974 papers
The paper introduces PerturbGraph, a graph neural network framework that integrates biological priors such as protein-protein interaction networks and functional annotations to accurately predict transcriptional responses for unseen gene perturbations, significantly outperforming existing machine learning and deep learning baselines.
Nextstrain is an automated platform that utilizes open genomic data to provide continually updated, real-time phylodynamic surveillance and visualization for 21 viruses and *Mycobacterium tuberculosis*, thereby enabling targeted public health interventions.
This paper demonstrates that discrete graph diffusion models, specifically graph transformers trained on synthetic tumor phylogenies, can effectively learn and generate structurally valid evolutionary trees, revealing that mid-scale architectures and diverse training regimes yield superior generalization compared to deeper models or specialized single-regime training.
This paper reanalyzes datasets comparing PELSA and LiP-MS to demonstrate that the reported superior sensitivity of PELSA, particularly a 21-fold difference in FKBP1A quantification, stems from non-matched experimental conditions and undisclosed data imputation, thereby warranting caution in accepting claims of its quantitative superiority.
FoundedPBI is an ensemble deep learning framework that leverages genomic foundation models and novel long-context aggregation strategies to accurately predict phage-bacterium interactions from DNA sequences, significantly outperforming existing state-of-the-art methods on benchmark datasets.
The paper introduces GraphHDBSCAN*, a hyperparameter-free, graph-based hierarchical clustering method that effectively recovers both fine-grained flat partitions and biologically meaningful hierarchical structures in high-dimensional, sparse single-cell RNA sequencing data, outperforming existing state-of-the-art approaches.
The OPTIMIS framework addresses the challenge of controlling complex multiscale biological systems by integrating a hybrid stochastic-deterministic model into a differentiable Neural ODE surrogate, enabling deep reinforcement learning agents to successfully predict and prevent dangerous immune reactions in engineered cellular therapies with over 70% success rates.
Allos is an open-source, modular Python toolkit built on the AnnData model that unifies isoform-level analysis, visualization, and interpretation for single-cell and spatial transcriptomics across both long- and short-read sequencing data.
This study utilizes structural modeling and computational analysis of 107 consecutive enzyme pairs in *E. coli* to demonstrate that, despite a tendency for these enzymes to interact, their catalytic sites are not systematically positioned in spatially optimized configurations to facilitate metabolite transfer.
The paper introduces Tripso, a self-supervised transformer model that moves beyond single latent representations to learn interpretable, gene program-centric embeddings of cell states, thereby enabling the discovery of biologically meaningful patterns and actionable hypotheses across development, disease, and experimental systems.
This paper demonstrates that applying Reinforcement Learning with Group Relative Policy Optimization to DNA language models not only significantly improves plasmid generation quality but also induces emergent biological realism, such as natural thermodynamic stability and codon usage patterns, without explicit optimization for these specific traits.
The paper introduces SpaMosaic, a scalable tool leveraging contrastive learning and graph neural networks to integrate heterogeneous spatial multi-omics datasets into a unified latent space for robust domain identification, batch correction, and accurate imputation of missing modalities.
This paper introduces Signature Distance (SD), a structural generalization of energy distance that compares empirical distributions via sorted pointwise distance profiles to detect local density and topological changes, offering a differentiable and computationally efficient metric for improved generative model evaluation, hypothesis testing, and data augmentation in high-dimensional biological data.
This paper introduces "fitness translocation," a data augmentation strategy that leverages variant fitness data from homologous proteins to generate synthetic training examples in embedding space, thereby significantly improving the accuracy of protein variant effect prediction models, particularly when training data is scarce.
This paper introduces Chromatix, an open-source, GPU-accelerated, differentiable wave-optics library built on JAX that standardizes optical simulations to overcome reusability and performance limitations in computational microscopy, achieving significant speedups across applications like holography and phase retrieval.
The paper introduces cellSight, an automated, user-friendly workflow designed to streamline single-cell RNA-sequencing analysis by reducing manual effort, enhancing reproducibility, and accelerating biological discoveries.
This paper introduces a new phylogenetic statistical method implemented in the Pelican software to detect protein sites associated with continuous traits, demonstrating improved sensitivity for genome-wide scans while revealing weak evidence for longevity-associated sites in previously studied mammalian protein families.
By integrating genomic, metabolomic, and microbiome data from 198 soybean accessions, this study employs an interpretable machine learning approach to reveal that the isoflavone daidzin and the bacterium *Candidatus Nitrosocosmicus* are key drivers of drought resilience through specific plant-microbe interactions in the rhizosphere.
The paper introduces Cenote-Taker 3, a high-performance, open-source command-line tool that significantly outperforms existing methods in speed and accuracy for discovering and annotating viral genomes and prophages from sequencing data, while also offering complementary results to other leading tools like geNomad.
The paper introduces PATTY, a computational method that leverages ATAC-seq data and machine learning to correct Tn5 transposase-induced open chromatin bias in both bulk and single-cell CUT&Tag datasets, thereby enabling more accurate detection of histone modification occupancy and improved cell clustering.