Honda Naoki

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.

The paper infers spatially resolved tau amplification rates from tau PET data using a differentiable reaction-diffusion framework and links these rates to amyloid PET and gene expression patterns.

The paper introduces mindGAP, a framework to quantify individual-population gaps in depressive symptom dynamics using longitudinal data, finding divergence in symptom organization associated with depression severity and traits.

InferAging quantifies biological aging by modeling deviations from chronological age, identifying accelerated and delayed agers in zebrafish using non-invasive phenotypes.

This study uses a computational model and machine learning to analyze monkey decision-making in risk-taking tasks, revealing dynamic optimism and pessimism states and a unique control rule for bias.

This paper introduces a machine-learning framework to estimate latent world-model dynamics from behavioral data and applies it to analyze insight-like shifts in mouse behavior.

Monju performs multi-criteria clustering on single-cell omics data by integrating multiple clustering solutions to assess cluster stability.

This paper introduces a zero-shot framework to predict spatial transcriptomes of mutants using single-cell RNA sequencing data without requiring spatial reference atlases.

The paper develops a filament-resolved model of actin self-organization, showing how Arp2/3 and fascin concentrations determine lamellipodia, filopodia, and reticulated meshwork morphologies.

This paper develops a continuous reinforcement learning framework to model anticipatory homeostatic regulation and demonstrates its application to thermoregulation and multi-variable physiological control.

The paper develops a machine-learning framework to infer governing equations of mechano-chemical dynamics in epithelial migration from live-cell imaging data, revealing cell-cell heterogeneity in migratory rules.

This study uses a Bayesian hierarchical model to decode decision-making properties from trust game data, identifying biomarkers linked to trusting behavior in MDD patients and healthy controls.

SPERRFY applies Sperry's chemoaffinity theory to whole-brain connectivity using spatial transcriptomics and connectomic data, revealing positional gradients that predict axonal wiring with high accuracy.

The paper introduces Foldinsight, a structure-based framework for predicting protein function using structural information, achieving performance comparable to sequence-based models.

This paper introduces Kanade, a variational autoencoder-based method that disentangles batch effects from biological signals in single-cell transcriptomics, improving comparative analysis across conditions.

The paper introduces a machine learning framework, ISI, to infer how organisms prioritize environmental signals from time-series data, applied to medaka fish gonadal development under natural conditions, revealing dynamic signal importance and gene expression correlations.

The paper develops a machine learning method to recognize individual F-actin filaments in HS-AFM images, revealing their orientation patterns in lamellipodia and cell cortex.

The paper presents a single-cell transcriptome atlas of Drosophila gastrulae, identifying 65 clusters and revealing spatial gene expression patterns across germ layers.

This study develops a decision-making model and machine learning method to decode curiosity from behavioral data, revealing that irrational choices in rats correlate with negative curiosity levels.

This study develops a mathematical model that combines hierarchy and neutral competition to explain stem cell homeostasis, showing unique clonal expansion patterns and clone size distributions.

This paper develops a hierarchical Bayesian model to predict amyloid-beta accumulation from biomarker candidates using mouse behavioral data, showing improved prediction when multiple features are used.

Perler predicts spatial gene expression from scRNA-seq data by integrating with in situ hybridization data using a generative linear model, achieving accurate and interpretable results across multiple tissues.

This study develops a computational model of ERK-mediated somitogenesis to show how the segmentation clock maintains noise-resistant developmental reproducibility in vertebrate somite formation.

This paper develops an inverse reinforcement learning framework to identify animal behavioral strategies from time-series data, applied to C. elegans thermotaxis, revealing distinct strategies for fed and starved worms.

This paper revisits chemoaffinity theory by developing a mathematical model of growth cone chemotaxis to explain how axons navigate to their correct targets in topographic maps.

The paper develops a biophysical model to explain how microtubule orientations differ between axons and dendrites based on neurite growth speed and hydrolysis rate.