Olaf Wiest’s research while affiliated with Université Notre Dame d'Haïti and other places

Practical applications of machine learning (ML) to new chemical domains are often hindered by data scarcity. Here we show how data gaps can be circumvented by means of transfer learning that leverages chemically relevant pre-training data. Case studies are presented in which the outcomes of two classes of pericyclic reactions are predicted: [3,3] rearrangements (Cope and Claisen rearrangements) and [4 + 2] cycloadditions (Diels–Alder reactions). Using the graph-based generative algorithm NERF, we evaluate the data efficiencies achieved with different starting models that we pre-trained on datasets of different sizes and chemical scope. We show that the greatest data efficiency is obtained when the pre-training is performed on smaller datasets of mechanistically related reactions (Diels–Alder, Cope and Claisen, Ene, and Nazarov) rather than >50× larger datasets of mechanistically unrelated reactions (USPTO-MIT). These small bespoke datasets were more efficient in both low re-training and low pre-training regimes, and are thus recommended alternatives to large diverse datasets for pre-training ML models.

The rapid advent of machine learning (ML) and artificial intelligence (AI) has catalyzed major transformations in chemistry, yet the application of these methods to spectroscopic and spectrometric data, referred to as Spectroscopy Machine Learning (SpectraML), remains relatively underexplored. Modern spectroscopic techniques (MS, NMR, IR, Raman, UV-Vis) generate an ever-growing volume of high-dimensional data, creating a pressing need for automated and intelligent analysis beyond traditional expert-based workflows. In this survey, we provide a unified review of SpectraML, systematically examining state-of-the-art approaches for both forward tasks (molecule-to-spectrum prediction) and inverse tasks (spectrum-to-molecule inference). We trace the historical evolution of ML in spectroscopy, from early pattern recognition to the latest foundation models capable of advanced reasoning, and offer a taxonomy of representative neural architectures, including graph-based and transformer-based methods. Addressing key challenges such as data quality, multimodal integration, and computational scalability, we highlight emerging directions such as synthetic data generation, large-scale pretraining, and few- or zero-shot learning. To foster reproducible research, we also release an open-source repository containing recent papers and their corresponding curated datasets (https://github.com/MINE-Lab-ND/SpectrumML_Survey_Papers). Our survey serves as a roadmap for researchers, guiding progress at the intersection of spectroscopy and AI.

HMG CoA Reductase catalyzed the interconversion of a thioester, HMG CoA, and mevalonic acid, a key step in the isoprenoid pathway, through a complex reaction mechanism involving three distinct chemical steps with two molecules of cofactor and large-scale rearrangements of the enzyme. Here, we investigate the second step, the formation of a thiohemiacetal from CoA and mevaldehyde, using time resolved crystallography and molecular dynamics (MD) simulations. After triggering the reaction by a pH jump from pH 6.7 to pH9, the formation of the carbon-sulfur bond can be observed in the two structures at 2.5 and 4 minutes. The structures obtained close to the activated complex of the reaction serve as the starting point for MD simulations of different possible protonation states of the catalytically active residues. Changes to the active site geometry, specifically the residues Ser 85, Glu 83 and His 381 that are important for catalysis of the reaction are discussed in detail. This work demonstrates the applicability of the combination of time resolved crystallography using a pH trigger with D simulations to obtain a detailed view of a complex reaction in an enzyme active site.

The synthesis of photocaged substrates of the biologically important enzyme HMG-CoA reductase is reported. HMG-CoA bearing a p-hydroxyphenacyl (pHP) photocage moiety was synthesized in an overall yield of 14% over seven steps in addition to caged forms of mevalonate and mevaldehyde. The absorption maximum and quantum yield for decaging of the photocaged compounds is pH dependent with a lambda (max) = 330 nm and a quantum yield of 5%, respectively, at pH 9.1 but lambda (max) = 290 nm and a quantum yield of 16% at pH 6.7.

Therapeutics that rescue folding, trafficking, and function of disease-causing missense mutants are sought for a host of human diseases, but efforts to leverage model systems to test emerging strategies have met with limited success. Such is the case for Niemann-Pick type C1 disease, a lysosomal disorder characterized by impaired intracellular cholesterol trafficking, progressive neurodegeneration, and early death. NPC1, a multipass transmembrane glycoprotein, is synthesized in the endoplasmic reticulum and traffics to late endosomes/lysosomes, but this process is often disrupted in disease. We sought to identify small molecules that promote folding and enable lysosomal localization and functional recovery of mutant NPC1. We leveraged a panel of isogenic human induced neurons expressing distinct NPC1 missense mutations. We used this panel to rescreen compounds that were reported previously to correct NPC1 folding and trafficking. We established mo56-hydroxycholesterol (mo56Hc) as a potent pharmacological chaperone for several NPC1 mutants. Furthermore, we generated mice expressing human I1061T NPC1, a common mutation in patients. We demonstrated that this model exhibited disease phenotypes and recapitulated the protein trafficking defects, lipid storage, and response to mo56Hc exhibited by human cells expressing I1061T NPC1. These tools established a paradigm for testing and validation of proteostatic therapeutics as an important step towards the development of disease-modifying therapies.

Large Language Models (LLMs) have achieved remarkable success across a wide array of tasks. Due to their notable capabilities in planning and reasoning, LLMs have been utilized as autonomous agents for the automatic execution of various tasks. Recently, LLM-based agent systems have rapidly evolved from single-agent planning or decision-making to operating as multi-agent systems, enhancing their ability in complex problem-solving and world simulation. To offer an overview of this dynamic field, we present this survey to offer an in-depth discussion on the essential aspects and challenges of LLM-based multi-agent (LLM-MA) systems. Our objective is to provide readers with an in-depth understanding of these key points: the domains and settings where LLM-MA systems operate or simulate; the profiling and communication methods of these agents; and the means by which these agents develop their skills. For those interested in delving into this field, we also summarize the commonly used datasets or benchmarks. To keep researchers updated on the latest studies, we maintain an open-source GitHub repository (github.com/taichengguo/LLM_MultiAgents_Survey_Papers), dedicated to outlining the research of LLM-MA research.

Recently, Large Language Models (LLMs) with their strong task-handling capabilities have shown remarkable advancements across a spectrum of fields, moving beyond natural language understanding. However, their proficiency within the chemistry domain remains restricted, especially in solving professional molecule-related tasks. This challenge is attributed to their inherent limitations in comprehending molecules using only common textual representations, i.e., SMILES strings. In this study, we seek to enhance the ability of LLMs to comprehend molecules by designing and equipping them with a multi-modal external module, namely MolX. In particular, instead of directly using a SMILES string to represent a molecule, we utilize specific encoders to extract fine-grained features from both SMILES string and 2D molecular graph representations for feeding into an LLM. Moreover, a human-defined molecular fingerprint is incorporated to leverage its embedded domain knowledge. Then, to establish an alignment between MolX and the LLM's textual input space, the whole model in which the LLM is frozen, is pre-trained with a versatile strategy including a diverse set of tasks. Extensive experimental evaluations demonstrate that our proposed method only introduces a small number of trainable parameters while outperforming baselines on various downstream molecule-related tasks ranging from molecule-to-text translation to retrosynthesis, with and without fine-tuning the LLM.