De Novo Design of SIK3 Inhibitors via Feedback-Driven Fine-Tuning of Seq2Seq-VAE [new] De novo design... uses active learning to refine a Seq2Seq-VAE, generating potential SIK3 inhibitors without extensive data.

Entangled Schrödinger Bridge Matching][new] Models interacting particle dynamics by entangling velocities via coupled bias forces, improving trajectory simulation for systems with evolving interactions.

Graph attention with structural features improves the generalizability of identifying functional sequences at a protein interface [new]

Billion-Scale Deciphering of Human Gene Regulatory Grammar [new] Million-scale data links degenerate seqs to gene expr in human cells. Reveals TF occupancy, coop regulation, & generalizable seq rules.

Learning Biomolecular Motion: The Physics-Informed Machine Learning Paradigm [new] Integrates physics with ML to model biomolecular systems, addressing the "closure problem" for kinetics, rare events, and free-energy estimation.

Dual-Pathway Fusion of EHRs and Knowledge Graphs for Predicting Unseen Drug-Drug Interactions [new] Predicts DDIs using knowledge graphs & EHRs via teacher/student model, generalizing to new drugs with interpretable mechanisms.

Integrating Epigenetic and Phenotypic Features for Biological Age Estimation in Cancer Patients via Multimodal Learning [new] Combines epigenetic and phenotypic data via multimodal learning to estimate biological age in cancer patients.

A Diffusion Model to Shrink Proteins While Maintaining Their Function [new] Develops a diffusion model to shrink proteins by learning to delete sequence letters, creating shorter, functional variants resembling natural proteins.

Assessing the feasibility of machine learning for ancient DNA age prediction: limitations and insights [new] aDNA damage-based ML age prediction fails. Context/decay models may improve it.

Can Lightweight LLM Agents Improve Spatial TranscriptomicsAnnotation? [new] Lightweight LLMs annotated spatial transcriptomics data by integrating rule-based heuristics and multi-role reasoning within an agentic framework.

Lightweight open-source fine-tuning of SAM2 enables domain-specific microscopy segmentation [new] Domain microscopy seg. w/ SAM2 fine-tuning & informed post-processing.

Enhancing Intra-Continental Biogeographical Ancestry Prediction Through a Machine Learning Marker Selection Method [new] ML refines ancestry prediction via marker selection, boosting intra-continental accuracy.

Learning the Unseen: Data-Augmented Deep Learning for PTM Discovery with Prosit-PTM [new] Data-augmented DL enables zero-shot PTM prediction, improving site ID & localization in proteomics.

SpNeigh: spatial neighborhood and differential expression analysis for high-resolution spatial transcriptomics [new] Spat. transcriptomics: context, boundaries, neighbs, & genes.

MetaboliteChat: A Unified Multimodal Large Language Model for Interactive Metabolite Analysis and Functional Insights [new] Multimodal LLM integrates metabolite graphs, images, and language for interactive analysis/prediction.

Language may be all omics needs: Harmonizing multimodal data for omics understanding with CellHermes [new] LLMs integrate omics via Q\&A pairs, emulating self-supervised learning.

Improving DNA Modeling with WaveDNA: Enhancing Speed, Generalizability, and Interpretability through Wavelet Transformation [new] DNA to 2D via wavelets. Enables light, interpretable deep learning using CV.

CONCERT predicts niche-aware perturbation responses in spatial transcriptomics [new] CONCERT: Predicts niche-aware perturbation responses in spatial transcriptomics using spatial kernels within a generative model.

HEIMDALL: A Modular Framework for Tokenization in Single-Cell Foundation Models [new] HEIMDALL: ...evaluates tokenization strategies in single-cell foundation models by modularizing components for fine-grained control and analysis.

An Explainable AI Framework for Identifying Universal Aging Signatures in Cell Embeddings [new] Identifies universal aging signatures in cell data by disentangling aging-related gene expression from other biological variations.

Dynamics-enhanced Molecular Property Prediction Guided by Deep Learning [new] Leverages molecular dynamics simulations and deep learning to enhance property prediction by considering molecular dynamics.

PyrMol: A Knowledge-Structured Pyramid Graph Framework forGeneralizable Molecular Property Prediction [new] Hierarchical GNNs boost molecule property prediction generalization by integrating multi-level chem knowledge.

Prognosis prediction using autophagy gene expression in osteosarcoma [new] Autophagy gene data informs a classifier (Auto-RS) for individualized osteosarcoma risk, capturing metastatic trajectories and revealing drug vulnerabilities.

Challenges in predicting chromatin accessibility differences between species [new] Models predict chromatin accessibility across species but struggle with quantitative differences in orthologous regions, highlighting limitations.

Inferring the regulation dynamics of oscillatorynetworks from scRNA-seq data [new] Inferred cell cycle positions enhance gene regulatory network inference, especially in early progenitor cells, by improving on temporal ordering.

