2024-2025

Institute for Materials Chemistry and Engineering, Kyushu University

Structural fluctuations and conformational transitions of proteins have been realized to be essential for function by recent experiments such as single-molecule and NMR relaxation dispersion measurements. However, understanding how these conformational dynamics contribute to function has been challenging and often remains controversial. Molecular dynamics simulations can be a powerful tool for dissecting the dynamics of proteins. Yet, computational studies often focus on the static perspective by looking at the free energy profiles and rate constants along a reaction coordinate chosen by chemical intuition. Furthermore, defining the appropriate reaction coordinate is often challenging. In this talk, I will discuss how the dynamics of proteins, which are hierarchical by nature, occur and contribute to enzyme catalysis. The key is that chemistry occurs locally and rapidly (but are rare!), while protein dynamics are collective and slow [1,2]. Due to this spatiotemporal mismatch, enzymatic reaction dynamics deviate strikingly from the minimum free energy path [3]. I will also review how we to obtain an optimal reaction coordinate for better free energy surface, and present our group’s developments in this direction incorporating machine learning techniques [4,5].

1. T. Mori, S. Saito, J. Phys. Chem. Lett. 10, 474–480 (2019).
2. T. Mori, S. Saito, J. Phys. Chem. B 126, 5185–5193 (2022).
3. T. Mori, S. Saito, J. Chem. Theory Comput. 16, 3396–3407 (2020).
4. T. Kikutsuji et al., J. Chem. Phys. 156, 154108 (2022).
5. K. Kawashima et al., submitted (arXiv:2408.02132)

The field of computational chemical science cannot ignore the growing need for accurate electronic structure methods that transcend the framework of the Schrödinger equation. Fundamental to this need are molecular and material systems with complex electronic structures and photophysical properties that cannot be predicted by ordinary periodic trends. Despite significant advancements, most current electronic structure methods struggle to accurately predict energetic ordering, chemical reactivity, and spectroscopic features for complexes containing late-row transition metals, rare earth, or heavy elements without incorporating relativistic effects. Unlike the non-relativistic Schrödinger framework, where spin is a good quantum number and the Coulomb operator is the main source of particle-particle interactions, relativistic methods, based on the Dirac equation and quantum field theory, introduce additional operators and considerations, such as spin-orbit and spin-spin couplings, within a multi-component framework. This presentation offers a pedagogical perspective on the evolution and current state of relativistic electronic structure theory, illustrated with practical examples. By contrasting the non-relativistic and relativistic frameworks, it highlights the significance of various relativistic corrections for accurate chemical modeling. Perturbative and variational techniques to incorporate relativistic operators are compared and analyzed. Finally, the talk explores the challenges and opportunities shaping the future of relativistic electronic structure theory in the modern era.

Recent advances in relativistic electronic structure theory have unlocked new insights into chemical processes through molecular spectroscopic techniques. Relativistic spectroscopies, which explore the optical and magnetic properties of core electrons (e.g., XANES, RIXS, XES) and the magnetic responses of chemical systems (e.g., MCD, XMCD), have become instrumental in these studies. Furthermore, spin-driven phenomena, such as intersystem crossing, spin echo, and spin entanglement, demand computational methods grounded in the relativistic framework. In this presentation, I will discuss our recent developments in relativistic quantum dynamics and their applications in studying conical intersections, chiral-induced spin selectivity, and M-edge spectra of heavy-element complexes. These methodological innovations are poised to transform emerging scientific fields, with potentially revolutionary impacts on research.

University of California, Irvine

Novel X-ray pulse sources from free-electron lasers and high-harmonic generation setups enable the monitoring of elementary events molecular such as the ultrafast passage through conical intersections on unprecedented temporal, spatial and energetic scales. The attosecond duration of X-ray pulses, their large bandwidth, and the atomic selectivity of core X-ray excitations offer new windows into photochemical processes. We show how the Orbital Angular Momentum of twisted X-ray light can be leveraged to detect vibronic coherences and time evolving chirality emerging at conical intersections due to the bifurcation of molecular wavepackets. Employing quantum light in multidimensional spectroscopy is opening up many exciting opportunities to enhance the signal-to-noise ratio, improve the combined temporal, spatial, and spectral resolutions, and simplify nonlinear optical signals by selecting desired transition pathways in second and third order signals. We show how photoelectron signals generated by time-energy entangled photon pairs can monitor ultrafast excited state dynamics of molecules with high joint spectral and temporal resolutions, not subjected to the Fourier uncertainty limitation of classical light. Two-entangled-photon absorption scales linearly with the pump intensity, allowing the study of fragile biological samples with low photon fluxes, and quantum interferometry can enhance the signals. Optical cavities provide another means for controlling the photophysics of molecules by making use of strong light–matter coupling without employing strong external laser pulses. A quantum dynamical study of charge migration in molecules in an optical cavity demonstrates how to trigger and manipulate charge migration pathways that are originally inactivated or suppressed in the bare molecule.

