deep learning

flashscenic: Accelerated GRN Analysis with RegDiffusion for Large-Scale Single-Cell Data

flashscenic accelerates single-cell GRN analysis using GPU acceleration, improving speed and enabling benchmarking of integration strategies in AML data.

Interpretable Deep Learning of Endogenous Fluorophore Concentration and Entropy for Lung Tissue Classification

The paper uses an interpretable deep learning model to classify lung tissue as normal or cancerous by combining fluorophore concentration and entropy features, achieving high accuracy.

General prediction of T cell receptor antigen specificity from sequence using AlphaFold 3

This paper uses AlphaFold 3 to predict T cell receptor antigen specificity from sequence, achieving high accuracy on unseen epitopes.

Few-shot regulatory sequence design with a genomic language model

This paper demonstrates that a genomic language model can generate functional regulatory DNA sequences using few examples, showing improved performance in cell-type-specific gene expression and neuronal labeling.

GBactPro: cross-species bacterial promoter prediction using element-resolved promoter definitions

GBactPro improves cross-species bacterial promoter prediction using element-resolved promoter definitions, achieving high AUROC across multiple species and outperforming existing predictors.

Scooby-2: fine-tuning AlphaGenome for single-cell genomic track prediction

Scooby-2 fine-tunes AlphaGenome to predict single-cell genomic tracks, improving correlation and variant effect prediction compared to Scooby and AlphaGenome.

Matrix biophysical cues drive transient or prolonged breast cancer dormancy programmes

This study shows that matrix stiffness influences breast cancer cell dormancy states, with soft matrices inducing transient dormancy and stiff matrices causing prolonged dormancy.

Integrative taxonomy using traits and genomic data for Species Delimitation with Deep learning

This paper introduces a deep learning method that integrates genomic and trait data to improve species delimitation in complex speciation scenarios.

ESMRank reveals a transferable axis of protein mutational constraint from overlapping variant effect assays

ESMRank is a sequence-based predictor that integrates protein language models and physicochemical features to rank mutational constraint across proteins, outperforming existing methods and capturing disease-relevant variants.

Multimodal epi-mode 3D quantitative phase and fluorescence microscopy enables supervised volumetric virtual fluorescence in bulk tissue

This paper introduces a multimodal microscope that uses a cGAN to generate virtual fluorescence images from phase contrast data, enabling virtual histological staining of bulk tissues.

Agentic Discovery Transfer Enables Cross-Cancer Discovery at Scale

SMART, a multi-agent framework, transfers findings from single-cell RNA-sequencing cancer studies to other cancer types, identifying cross-cancer prognostic signatures and generating new hypotheses.

Illuminating phage dark matter beyond the reach of structural alignments with PhageFlare

PhageFlare predicts functions for phage proteins with no structural matches, outperforming baselines and annotating many previously unresolved proteins.

Data-driven gait cycle decomposition based on whole-body coordination dynamics

This study decomposes gait cycles using a dynamic system framework to identify coordination patterns in human locomotion.

Multi-resolution humanness scoring of antibodies via a region-isolated language model

The paper introduces IgScope, a language model for antibody humanness scoring at multiple resolutions, showing improved accuracy over existing methods and revealing issues with generative models.

Preserving Information and Integrity in Cryo-EM: A Fourier-Space Deposition Approach

The paper identifies limitations in current cryo-EM data deposition practices and proposes an alternative approach to better preserve information and support validation.

Intelligent SEEQPI enables reconstruction of inhibitory neuronal nuclei and dry-mass mapping deep in the living mouse brain

This paper introduces SEEQPI, a label-free imaging technique, to reconstruct inhibitory neuronal nuclei and estimate dry-mass density in deep brain regions of living mice.

XChrom: a cross-cell chromatin accessibility prediction model integrating genomic sequence and cellular context

XChrom predicts chromatin accessibility across cells and genomic regions using genomic sequence and cell identity, outperforming existing methods and enabling cross-species analysis.

Impact of Rare and Common Genetic Variation on Cell Type-Specific Gene Expression

Single-cell eQTL mapping in 5.4 million PBMCs from 1,925 individuals with matched WGS finds common-variant eQTLs for 83% and rare-variant signals for 47% of 21,404 genes.

Integrating sequence-to-omics predictions and personal multi-omics for rare variant prioritization

The paper introduces Watershed-DL, a framework integrating sequence-to-omics predictions and multi-omics data to prioritize rare variants, finding it increases prioritized variants by 61.4% and identifies candidate disease alleles.

Neopep: Reinforcement Learning From Machine-Learned Force Field Energy for Non-Canonical Peptide Structure Prediction

Neopep uses reinforcement learning with machine-learned force field energy to predict structures of non-canonical amino acid-containing peptides, achieving lower RMSD than existing methods.

DINOSim: Zero-Shot Object Detection and Semantic Segmentation on Microscopy Images

DINOSim detects and segments objects in microscopy images without annotated data, achieving performance comparable to supervised methods using a pretrained DINOv2 encoder and k-nearest neighbors refinement.

Who contributes to whom: following the evidence behind protein language-model predictions of backbone variability

ESMfluc predicts local backbone variability in proteins using attention-based evidence decomposition, showing how sequence positions contribute to predictions of single or multi-PB states.

Antimicrobial peptide databases and prediction tools: Toward a standard evaluation framework

The paper evaluates antimicrobial peptide prediction tools and proposes a standardized evaluation framework to improve comparability and reproducibility.

Predicting conformational changes induced by biomolecular interactions

This paper introduces K-Fold, a method that improves prediction of conformational changes in protein complexes by modeling structural transitions from unbound to bound states.

Benchmarking and Experimental Validation of Machine Learning Strategies for Enzyme Engineering

This study benchmarks machine learning models for enzyme engineering using a curated dataset, finding zero-shot predictors are most generalizable but still modestly correlated with activity, with consensus models improving precision.

Widespread Presence of Catalytic RNAs in All Domains of Life

The paper identifies millions of self-cleaving ribozymes in humans and other species, revealing a much larger ribozyme-ome than previously thought and their potential roles in gene regulation and biotechnology.

Cross-family learning transfers thermal adaptation between enzymes

This paper shows that thermal adaptation can be transferred between unrelated enzymes using a generative framework, improving thermostability without evolutionary homology.

Simulating protein evolution with programmable fitness objectives derived from modern protein models

This paper simulates protein evolution using programmable fitness objectives derived from modern protein models, showing that combined objectives improve evolutionary realism and molecular plausibility.

GlioVision: A Multi-Modal MRI Framework for Non-Invasive Glioma Molecular Biomarkers Prediction

GlioVision predicts four glioma molecular biomarkers from MRI data using a multi-modal framework, achieving high AUC scores across multiple cohorts.

High-throughput mapping of kinase-substrate interactions with protein language models and phospho-PCA

This paper develops a framework combining a protein language model and a screening assay to predict kinase-substrate interactions and map the human kinome.