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.
We often encounter mental conflict in our lives. Such mental conflict has long been regarded as subjective. However, a machine learning method can be used to quantify the temporal dynamics of conflict between reward and curiosity from behavioral time-series.
An image-inspired deep-learning model is developed to generate realistic de novo protein structures and scaffolds around functional sites, which helps the search for new structures and functions in protein engineering.
A graph neural network — GAME-Net — has been developed to predict the adsorption energy of organic molecules on metal surfaces, which is a key descriptor of heterogeneous catalytic activity. This method allows for the study of large molecules derived from raw materials such as plastic waste, avoiding the use of costly and time-intensive first-principles simulations.
While digital twins have been recently used to represent cities and their physical structures, integrating complexity science into the digital twin approach will be key to deliver more explicable and trustworthy models and results.
Two computational methods — one physics-based, and the other one deep-learning based — are proposed to enable the systematic investigation of magnetic order in moiré magnets from first principles.
A rotational and time-reversal equivariant neural network designed to represent the spin–orbital density functional theory Hamiltonian as a function of the atomic and magnetic structure enables ab initio electronic-structure calculations of magnetic superstructures. These calculations can efficiently and accurately predict subtle magnetic effects in various chemical environments.
A machine learning algorithm has been developed to capture and analyze rare molecular processes, revealing how molecules self-organize and function. The algorithm is general and can be applied whenever a dynamic system has a notion of ‘likely fate’.
Identifying the origins and nature of non-genetic heterogeneity in cancer has widespread clinical ramifications. In this Review, the authors discuss how computational models and tools have been used to provide insights into this phenomenon and how they can help tackle the disease in the future.
Discovering biological patterns from omics data is challenging due to the high dimensionality of biological data. A computational framework is presented to more efficiently calculate correlations among omics features and to build networks by estimating important connections.
Proton-coupled electron transfer occurs at a variety of length and time scales and often in complex environments. This Perspective summarizes a range of modeling strategies that can be used together to address remaining challenges and provide a better understanding of such reactions.
The computational characterization of short-range order in compositionally complex materials relies on effective interatomic potentials. In this Review, challenges and opportunities in developing advanced potentials for such systems are discussed, with a focus on machine learning-based potentials.
Machine learning models have been widely applied to boost the computational efficiency of searching vast chemical space of compositionally complex materials. This Perspective summarizes the recent developments and proposes future opportunities, such as the physics-informed machine learning models.
Complex materials offer promises for exotic materials properties that enable novel applications. Nevertheless, there are numerous computational challenges for a rational design of defects in such materials, thus inspiring opportunities for developing advanced defect models.
This work involved the design of a multi-view manifold learning algorithm that capitalizes on various types of structure in high-dimensional time-series data to model dynamic signals in low dimensions. The resulting embeddings of human functional brain imaging data unveil trajectories through brain states that predict cognitive processing during diverse experimental tasks.
We present a computational method to generate a single-cell-resolution model of human brain regions starting from microscopy images. The developed method has been benchmarked to reconstruct the CA1 region of a right human hippocampus, including anatomical cell organization, connectivity, and network activity.
We propose a minimal and analytically tractable class of neural networks, the adaptive Ising class. By inferring the model’s parameters from resting-state brain activity recordings, we show that scale-specific oscillations and scale-free avalanches can coexist in resting brains close to a non-equilibrium critical point at the onset of self-sustained oscillations.
A biasing potential is derived from the uncertainty of a neural network ensemble and used to modify the potential energy surface in molecular dynamics simulations and facilitate the determination of underrepresented structural regions.
A computational tool has been developed for the multiscale design of open disordered material systems, bridging network science, computational materials, and wave physics.
Inferring gene networks from discrete RNA counts across cells remains a complex problem. Following Bayesian non-parametrics, a computational framework is proposed to perform non-biased inference of transcription kinetics from single-cell RNA counting experiments.