deep learning

Rising water is an underappreciated driver of forest mortality

The paper tracks forest mortality across the U.S. Great Lakes and coastlines from 2012 to 2023, finding rising water levels as a major driver of tree death, especially in wetlands and low-lying forests.

ViTAMIn-O: Democratizing computer vision-based machine learning for stem cell research

ViTAMIn-O is an open DL model for organoid differentiation prediction, outperforming existing models and adaptable to various datasets and imaging modalities.

Folate deficiency disrupts key metabolic transitions within the developing neural ectoderm

This study uses a deep-learning model to analyze scRNA-seq data and identifies how folate deficiency disrupts metabolic transitions in developing mouse neural ectoderm, leading to morphological defects.

PathEdit: Diagnostic-Preserving Counterfactual Medical Image Editing through State-Factorized Generative Intervention

PATHEDIT edits medical images to change diagnoses while preserving other patient characteristics, achieving high target success and minimal off-target changes.

Space partitioning by self-organized epithelial networks

This study shows epithelial cells self-organize into reticulate networks, revealing a conserved mechanism for space partitioning through mechanical and topological dynamics.

DeepSCENIC: transfer learning from sequence-to-function models enables causal gene regulatory network inference

DeepSCENIC uses transfer learning from sequence-to-function models to infer causal gene regulatory networks from single-cell multiome data.

TCF25 emerges as a cytoplasmic regulator of GPRASP2 stability

TCF25 is mostly cytoplasmic with a perinuclear membrane-proximal pool, stabilises GPRASP2 against proteasomal degradation, and its C-terminal residues 604-664 are required for their association.

Multi-Objective Engineering of Fibrin-Selective Thrombolytic Proteases with Enhanced Biocatalytic Efficiency and Inhibition Resistance

This paper designs next-generation thrombolytic proteases with improved catalytic efficiency, fibrin selectivity, and inhibition resistance through multi-objective enzyme engineering.

Communication through autoattractants can enhance and limit collective migration of immune cells

This study shows autoattractants enhance immune cell migration but can also cause aggregation, with optimal communication balancing cell coordination and movement efficiency.

Programmed inhibition of an innate immune receptor via de novo designed transmembrane proteins

This paper designs synthetic proteins that inhibit TLR4 signaling by binding its transmembrane domain, demonstrating the importance of transmembrane interactions in immune signaling.

Key Regulatory Elements of the TGFbeta-LRRC15 Axis Predict Disease Progression and Immunotherapy Resistance Across Cancer Types

MINT infers the latent single-cell spatial transcriptome from paired histology

MINT infers cell-resolved spatial transcriptomes from paired histology and Slide-tags data, improving alignment and expression recovery in lung adenocarcinoma.

Exploring Healthy Neurocognitive Ageing with Deep Learning Interpretability Methods

This study uses deep learning interpretability methods on resting-state fMRI data to show how functional brain networks reorganize with age and relate to cognitive performance.

ANIMA: predicting protein-protein interactions across species

ANIMA predicts protein-protein interactions across species using an AI model trained on diverse animal species, achieving high accuracy in under-researched species.

RheoScale 2.0: Revealing the Hidden Roles of Protein Positions via Substitution Patterns

RheoScale 2.0 identifies protein positions with varied substitution outcomes, revealing roles beyond neutral or deleterious, improving interpretation of high-throughput mutagenesis data.

NeuralRNN: a unified framework for recurrent neural network methods in cognitive neuroscience

NeuralRNN unifies RNN methods for cognitive neuroscience tasks by providing a single framework for model implementation, comparison, and extension.

Counterfactual Explanations for Graph Neural Networks in Patient Outcome Prediction

This paper introduces a counterfactual explanation algorithm for graph neural networks to explain patient outcome predictions in patient similarity networks.

Integrative Structure of Norovirus NS3 Suggests a Role in RNA Transport

The paper determines the structure of norovirus NS3 and proposes it functions as an RNA translocase.

Sparse Linear Algebra Accelerates Genotype Representation Graph Computation at Biobank Scale

This paper introduces Mikado, a method using sparse linear algebra to accelerate genotype representation graph computations, significantly improving speed over existing tools on large biobank datasets.

StressNET: an adaptable deep-learning model for mechanical stress inference in tissues

StressNET infers intercellular mechanical stresses from tissue geometry using a Graph Neural Network, achieving state-of-the-art performance in zebrafish and Xenopus embryos.

Code-Multiplexed Multi-Frequency Impedance Cytometry with a Unified Deep-Unfolding Network

This paper develops a deep-unfolding network for code-multiplexed, multi-frequency impedance cytometry to resolve signal overlaps and enable label-free single-cell analysis.

Rapid volumetric reconstruction and tracking for Fourier light-field microscopy enables real-time calcium imaging in freely behaving Hydra.

This paper introduces OpenFLR, a software framework for real-time volumetric reconstruction and calcium imaging using Fourier light-field microscopy in Hydra.

Enhanced Tactile Coding in Rat Neocortex Under Darkness

This study shows that tactile neural representations in rat somatosensory cortex become more distinct under darkness, revealing reorganization of sensory processing when visual input is removed.

Human body-site-stratified fungal genomic catalogs enhancing human mycobiome analysis

The paper introduces MycobiomeDB, a fungal genome catalog, and site-specific MycoProfilers that enhance mycobiome analysis by improving taxonomic recall and distinguishing disease-associated mycobiomes from healthy controls.

Temporal Hydrogen-Bond Network Analysis Reveals Substrate-Directed Connectivity in Dihydrofolate Reductase

The paper introduces a temporal hydrogen-bond network analysis framework to study ligand-directed connectivity in DHFR, revealing mutation-sensitive pathways and evolutionarily supported communication regions.

The pod components of the Shigella T3SS sorting platform accommodate multiple copies of Spa33 (SctQ)

This paper models the Shigella T3SS pod complex using AlphaFold, showing two Spa33 proteins bind MxiK and SPOA2-SPOA2 domains, supporting a 1:4:2 stoichiometry and explaining electron density data.

Inhibiting the interaction between the mitochondrial receptor Tom70 and SARS CoV 2 Orf9b with small molecules

This paper identifies two strategies to modulate the Orf9b-Tom70 interaction by screening for small molecules that either stabilize Orf9b dimer or block its binding to Tom70, with potential to restore interferon signaling against SARS-CoV-2.

AlphaGenome Atlas: in silico mutagenesis of the entire human genome improves prioritization and interpretation of non-coding variants

AlphaGenome Atlas predicts regulatory effects of non-coding variants across the human genome, improving their prioritization and interpretation.

CUBE: Multimodal Representation Learning Reveals Biological Structure Across Histomorphology, Spatial Protein Phenotypes, and Transcriptome-Associated Signals

CUBE integrates spatial protein and transcriptomic data via histology, creating a multimodal representation that captures biological structure from incompletely paired data.

Segment Any Plant (SAP): Foundation-Model Segmentation for Plant Time-Series Phenotyping

SAP provides a transferable, training-free framework for automated segmentation of plant time-series imagery across species and conditions.