HalluDesign: Protein Optimization and de novo Design via Iterative Structure Hallucination and Sequence design [new] Iteratively refines protein struct./seq. using "hallucination" for optimiz., design & sampling.

Examining Selection Dynamics and Limitations in Multi-round Protein Selection of High Diversity Libraries [new] Analyzes multi-round selection: diversity, coverage, reproducibility affect prot/pep discovery.

Control of gene output by intron RNA structure [new] Intron RNA structure significantly impacts gene output by modulating splicing efficiency, particularly near splice sites. Structure prediction models explain observed outcomes.

NEAT-DNA: A Chemically Accurate, Sequence-Dependent Coarse-Grained Model for Large-Scale DNA Simulations [new] Models DNA structure & simulates genomes w/ sequence accuracy. Combines physics, atomistic & experimental data.

Multi-Dimensional Spatiotemporal Attention Neural Network for Next Generation Sequencing Basecalling [new] NGS basecalls via spatiotemporal attention; extracts key info, skipping steps.

HeartMap: An Integrated Cell Atlas of 2.4 million cells across 209 Individuals in Health and Disease [new] Human heart atlas: single-nucleus RNA-seq reveals cell types & disease gene sigs.

3D Mapping of Intact Ovaries Reveals the Aging Dynamics of the Ovarian Reserve [new] Mouse ovary: Age, 3D mapping, AI reveal oocyte decline, fixed activation, size bimodality.

SpecLig: Energy-Guided Hierarchical Model for Target-Specific 3D Ligand Design [new] Gen. molecules/peptides w/ better target affinity & specificity via energy-guided SE(3) modeling.

Nona: A unifying multimodal masking framework for functional genomics [new] Unifies functional genomics models via multimodal masked modeling, learning from DNA sequence and base-resolution data for diverse applications.

GONNECT: A Gene Ontology-guided Neural Network for Explainable Cancer Typing [new] GO-guided network enhances deep learning interpretability, revealing bio processes for cancer typing.

A chromatin fiber model explains cell-free DNA fragmentation signatures of active regulatory elements [new] cfDNA fragmentation reveals enhancer activity via chromatin modeling & histone depletion.

Multi-agent AI enables evidence-based cell annotation in single-cell transcriptomics [new] Multi-agent AI: evidence-based cell annotation in single-cell transcriptomics. Uses expression data & databases for cell ID.

Unlocking Scalable Ligand Residence Time Predictions with Koffee Unbinding Kinetics Simulations [new] Koffee: Fast, scalable ligand-protein residence time prediction for improved drug discovery and kinetics.

Tear fluid as noninvasive liquid biopsy reveals proteins associated with malignant transformation of oral lesions [new] Tear fluid reflects oral lesion malignancy; immune & epithelial markers ID noninvasive detection.

Genomic Perception Fusion: A Lightweight, Interpretable Kernel for Protein Functional Tuning [new] DNA->functional forms via linear signals. Predicts protein props using physics.

Deep learning methods for automated bone segmentation and biomechanical modelling in zebrafish [new] 3D U-Net zebrafish mandible segmentation, validated against experts via biomechanics.

ELITE: E3 Ligase Inference for Tissue specific Elimination: A LLM Based E3 Ligase Prediction System for Precise Targeted Protein Degradation [new]

ECLIPSE: Exploration of Complex Ligand-Protein Interactions through Learning from Systems-level Heterogeneous Biomedical Knowledge Graphs [new]

mdBIRCH for Fast, Scalable, Online Clustering of Molecular Dynamics Trajectories [new] Adapts BIRCH, clusters MD via RMSD, merging by threshold.

Pro4S: prediction of protein solubility by fusing sequence, structure, and surface [new] Predicts protein solubility by integrating sequence, structure & surface data via contrastive learning, unifying qualitative/quantitative tasks.

Uncertainty-Aware Deep Learning for Multi-Metric and Dose-Specific Prediction of Drug Synergy [new] Deep net predicts drug synergy, handles metrics/dose responses, & quantifies uncertainty.

Squidly: Enzyme Catalytic Residue Prediction Harnessing a Biology-Informed Contrastive Learning Framework [2025-06-20] Predicts catalytic residues via contrastive learning on protein embeddings for large-scale function prediction.

Deep Learning for RNA Secondary Structure Determination: Gauging Generalizability and Broadening the Scope of Traditional Methods [new]

PETRA: Pretrained Evolutionary Transformer for SARS-CoV-2 Mutation Prediction [new] Predicts SARS-CoV-2 mutations using evolutionary trajectories from phylogenetic trees, addressing noise and data imbalances in genomic sequences.

Use of Continuous Glucose Monitoring with Machine Learning to Identify Metabolic Subphenotypes and Inform Precision Lifestyle Changes [new] CGM/ML reveal metabolic subtypes beyond glucose. PPGR/lifestyle data informs tailored dysglycemia interventions.