variational autoencoder

SPATNIC enables H&E-grounded cancer-cell identification across spatial and single-cell transcriptomic modalities in colorectal cancer

SPATNIC is a variational autoencoder that identifies cancer cells from spatial and single-cell transcriptomic data in colorectal cancer, outperforming existing tools and identifying a rare cancer stem cell population.

A Generative AI Framework to Predict Cardiomyocyte Contraction Function from Single Static Images.

This paper introduces a generative AI framework that predicts cardiomyocyte contraction from static images, achieving high accuracy and revealing structure-function relationships.

Extracting interpretable single-cell metabolic states with graph-guided representation learning

The paper introduces MeRN, a graph-guided variational autoencoder for learning single-cell metabolic states from transcriptomes, and identifies metabolic drivers in various biological contexts.

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.

An Integrated Single-Nucleus Atlas Resolves Cell-Type-Specific Programs and Molecular Subtypes in Alzheimer's Disease

The paper constructs a single-nucleus transcriptomic atlas of Alzheimer's disease and identifies cell-type-specific molecular subtypes and therapeutic targets using a new framework.

Invasion status stratifies the composition of the human pancreatic cancer perineural niche

Xenium and single-nucleus profiling of 14 human pancreatic tumors shows invaded nerve neighborhoods enriched for classical cancer cells, myCAFs and lipid-handling macrophages but depleted of lymphocytes.

Deciphering Cell Cycle Dynamics and Cell States in Single-cell RNA-seq data with SPAE

This paper introduces SPAE, a model for analyzing cell cycle dynamics and states in scRNA-seq data, showing improved accuracy and robustness in cell cycle characterization and cancer cell cycle transition prediction.

Learning proteomic disease trajectories with flow matching

This paper introduces proteome velocity, a framework using flow matching to infer continuous proteome trajectories from proteomics data, showing its application in mouse sepsis and clinical COVID-19.

Dynamics of ML-based Morphological Features Indicate a Shear Stress-Dependent Bifurcation of hiPSC-Derived Endothelial Cell States

The paper uses ML to analyze morphological dynamics of hiPSC-derived endothelial cells under shear stress, identifying a bifurcation between two stable cell states.

A variational sparse Gaussian-process method for detecting spatially variable genes and cellular interactions from spatial transcriptomics

VISGP identifies spatially variable genes and cellular interactions from spatial transcriptomics data, outperforming existing methods in detecting ligand-receptor pairs.

CoLa-VAE: A Cell-Cell Communication-Aware Variational Autoencoder for Representation Learning and Expression Denoising

CoLa-VAE improves denoising and representation learning in single-cell data by integrating cell-cell communication information, enhancing downstream analyses like differential communication detection and spatial deconvolution.

Variational autoencoder for interpretable seizure onset phases detection

This paper introduces a variational autoencoder for detecting seizure onset phases in SEEG data, achieving high recall and interpretable latent features.

Granger Sensori-Behavioral Taxonomy of Neuronal Ensemble Activity from Two-Photon Calcium Imaging Data

This paper introduces a framework to extract directional functional interactions between neurons, stimuli, and behavior from two-photon calcium imaging data, revealing distinct sensori-behavioral roles in the mouse auditory cortex.

Translating Brain Anatomy and Disease from Mouse to Human in Latent Gene Expression Space

This paper develops a VAE-based framework to translate mouse brain gene expression to human, showing that mouse disease models predict human Alzheimer's and Parkinson's patterns and identify the best mouse model for AD patients.

Geometry aware graph attention networks to explain single-cell chromatin state and gene expression

SEAGALL uses geometry-aware graph attention networks to extract biologically relevant features from single-cell data, revealing cell identity drivers beyond marker genes.

Canalized gene regulatory networks stabilize floral polymorphism and enable modular transgressive expression

The paper investigates how canalized gene regulatory networks maintain floral polymorphism in Stellera chamaejasme by analyzing transcriptomic and regulatory variation across morphs.

CoPhaser: generic modeling of biological cycles in scRNA-seq with context-dependent periodic manifolds

CoPhaser models biological cycles in scRNA-seq by decomposing data into periodic and non-periodic sources, revealing insights into cell cycle dynamics and other periodic processes across diverse contexts.

A hierarchical generative model reveals enhanced latent precision of brain-body interaction dynamics during interoceptive attention

The paper uses a hierarchical generative model to analyze brain-body interaction dynamics during interoceptive attention, finding increased latent precision in multimodal physiological signals.

Transcriptional Hysteresis and Irreversibility in Periodontitis Revealed by Single-Cell Latent Manifold Modeling

This study uses single-cell RNA sequencing to reveal transcriptional irreversibility in periodontitis, introducing the Regenerative Permission Index to predict treatment failure.

A Unified Dynamical-Systems and Control-Theoretic Model for Single-Cell Fate Dynamics

This paper introduces a unified dynamical-systems and control-theoretic model to predict single-cell fate dynamics by integrating pseudotime, RNA velocity, optimal transport, and stochastic trajectories under partial observability.

Mapping trans-eQTLs at single-cell resolution using Latent Interaction Variational Inference.

This paper introduces LIVI, a deep learning framework for mapping trans-eQTLs at single-cell resolution, and shows it detects more trans-eQTLs than existing methods and identifies cell types and genes involved in autoimmune disease risk.

LoopBin, a VaDE-based neural network for chromatin loop classification

LoopBin classifies chromatin loops using a VaDE neural network, identifying distinct functional subgroups from 3D genomics data with minimal input.

Machine Learning-Based Tumor Segmentation and Classification Using Dynamic Optical Contrast Imaging (DOCI) for Thyroid Cancer

This paper uses machine learning with DOCI imaging to improve tumor segmentation and classification in thyroid cancer, achieving high accuracy in tissue type distinction.

Multi-view deep learning of highly multiplexed imaging data improves association of cell states with clinical outcomes

This study uses multi-modal VAEs to integrate multiple views of highly multiplexed imaging data, improving the prediction of clinical outcomes and identification of cell states.

MisTIC: Missegmented Transcript Inference Correction for Improved Spatial Transcriptomics Analysis

MisTIC corrects transcript misassignment in spatial transcriptomics data, improving cell type identification, differential expression analysis, and RNA localization.

Learning Invariant Graph Representations for Cox Survival Modeling under Distribution Shifts

This paper introduces InvGraphCox, a framework that improves survival prediction by learning invariant graph representations under distribution shifts, outperforming baselines in discrimination and risk stratification.

A multi-modal cell-free RNA language model for liquid biopsy applications

Exai-1 is a transformer-based model that integrates RNA sequence and expression data to improve cfRNA analysis for disease detection.

Multi-modal, multi-species, and multi-task latent-space model for decoding level of consciousness

This paper introduces a multi-modal, multi-species, and multi-task latent-space model to decode consciousness levels from neural recordings, achieving cross-species and zero-shot transfer.

Unified Probabilistic Analysis of CyTOF: A Deep Generative Approach using CytoOne

CytoOne is a probabilistic framework for CyTOF data that integrates batch correction, differential analysis, and visualization, effectively modeling sparse and noisy data.

An integrated platform for high-throughput phenospace learning of 3D multilineage organoid systems

The paper introduces a platform for high-throughput spatial phenotyping of multilineage organoids, identifying pathway modulators and treatment strategies for PDAC.