deep learning

Integrating morphology and gene expression of neural cells in unpaired single-cell data using GeoAdvAE

GeoAdvAE integrates unpaired single-cell morphology and transcriptomics to align cell types and reveal disease-associated microglia signatures.

Acoustic Monitoring of Tropical Bats: Saccopteryx bilineata's response to habitat and time across a restoration gradient using a 2D-CNN

The paper develops a 2D-CNN to detect bat calls in acoustic data across a restoration gradient, finding that restoration plots have higher detection rates than forest habitats.

Inferring cascade drivers of VEXAS syndrome by a causal machine learning tool CauNagi

The paper introduces CauNagi, a causal machine learning tool, to identify causal drivers of VEXAS syndrome using single-cell transcriptomic data, revealing key regulatory hubs involved in myeloid dysfunction.

Genome-resolved ecology and evolution of the microbiome enabled by quantitative long-read metagenomics

This paper introduces LRQ, a long-read metagenomics method that enables absolute strain abundance measurement and resolves microbial evolution in a Crohn's disease patient over eight years.

Artificial Intelligence-Driven Nanosensing to Identify Circulating Bacterial DNA: A Novel Approach to Breast Cancer Risk Profiling

This paper develops a PQD-based nanosensor using LSTM deep learning to detect bacterial species linked to breast cancer, showing high sensitivity and selectivity for early detection.

Probing the sequence variability tolerance in a de novo α-helical barrel biocatalyst

This study modifies de novo α-helical barrel biocatalysts using deep learning and computational methods to improve solubility and catalytic performance while maintaining structural stability.

Drought-Spec-Net: Early Tomato Drought Detection and Potential Yield-Impact Assessment Using Vis NIR Data

This paper develops Drought-Spec-Net, a model for early tomato drought detection using Vis NIR data, achieving high accuracy and providing yield impact assessments.

Anatomically Guided Deep Learning Reconstruction of Accelerated Snapshot CEST MRI

This study shows that using subject-specific structural MRI data improves the reconstruction of highly accelerated CEST MRI, especially at higher acceleration factors.

De novo minibinders targeting RH5 achieve nanomolar inhibition of blood-stage malaria replication

This paper designs de novo minibinders that inhibit blood-stage malaria replication with nanomolar potency by targeting the RH5-basigin interaction.

ScProteoAgent enables natural-language-driven single-cell proteomics analysis and interpretation

ScProteoAgent translates natural-language requests into single-cell proteomics analyses, preserving analytical state and enabling follow-up queries with high benchmark performance.

A directed TM->JM coupling in receptor tyrosine kinase dimers, set by activating mutations and the membrane environment

The paper uses MD simulations and computational methods to determine the direction of conformational coupling in RTK dimers, showing it correlates with receptor activity states.

Deep Representation Learning on Whole-Brain Population Dynamics Uncovers Geometrically Separable Neural Codes

The paper develops a deep-learning framework to learn interpretable representations of whole-brain neural dynamics in Drosophila without neuronal identification, revealing factorized organization along metabolic state, sensory modality, and stimulus valence axes.

Evaluation of LiDAR-based Canopy Trait Estimation in Midwestern Row Crops

This study evaluates LiDAR's ability to estimate canopy traits in corn and soybean, finding comparable LAI quality in corn but poor performance in soybean due to dense canopies, and accurate stand count in corn with faster data collection.

Real-time AI integration for MR to detect artifacts and guide pulse sequence adaptations

This paper introduces an AI-integrated MRI pulse sequence that detects and adapts to out-of-voxel artifacts in real-time, improving data quality through gradient adjustments.

Biomarker-Aware Super-Resolution for 4D Flow MRI in a Synthetic Pulmonary-Artery Benchmark

This paper evaluates if biomarker-aware super-resolution improves hemodynamic endpoint fidelity in 4D flow MRI using a synthetic pulmonary-artery benchmark.

