Discover, rate, and discuss biology preprints.

CPPLocPred predicts subcellular localization of cell-penetrating peptides using a hierarchical machine learning framework, achieving high accuracy on validation datasets.

The paper introduces a tensor decomposition method for integrating CITE-seq RNA and ADT data without modality-weight tuning, showing competitive but not superior performance compared to existing methods.

This paper introduces AMPainterV2, a model that designs membrane-lytic antimicrobial peptides using generative adversarial imitation learning, achieving active peptides with improved activity and miniaturization.

ComBatFamQC provides an R package for batch-effect diagnosis and harmonization in neuroimaging data, integrating visualization, statistical tests, and harmonization methods.

MHChron improves pMHC binding prediction by creating a diverse dataset and testing models under strict generalization conditions, achieving strong performance despite smaller training data.

The paper develops variational inference methods for amino acid coevolution models, showing improved efficiency and accuracy over existing approaches.

BWR-finder detects interspersed repeats in large genomes using a BWT-based algorithm, outperforming existing tools in speed and memory usage.

DrugLM enhances drug-target interaction predictions by integrating textual embeddings from large language models into DTI models, showing strong performance without fine-tuning.

This paper benchmarks 7 TR genotyping tools using Nanopore long-read data, finding that no single tool is best across all metrics, and that sequence-level accuracy is crucial for clinical and population studies.

The paper identifies a microbial module, the FERM guild, associated with hypertension through metabolic flux changes using pFBA and DoubleML.

The paper presents a framework using BERT embeddings and WGAN-GP to predict pathway associations for human dark kinase proteins, achieving high accuracy in classification and pathway clustering.

This study uses single-cell RNA sequencing to analyze human retinal samples and identifies microglial states and neural-immune interactions in diabetic retinopathy.

This paper develops the Fontan EV Score, a non-invasive biomarker using circulating extracellular vesicles to predict Fontan-associated liver disease severity, outperforming existing serological indices.

This paper introduces CLSS, a contrastive learning model that co-embeds protein sequences and structures, revealing evolutionary relationships and structural transitions.

The paper evaluates local ancestry inference methods across varying FST values, finding a feasibility threshold and showing that input quality, not architecture, limits performance.

crispAIPE quantifies uncertainty in prime editing pegRNA efficiency predictions using a transformer-based probabilistic model, achieving high correlation with experimental data and providing coverage guarantees.

This study evaluates whether geometric outliers in single-cell foundation models identify important genes, finding model-specific patterns without consistent biological significance.

The paper develops CGAP-HMM and a voting system to improve transcription unit annotation from run-on and sequencing data, showing significant performance gains over existing methods.

Nipoppy standardizes neuroimaging data workflows to enable reproducible and comparable multi-center data sharing.

TM-Vec 2s improves protein remote homology detection by being faster and more accurate than previous methods, enabling efficient large-scale database searches.