Themed collection Most popular 2025 machine learning and automation articles

Spectroscopy enables studying matter via its interaction with electromagnetic radiation, supporting analysis, with machine learning further advancing its capabilities.

Through artificial intelligence and robotics, autonomous labs are transforming chemical and materials research by enabling high-throughput, data-driven experiments with minimal human input.

This review examines current deep learning methods for structure-based drug discovery, detailing how various representations of protein pockets can be integrated to encode structural information and design new molecules.

Imbalanced data, where certain classes are significantly underrepresented in a dataset, is a widespread machine learning (ML) challenge across various fields of chemistry, yet it remains inadequately addressed.

This review examines the roles of large language models (LLMs) and autonomous agents in chemistry, exploring advancements in molecule design, property prediction, and synthesis automation.

Experimental ¹H NMR and IR spectra can be scored against calculated data to verify candidate molecules. We show that combining these techniques is significantly more powerful for automated structure verification than using either one individually.

A machine learning framework predicts suitable agents, temperature, and equivalence ratios for reactants and agents. The model consistently outperforms strong baselines, enabling more complete and automation-ready reaction protocols.

Combining the FORMED database with a generative model and the prediction of excited state propertoes, we generate molecular candidates for singlet fission (SF). Amidst known candidates, we find a promising neocoumarin (2-benzopyran-3-one) scaffold.

AIQM2's high speed, competitive accuracy, and robustness enable organic reaction simulations beyond what is possible with the popular DFT methods. It can be used for TS structure search and reactive dynamics, often with chemical accuracy.

IDOLpro is a modular framework for guided diffusion which can generate molecules with a plurality of optimized properties for structure-based drug design, accelerating the drug discovery process.

Multi-modal transfer learning for predicting and explaining universal properties of organic crystals.

NMRExtractor is a large language model-powered pipeline that automatically extracts experimental NMR data from massive open-access publications, resulting in the construction of NMRBank—the largest open-access NMR dataset available to date.

Machine learned interatomic potentials (MLIPs) are reshaping computational chemistry practices because of their ability to drastically exceed the accuracy-length/time scale tradeoff.

PepINVENT introduces a generative model for designing peptides that extend beyond natural amino acids, enabling non-traditional peptide discovery and optimization.

GPT-based generative modeling of MD trajectories enables efficient prediction of state transitions by capturing long-range correlations, offering accurate kinetic and thermodynamic forecasts for diverse physicochemical systems.

This work presents automated non-linear workflows for studying problems in low-data regimes alongside traditional linear models.

Virtual agents and cloud computing have enabled chemists to easily access automated simulations of explicitly solvated molecules.

We propose Grappa, a machine learned molecular mechanics force field for proteins. Grappa, operating on the molecular graph, accurately predicts energies and forces and agrees with experimental data such as J-couplings and folding free energies.