Reviews & Analysis

We developed a machine learning method that consistently and accurately identified dominant patterns of disease progression in amyotrophic later sclerosis (ALS), Alzheimer’s disease and Parkinson’s disease. Of note, the model was able to identify nonlinear progression trajectories in ALS, a finding that has clinical implications for patient stratification and clinical trial design.

A new Bayesian analysis of remote work data supports one of the oldest theories in social networks, with fresh implications for the future of work environments.

A systematic procedure is reported for calculating effective carrier lifetimes in semiconductor crystals from first-principles calculations. Consideration of three major recombination mechanisms and the use of realistic carrier and defect densities is key in resolving the discrepancy between experimental measurements and lifetimes calculated from nonadiabatic molecular dynamics simulations.

An algorithm is presented for the simulation of reaction–diffusion systems on complex geometries, providing insight on how the interplay of cell geometry and biochemistry can control polarity in living cells.

Multi-messenger astronomy offers promises for exploring Universe events in distance. Nevertheless, there are numerous computational challenges when analyzing the massive heterogeneous messenger data from various detectors, creating research opportunities to the community, such as developing multimodal machine learning.

Determining how information flows throughout a network of interconnected components is a challenging task in many scientific domains. A framework is presented to deconstruct the flow of signals that are transmitted across any two areas (such as brain areas) and define how each area represents these signals.

A machine learning method is developed and used to predict the adsorption configurations and energies of complex molecules at the surfaces of transition metals and alloys. This method will be useful for investigating complex reaction networks at complex catalyst materials to understand and improve the performance of heterogeneous catalysts.

The problem of automatically determining state variables for physical systems is challenging, but essential in the modeling process of almost all scientific and engineering processes. A deep neural network-based approach is proposed to find state variables for systems whose data are given as video frames.

Quantum embedding theory promises the simulation of realistic materials in quantum computers. In this Perspective, challenges and opportunities of applying different embedding frameworks to calculate solid materials properties are discussed, with a focus on electronic structures of spin defects.

The design of protein sequences that can precisely fold into pre-specified 3D structures is a challenging task. A recently proposed deep-learning algorithm improves such designs when compared with traditional, physics-based protein design approaches.

Inspired by active learning approaches, we have developed a computational method that selects minimal gene sets capable of reliably identifying cell-types and transcriptional states in large sets of single-cell RNA-sequencing data. As the procedure focuses computational resources on poorly classified cells, active support vector machine (ActiveSVM) scales to data sets with over one million cells.

A cell clustering model for spatial transcripts that uses cell embedding obtained by graph neural networks can be applied to datasets from multiple platforms for cell type or subpopulation identification and further analysis of the spatial microenvironment.

A graph neural network-based tool is introduced to perform unsupervised cell clustering using spatially resolved transcriptomics data that can uncover cell identities, interactions, and spatial organization in tissues and organs.

Machine learning has been used to accelerate the simulation of fluid dynamics. However, despite the recent developments in this field, there are still challenges to be addressed by the community, a fact that creates research opportunities.

A deep neural network method is developed to learn the density functional theory (DFT) Hamiltonian as a function of atomic structure. This approach provides a solution to the accuracy–efficiency dilemma of DFT and opens opportunities to investigate large-scale materials, such as twisted van der Waals materials.

Aptamers are expected to be next-generation drugs, but identifying candidate aptamers is a challenging task given the large search space. Now, an artificial intelligence (AI)-powered tool called RaptGen is proposed for improving the successful identification of aptamer sequences.

Characterizing the brain’s connectome at multiple scales is essential for unraveling fundamental principles of cortical information processing and how it impacts behavior. A GPU-based implementation for connectome pruning is proposed, achieving greater than 100-fold speedups over previous CPU-based implementations.

The identification of robust and generalizable biomarkers based on microbial abundance data is a challenging task. An algorithm shows an enhanced classification performance by quantifying shifts in microbial co-abundances.

Variational Monte Carlo is one of the most accurate methods to solve the many-electron Schrödinger equation, but suffers from high computational cost. A recent study uses a weight-sharing technique to accelerate the neural network-based variational Monte Carlo method, allowing accurate and effective simulations of molecules.

Determining the origin of engineered DNA can help to foster responsible innovation within the biotechnology community. A convolutional neural network approach that learns distances between engineered DNA sequences and various labs that could have created them is used to accurately predict the lab-of-origin.