deep learning

Long-read shotgun metagenomics identifies distinct urogenital microbial community states associated with functional potential and urinary tract infection during pregnancy

This study uses long-read shotgun metagenomics to identify distinct urogenital microbial community states in pregnant women, linking them to functional potential and UTI risk.

Earl Grey ParTEA: scalable, consistent transposable element annotation from pangenomes to deep phylogenetic scopes

Earl Grey ParTEA annotates transposable elements across multiple genomes, revealing shared and private TE families and improving annotation consistency.

FeatureMAP enables network-guided interpretation of untargeted metabolomics

FeatureMAP improves interpretation of untargeted metabolomics data by integrating annotation, network refinement, and functional analysis, revealing metabolic patterns in Alzheimer's disease.

PartiDesign: A structure-sequence co-design paradigm for practical all-to-all biomolecules design

PartiDesign is a structure-sequence co-design model that generates molecular structures and sequences for proteins, nucleic acids, and small molecules, achieving high design throughput across benchmarks.

FoldAbS: Repurposing the Protein Folding Model as a Foundation Encoder for Antibody Screening

FoldAbS repurposes protein folding models as foundation encoders to improve antibody screening by evaluating specificity, affinity, and developability in a unified framework.

From smartphone to gene: Automated image-based phenotyping captures flowering diversity of common buckwheat

The paper uses smartphone imaging and automated analysis to phenotype flowering traits in buckwheat, identifying heritable traits and genomic regions associated with flowering patterns.

Vision-language representations link food perception to subjective valuation

The paper shows that a vision-language model better predicts food valuation than vision-only models by encoding higher-level food attributes.

Decoding Cell-type Specific Gene Regulation with CellReg3D by Integrating DNA Large Language Models and 3D Genome Structure

CellReg3D integrates DNA language models and 3D genome structure to model cell-type-specific gene regulation, achieving state-of-the-art performance in predicting epigenetic marks and gene expression.

Inferring modularity in genotype-phenotype maps using patterns of epistasis and pleiotropy

This paper introduces two methods, SLED and GTD, to infer modular structure in genotype-phenotype maps by combining epistasis and pleiotropy, showing they yield similar modular architectures in yeast data.

Spectral Adaptive Repositioning for Flow-Based Single-Cell Perturbation Modeling

SPEAR predicts cellular responses to perturbations by adaptively repositioning genes in a co-expression network using spectral encoding and generative modeling.

SafeLens: A Statistically Controlled Nonlinear Machine Learning Framework for Characterizing Hidden Dynamics in Event-Locked Neural Activity

SafeLens uses nonlinear machine learning with statistical control to detect subtle, trial-varying neural dynamics in event-locked activity, outperforming traditional methods in sensitivity and specificity.

Reference-Guided De Novo Retrieval of 3D Chromatin Organizational Features with Representation Learning

The paper introduces C-Fetch, a framework that retrieves 3D chromatin features using reference examples, showing improved enrichment for biologically relevant contacts compared to existing methods.

Unified cell segmentation for spatial transcriptomics via transcript reassignment

Tracel improves cell segmentation in spatial transcriptomics by assigning transcripts to cells using optimal transport, enhancing cell-type separation and reducing contamination.

MicrobeRT: A Framework and Suite of Fine-tuned Language Models for Genomic Sequences

MicrobeRT is a framework of fine-tuned language models for metagenomic analysis, achieving strong performance in taxonomic classification, antibiotic resistance detection, and pathogenicity assessment.

COSMOS: integrated vector-pathogen metagenomics for mosquito-borne disease

COSMOS integrates vector-pathogen metagenomics to monitor mosquito-borne diseases, detecting pathogens and mosquito communities in field samples with high accuracy.

Massively parallel quantification and prediction of monomeric protein yield in Escherichia coli

The paper develops a scalable assay and a protein language model to predict monomeric yield of proteins in E. coli, showing it can be learned and designed.

Interpretable Machine Learning Links Group-Specific Plasma Proteomics to Regional Structural Neuroimaging in Alzheimer's Disease

The paper identifies group-specific plasma proteomic signatures in Alzheimer's disease stages and links them to regional brain structural changes using machine learning and neuroimaging.

AlphaFold 3 Prediction Accuracy and Reliability across RNA Structures and Protein-RNA Interfaces

This study evaluates AlphaFold 3's accuracy and reliability in predicting RNA structures and protein-RNA interfaces, finding that confidence scores do not always correlate with correct predictions.

Predicting single-molecule chromatin distribution from DNA sequence

This paper introduces a sequence-to-distribution framework to predict single-molecule chromatin states from DNA sequence, revealing new regulatory mechanisms and variant effects.

Immersion as inference: An active inference account of induced realities and its relation to the two-competitor model

This paper develops an active inference framework to explain immersion in virtual and augmented realities by modeling induced realities as competing generative models.

scPCIP: cross-species annotation of plant cell types with conserved gene programs

scPCIP annotates plant cell types across species using conserved gene programs, outperforming marker-based and deep-learning methods in held-out datasets.

brisc: a multithreaded CPU toolkit for single-cell analysis

brisc is a multithreaded CPU toolkit for single-cell analysis that is faster and more reproducible than existing tools like Scanpy and Seurat.

Cross-cancer knowledge transfer for lymph node metastasis prediction: Applicability of an axillary lymph node-based deep learning model

This paper evaluates a deep learning model trained on axillary lymph nodes for predicting metastasis in lymph nodes from different anatomical regions across cancers.

Computational Analysis of Antioxidant Effects on Stress Response Mechanisms in Candida albicans

The paper uses computational methods to analyze antioxidant effects on stress response mechanisms in Candida albicans, identifying naringenin and quercetin as potential modulators of Sir2/SIRT3 pathways.

SQANTI-evidence: isoform-resolved, evidence-tiered structural annotation of unannotated genomes from long-read RNA-seq

SQANTI-evidence annotates genomes from long-read RNA-seq without a reference, achieving high precision in gene and isoform prediction across species.

Auditing Biomedical Literature with Blinded AI Peer Review: Where Peer Review Works, Where It Fails, and How to Strengthen It

This paper evaluates the effectiveness of a blinded AI peer review system (BARS) in auditing biomedical literature, finding it improves consistency and detects flaws while highlighting issues with megajournals and template studies.

Finely tuned DDA-PASEF-based immunopeptidomics workflow enhances HLA class I coverage

This study optimizes a DDA-PASEF-based immunopeptidomics workflow to enhance HLA class I peptide coverage in low-input samples.

SeverYX: An ensemble framework for robust variant severity classification

SeverYX is an ensemble framework that classifies genetic variants into severity tiers, achieving high concordance with ClinVar and improving predictive accuracy through threshold optimization and additional predictors.

Automated Segmentation of the Hypothalamus: Associations of Age and Socioeconomic Status with Subunit Volumes in Cognitively Healthy Adults

Marker-specific pixel attribution for single-cell protein quantification with PixAQ

PixAQ improves single-cell protein quantification in imaging mass cytometry by assigning pixel signals to specific markers using spatial and intensity models.