1199 papers
Bioinformatics sits at the exciting intersection where biology meets data science, using powerful computer tools to decode the vast complexity of living systems. From mapping the human genome to tracking how viruses evolve, this field transforms raw biological information into actionable insights that drive modern medicine and research forward without requiring a supercomputer to understand the basics.
On Gist.Science, we ensure you never miss a breakthrough by processing every new preprint in this category directly from bioRxiv. Our team provides both plain-language explanations and detailed technical summaries for each paper, making cutting-edge discoveries accessible to everyone regardless of their background.
Below are the latest bioinformatics papers added from bioRxiv, ready for you to explore with clarity and depth.
The paper introduces Handshake, a sequence-only deep learning model that leverages ProstT5 and cross-chain attention to accurately predict partner-specific protein-protein binding sites at scale, while also revealing that high sequence redundancy in training data significantly inflates performance metrics in existing benchmarks.
This study identifies a divergent surface loop forming "Pocket2" as a key source of structural and functional variation in shark legumains, contrasting with their otherwise highly conserved catalytic cores and suggesting this region as a target for future investigation into species-specific protease adaptation.
The paper introduces CryoDiff, an uncertainty-aware diffusion model that outperforms existing methods in enhancing cryo-EM maps by jointly improving structural resolution and providing voxel-wise confidence estimates, thereby significantly boosting map interpretability and downstream model building.
The paper introduces cellpin, a scalable variational autoencoder trained exclusively on single-cell RNA sequencing data that utilizes teacher-student latent distillation and noise simulation to effectively impute unmeasured genes and denoise spatial transcriptome profiles without requiring cross-modality alignment.
OmniGene-4 introduces a unified, router-interpretable Mixture-of-Experts foundation model that demonstrates how continued pretraining drives cross-task specialization while expert computation handles sequence-grounded biological reasoning, achieving state-of-the-art performance in protein homology and general biological knowledge with significantly reduced compute costs even when extended to multi-modal inputs.
The paper introduces UnBlender, a pipeline designed to enable respiratory scientists to perform customizable cell type deconvolution on bulk RNA-seq data while routinely validating the accuracy of the estimated cell type proportions to ensure reproducible and reliable conclusions.
This paper presents an interpretable machine learning framework that combines dimensionality reduction with a multi-output random forest model to accurately infer dog breed identity and ancestry from genome-wide SNP data, achieving 91.7% accuracy on the Dog Aging Project dataset while identifying biologically relevant genetic loci associated with breed-specific traits.
This paper introduces ROTS 2.0, an enhanced open-source framework available in R and Python that extends the reproducibility-optimized test statistic to diverse high-throughput omics study designs, including multi-group comparisons and survival analysis, to improve the reliability of feature selection in complex experimental settings.
This paper demonstrates that a correctness-gated coding agent can significantly outperform the established RapidNJ baseline in computational phylogenetics by producing SwiftNJ, an optimized Neighbour Joining implementation that achieves a 0.565 geometric-mean runtime ratio while maintaining exact correctness against reference standards.
Leviathan is an open-source software package that enables ultra-fast, memory-efficient, and accurate taxonomic and functional profiling of metagenomes and metatranscriptomes at genome and pangenome resolution by combining alignment-free taxonomic methods with DNA-space pseudo-alignment to bypass computationally expensive translated-search steps.