Department of Chemistry, University of California, Berkeley, & Chemical Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA

Core spectroscopy is undergoing rapid experimental progress, associated with new light sources, and improvements in time resolution. Together with the inherent element-specificity and atomic site specificity of x-ray absorption spectroscopy (XAS), this makes UV-visible pump / XAS probe experiments a powerful tool to follow reactive chemical dynamics in real time. There is accordingly a growing need for efficient and reliable simulation methods to predict XAS spectra of molecules in their ground and excited states using quantum chemistry methods.

This talk will discuss recent progress made in my group in developing and applying low-scaling electronic structure theory approaches to simulating XAS spectra. We will consider both wavefunction and density functional theory (DFT) approaches. First we will discuss some of the challenges that are faced by standard methods. In particular, standard linear response time-dependent DFT (TDDFT) exhibits very serious problems for XAS. The origin is in the standard adiabatic approximation, which leads to lack of orbital relaxation (in common with charge-transfer excitations). I will then discuss two alternatives that show considerable promise.

First is the use of state-specific orbital-optimized DFT (OO-DFT), which corresponds to finding saddle points of a ground state functional, without making the adiabatic approximation or doing linear response at all. A series of tests and examples shows that OO-DFT resolves most of the major problems of TDDFT for modeling XAS in terms of quality of results, although it is not as convenient to use because of the need to simulate each individual state that contributes to a given spectrum. Second is the possibility of resurrecting the ease of use of TDDFT, whilst correcting the major problems associated with the adiabatic approximation. This approach, when viewed from a wavefunction perspective, can be naturally extended to treat XAS of valence excited states, which I will also describe.

To show the promise of these approaches for modeling chemistry, some results from recent collaborations with experimental groups will be shown, illustrating the promise of observing (and simulating) reactive chemistry with time-resolved XAS.

Department of Chemistry, Emory University, Atlanta, GA

Machine learning (ML) and big data play increasingly critical roles in chemical discovery. However, datasets (both computational and experimental) and ML models for condensed-phase molecular systems, such as solvated molecules and molecule assemblies, remain scarce. My research group leverages GPU-accelerated quantum chemistry and machine learning to address these gaps.

For solvated molecules, we developed GPU-accelerated algorithms for implicit solvent and high-pressure models to achieve efficient quantum mechanical descriptions. To account for strong solute-solvent interactions that implicit solvent models cannot capture, we developed AutoSolvate. This open-source toolkit streamlines the classical and QM/MM simulation workflow for explicitly solvated molecules. To further enhance accessibility, we launched AutoSolvateWeb, a chatbot-assisted, cloud-based platform that automates simulation setup and execution using cloud resources. These tools have enabled the efficient generation of diverse computational datasets for solvated molecules. Leveraging these datasets, we trained Δ-ML models to enhance the accuracy of low-cost computational methods against experimental measurements. We also developed explainable ML models to uncover design principles for catalysts.

For molecular assemblies, we addressed computational challenges in predicting excited-state properties. We developed a size-transferable machine-learned exciton model that significantly reduces computational costs without sacrificing accuracy. Trained solely on a set of dimer exciton Hamiltonians, it can predict the exciton Hamiltonians of arbitrarily sized homogeneous aggregates containing up to 50 monomers.

Departments of Chemistry & Physics, University of California, Santa Barbara, Santa Barbara, CA

Tau is an intrinsically disordered protein that plays an important role in stabilizing microtubules. Under pathological conditions, this protein can also self-assemble into fibrillar structures, a process that has been associated with a class of neurodegenerative diseases known as Tauopathies. Interestingly, this protein is also capable of assembling into liquid droplets through a process of liquid-liquid phase separation (LLPS).

In the first part of my talk, I will introduce the methods that we use to look at protein self-assembly into fibrillar and droplet structures. These include field-theoretic simulations, coarse-grained models, and atomistic simulations. In the second part of my talk, I will discuss the application of these methods to Tau assembly. I will present studies on fragments of Tau that have a propensity to either phase separate or form fibrils, and I will discuss the sequence characteristics linked to these two modes of assembly. Next, I will introduce a 19-residue fragment of Tau that is capable of seeding the fibrillization of full-length Tau, and I will discuss the effect of point mutations in modulating aggregation in familial forms of Tauopathies.

Professor of Applied Physics, Physics and Materials Science, California Institute of Technology, Pasadena, CA

Abstract TBA

Microsoft Research AI4Science & Professor at Freie Universität Berlin

Abstract TBA

Biomedical Engineering, Washington University in St. Louis, St. Louis, MO

Abstract TBA

Institute of Physics, University of Freiburg, Freiburg im Breisgau, Baden-Württemberg, Germany

Abstract TBA

Departments of Chemistry & Physics, University of Illinois at Urbana-Champaign, Champaign county, IL

Abstract TBA

Department of Chemistry, Northwestern University, Evanston, IL

Abstract TBA

Institute of Theoretical Chemistry, University of Vienna, Vienna, Austria

Abstract TBA