively oxidised surface with liquid water, resulting in a higher heat of immersion (203 mJ/m2 vs. 166 mJ/m2), and may be explained in terms of the more pronounced water structuring close to the surface in alternating layers of larger and smaller density with respect to the liquid bulk. The computed force-displacement bonding curves may be a useful input for cohesive zone models where both the topographic details of the surfaces and the dependence of the attractive force on the initial surface separation and wetting can be taken into account.

cond-mat.mtrl-sci↗