deep learning

Identification of gene kinetics patterns improves RNA velocity and cell fate prediction using JuloVelo

JuloVelo improves RNA velocity and cell fate prediction by incorporating gene kinetics patterns, outperforming existing methods across multiple differentiation programs.

Targeting PRMT5 Across Cancers and AI-Driven Discovery of Candidate Compounds

The paper identifies PRMT5 as a cancer target, uses multi-omics and machine learning to screen compounds, and identifies a potential inhibitor with high docking scores.

Programmable acoustic single cell manipulation with model-free machine learning

This paper introduces VeLO, a model-free machine learning algorithm for programmable two-dimensional single-cell manipulation using acoustic standing waves, achieving robust control across multiple cell types.

Xenograft supervision and joint methylation-fragment-end encoding define a reusable molecular scoring layer for cell-free tumor DNA

ctDNA-BERT scores individual bisulfite reads for cell-free DNA methylation analysis, achieving high accuracy in cancer staging without cohort-specific retraining.

Extracting massive ecological data on state and interactions of species using large language models

This study uses large language models to extract ecological interactions from 83,910 scientific papers, creating a network of 144,402 interactions between 36,471 taxa with high precision and sensitivity.

Interactome-based deep learning and path-level decomposition reveal TGFβ-IFNα competition in MASH

The paper uses a deep learning framework to analyze drug-induced transcriptomics and identifies a four-node module involved in TGFβ-IFNα competition in MASH.

RNA-SpecFlow: RNA-Conditioned Spectral Flow for Histology Synthesis

RNA-SpecFlow synthesizes histology images conditioned on RNA data, showing that increased detection counts can coincide with reduced segmentation agreement.

In silico prediction of metabolic trait robustness in microbial cells

This paper introduces TRAEV, a computational framework to predict the evolutionary robustness of metabolic traits in microbial cells, validated using yeast strains in controlled environments.

Using SpliceAI to triage splice-altering variants in 7,220 individuals with rare conditions highlights limitations of the precomputed scores

The paper evaluates precomputed SpliceAI scores for splice-altering variant prioritization in NDD cohorts, finding inaccuracies and providing a tool to improve variant triage.

Recovery of a minor cryo-EM particle population reveals conformational equilibria linking cofactor loading, turnover, and reactivation in methionine synthase

The paper reanalyzes a cryo-EM dataset using advanced methods to uncover a minor reactivation-state structure of methionine synthase, revealing conformational changes during cofactor turnover.

VenusGenome: Genome-Aware Masked Autoencoding for Microbial Genome Representation Learning

VenusGenome improves microbial genome representation learning by using a genome-aware masked autoencoder, achieving high performance in phenotype prediction and gene-level tasks.

golgi: an open-source graphical platform for image-to-recruitment modeling of peripheral nerve stimulation

golgi provides an open-source platform for modeling peripheral nerve stimulation, enabling image-to-recruitment analysis with a graphical interface and reproducible results.

Building virtual cell dynamics that evolve with biological measurements

CellCompass, an agentic AI system, models cell dynamics from single-cell measurements and improves predictions by integrating new biological constraints and experimental data.

AltereGO: Prediction of Protein Function by Multimodal Fusion of Protein Language Models.

AltereGO fuses protein language models and structure representations to predict molecular functions, achieving better performance than individual models and some existing methods.

BLOOM: Biologically Informed Flow Matching for Transcriptomic Virtual Embryo Modeling

BLOOM generates developmental cell-expression distributions across time using conditional flow matching, outperforming baselines in gene and cell-expression correlations.

HERCULES: a deep-learning framework for RNA-binding region localization and mutation-effect prediction

HERCULES predicts RNA-binding regions and mutation effects using a deep-learning framework combining language models and physicochemical descriptors.

MPFIL: Multi-Level Protein Feature Integration for Predicting Mutation-Induced Protein-Protein Binding Affinity Changes

MPFIL predicts mutation-induced changes in protein-protein binding affinity by integrating structural and sequential features across multiple levels.

Mimic: A Unified Neural Network Architecture for Modelling Genomic Sequences and Module-level Comparison

This paper introduces Mimic, a unified neural network architecture for genomic sequence modeling, showing it outperforms existing models in protein-DNA interactions and enhancer activity.

Computational Design and Evaluation of De Novo Protein Minibinders Targeting Specific Sites of STAT3

The paper designs and evaluates de novo protein minibinders targeting specific STAT3 sites using AI and AlphaFold3, identifying high-confidence binders with novel sequences.

Increasing Cancer Stemness Drives Prostate Cancer Progression, Plasticity, Therapy Resistance and Poor Patient Survival

This study quantifies cancer stemness in prostate cancer across progression stages, showing its association with therapy resistance, poor survival, and lineage plasticity.

NK Cells Engineered with a Chimeric Antigen Receptor Delay HIV Rebound and Reshape HIV Reservoir Composition

This study shows that engineered NK cells with a CD4-based CAR delay HIV rebound and alter the reservoir's composition in humanized mice.

Scaling a context-aware genomic model for condition-specific edit design in plants

CROP-1 predicts gene activity based on DNA sequence and regulatory state, enabling context-specific edit design in plants with higher accuracy than sequence-only models.

Accurate prediction of protein stability changes from single mutations using self-distillation and antisymmetric constraint strategies

DPStab predicts protein stability changes from single mutations using self-distillation and antisymmetric constraints, achieving state-of-the-art performance in ΔΔG and ΔTm prediction.

An evidence-governed agent harness for spatial transcriptomics prioritizes ICAM-1 as a target in lung fibrosis

DeepMAST, an AI system for spatial transcriptomics, identifies ICAM-1 as a therapeutic target in lung fibrosis through evidence-based analysis and validation.

An AI-based and coding-free integration for forest Leaf Area Index calculation

This paper introduces a coding-free AI method for improving leaf area index (LAI) calculation from hemispherical images, showing higher consistency and reliability compared to existing tools.

Joint reconstruction of multi-contrast relaxation-diffusion imaging

This paper introduces DIFFUSE, a framework for joint k-q-TE image reconstruction in diffusion MRI, showing improved error reduction and multi-contrast reconstruction.

Sequence design for three-dimensional genome folding using Akita Semifreddo

This paper introduces Akita Semifreddo, a framework for designing DNA sequences that produce specific 3D genome folding patterns, demonstrating the ability to create boundaries, dots, and flames with controlled strengths.

From pose to behavior: SABER integrates identity-resolved multi-animal pose tracking with large-language-model-based behavioral factor discovery

SABER improves pose tracking and behavioral analysis in multi-animal videos by integrating identity-preserving tracking with behavioral factor discovery, showing reduced approach behavior in depression-susceptible mice.

Dissociating Resting-State Networks of Pedohebephilic Preference and Child Sexual Offending Using Network-Based Prediction

The paper distinguishes pedohebephilic preference from child sexual offending using resting-state functional connectivity and network-based prediction, identifying distinct neural network alterations.

Accelerated discovery and optimization of a potent andselective MMP12 inhibitor with the machine learningplatform CAMOLE-1

This paper describes the development of CAMOLE-1, a machine learning platform for designing potent and selective MMP12 peptide inhibitors, achieving results in about one year with 270 synthesized peptides.