Unified modeling of mutation-induced protein stability responses with OmegaTherm

OmegaTherm predicts mutation-induced protein stability responses by jointly modeling folding free energy and melting temperature, achieving state-of-the-art performance in prediction and variant ranking.

Kingdom-wide Discovery of Type VI Secretion Effectors via AI Uncovers Three New Bacterial Toxins

The paper uses an XGBoost classifier to identify novel T6SS effectors in bacterial genomes, validating three new toxins with cytotoxic effects and structural features.

RNASeek: A Cross-Phyla Generative Foundation Model for Multipurpose RNA Modeling and Reinforcement Learning-Based Design

RNASeek is a generative foundation model for RNA sequence design and functional prediction, using reinforcement learning to optimize RNA properties like stability and cleavage activity.

Tri-Modality Representation Learning for Molecular Property Prediction

This paper introduces TMCA, a multimodal framework that integrates SMILES, molecular graphs, and 3D conformations for molecular property prediction, achieving the best performance on benchmark datasets.

Decoding natural scenes from patterned optogenetic responses in mouse visual cortex

This paper decodes natural scenes from optogenetic responses in mouse visual cortex and shows that patterned stimulation can reproduce visual representations with high fidelity.

Eukaryotic MAGs recovered from deep metagenomic sequencing of the seagrass, Zostera marina, include a novel chytrid in the order Lobulomycetales

This study uses deep metagenomic sequencing to recover MAGs from Zostera marina, identifying a novel chytrid in the Lobulomycetales order and other eukaryotic lineages with genomic novelty.

A Taxonomy-Informed Sparse DNA Foundation Model for Microbial Genomics

A sparse mixture-of-experts DNA model conditioned on hyperbolic taxonomy embeddings learns better microbial representations than its taxonomy-free twin and comes close to Evo2-7B in probing and fine-tuning at lower compute.

Predicting the immediate and subsequent effects of commercials on product valuation using EEG and deep learning

This study uses EEG and deep learning to predict consumer valuation and preferences after watching commercials, showing better accuracy than standard EEG measures.

Physics-Informed Neural Network for Mapping Vascular and Tissue Dynamics Using Laser Speckle Contrast Imaging

This paper develops a physics-informed neural network to rapidly estimate vascular and tissue dynamics from laser speckle contrast imaging data without ground-truth labels.

Inferring division-associated stochasticity from time-series single-cell transcriptomes

This paper introduces scDIVIDE, a framework to infer division rates and cellular dynamics from time-series scRNA-seq data, showing it accurately captures birth rates in biological processes.

A Large-Scale Deep Normative Modeling of Primary Sulcal Patterns Reveals Deviations in a Spectrum of Disorders

This paper introduces a deep learning model to detect sulcal pattern deviations in neurological disorders using a large cohort of typically developing individuals.

A Physics-Informed Neural Network Surrogate for Patient-Specific Hepatic Arterial Hemodynamics in Yttrium-90 Radioembolization: Network Architecture, Boundary-Condition Enforcement, and Data Efficiency

This paper develops a physics-informed neural network surrogate for predicting hepatic arterial flow in Y-90 radioembolization, showing that data fraction is more critical than architecture for accuracy.

Biologically grounded cell profiling across microscopy modalities

The paper introduces MorphAgent, a biologically grounded AI framework for cell profiling across microscopy modalities, improving feature design and performance in drug screening and disease analysis.

Discriminating betacoronavirus receptor usage across subgenera using protein structure prediction and molecular dynamics

The paper predicts ACE2 receptor binding potential of emerging betacoronaviruses using structural prediction and molecular dynamics, identifying Zhejiang2013 as a potential ACE2 binder.

Accessing Enzyme Kinetic Data and Prediction Methods at Scale

OpenKinetics Predictor unifies thirteen enzyme kinetic predictors behind one interface and API; on E. coli iML1515, eight kcat predictors correlate pairwise at r 0.08 to 0.67.