Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain
the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in
Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles
and JavaScript.
Bolstering the broad and deep applicability of graph neural networks, Heydaribeni et al. introduce HypOp, a framework that uses hypergraph neural networks to solve general constrained combinatorial optimization problems. The presented method scales and generalizes well, improves accuracy and outperforms existing solvers on various benchmarking examples.
Deep learning has led to great advances in predicting protein structure from sequences. Ren and colleagues present here a method for the inverse problem of finding a sequence that results in a desired protein structure, which is inspired by various components of AlphaFold combined with Markov random fields to decode sequences more efficiently.
Achieving the promised advantages of quantum computing relies on translating quantum operations into physical realizations. Fürrutter and colleagues use diffusion models to create quantum circuits that are based on user specifications and tailored to experimental constraints.
Machine learning-based surrogate models are important to model complex systems at a reduced computational cost; however, they must often be re-evaluated and adapted for validity on future data. Diaw and colleagues propose an online training method leveraging optimizer-directed sampling to produce surrogate models that can be applied to any future data and demonstrate the approach on a dense nuclear-matter equation of state containing a phase transition.
Despite the existence of various pretrained language models for nucleotide sequence analysis, achieving good performance on a broad range of downstream tasks using a single model is challenging. Wang and colleagues develop a pretrained language model specifically optimized for RNA sequence analysis and show that it can outperform state-of-the-art methods in a diverse set of downstream tasks.
Large language models can be queried to perform chain-of-thought reasoning on text descriptions of data or computational tools, which can enable flexible and autonomous workflows. Bran et al. developed ChemCrow, a GPT-4-based agent that has access to computational chemistry tools and a robotic chemistry platform, which can autonomously solve tasks for designing or synthesizing chemicals such as drugs or materials.
Methods for predicting molecular structure predictions have so far focused on only the most probable conformation, but molecular structures are dynamic and can change when performing their biological functions, for example. Zheng et al. use a graph transformer approach to learn the equilibrium distribution of molecular systems and show that this can be helpful for a number of downstream tasks, including protein structure prediction, ligand docking and molecular design.
The central assumption in machine learning that data are independent and identically distributed does not hold in many reinforcement learning settings, as experiences of reinforcement learning agents are sequential and intrinsically correlated in time. Berrueta and colleagues use the mathematical theory of ergodic processes to develop a reinforcement framework that can decorrelate agent experiences and is capable of learning in single-shot deployments.
Current limb-driven methods often result in suboptimal prosthetic motions. Kühn and colleagues develop a framework called synergy complement control (SCC) that advances prosthetics by learning ‘cyborg’ limb-driven control, ensuring natural coordination. Validated in diverse trials, SCC offers reliable and intuitive enhancement for limb functionality.
Modelling the statistical and geometrical properties of particle trajectories in turbulent flows is key to many scientific and technological applications. Li and colleagues introduce a data-driven diffusion model that can generate high-Reynolds-number Lagrangian turbulence trajectories with statistical properties consistent with those of the training set and even generalize to rare, intense events unseen during training.
Fragment-based molecular design uses chemical motifs and combines them into bio-active compounds. While this approach has grown in capability, molecular linker methods are restricted to linking fragments one by one, which makes the search for effective combinations harder. Igashov and colleagues use a conditional diffusion model to link multiple fragments in a one-shot generative process.
Recent research has focused on restoring speech in populations with neurological deficits. Chen, Wang et al. develop a framework for decoding speech from neural signals, which could lead to innovative speech prostheses.
Identifying compounds in tandem mass spectrometry requires extensive databases of known compounds or computational methods to simulate spectra for samples not found in databases. Simulating tandem mass spectra is still challenging, and long-range connections in particular are difficult to model for graph neural networks. Young and colleagues use a graph transformer model to learn patterns of long-distance relations between atoms and molecules.
The 5′ untranslated region is a critical regulatory region of mRNA, influencing gene expression regulation and translation. Chu, Yu and colleagues develop a language model for analysing untranslated regions of mRNA. The model, pretrained on data from diverse species, enhances the prediction of mRNA translation activities and has implications for new vaccine design.
Using machine learning methods to model interatomic potentials enables molecular dynamics simulations with ab initio level accuracy at a relatively low computational cost, but requires a large number of labelled training data obtained through expensive ab initio computations. Cui and colleagues propose a geometric learning framework that leverages self-supervised learning pretraining to enhance existing machine learning based interatomic potential models at a negligible additional computational cost.
In early 2023, Bai and colleagues presented DrugBAN, an interpretable method for drug–target prediction. In this Reusability Report, Xu and colleagues reproduce the original findings and provide a careful exploration of cross-domain adaptability.
Generative models for chemical structures are often trained to create output in the common SMILES notation. Michael Skinnider shows that training models with the goal of avoiding the generation of incorrect SMILES strings is detrimental to learning other chemical properties and that allowing models to generate incorrect molecules, which can be easily removed post hoc, leads to better performing models.
AI methods can discover new antibiotics but existing methods have limitations. Swanson et al. develop a generative AI model that learns to design molecules that are easy to synthesize. The authors apply the model to design and validate novel antibiotics against the bacterial pathogen Acinetobacter baumannii.
Foundation models have transformed artificial intelligence by training on vast amounts of broad unlabelled data. Pai et al. present a foundation model leading to more accurate, efficient and robust cancer imaging biomarkers, especially in use cases with small training datasets.
Deep learning generative approaches have been used in recent years to discover new molecules with drug-like properties. To improve the performance of such approaches, Yang et al. add chemical binding knowledge to a deep generative framework and demonstrate, including by wet-lab verification, that the method can find valid molecules that successfully bind to target proteins.