1012 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.
MoE-Bind introduces the first sparse Mixture-of-Experts architecture for de novo protein binder generation, achieving faster, more efficient, and interpretable sequence-only design that matches or exceeds dense baselines without requiring prior structural information.
This paper introduces Latent Event Mapping (LEMING), a scalable and interpretable unsupervised modeling technique that reconstructs voxel-level trajectories of multi-modal neuroimaging changes in Alzheimer's disease, revealing new insights into progression-dependent mechanisms such as the late-stage association between acetylcholine receptor density and amyloid pathology.
The paper introduces STITCH, an interpretable method based on tangent principal component analysis in a Kendall shape manifold that effectively links size-independent cellular morphology to gene expression in spatial transcriptomics, outperforming existing deep learning approaches in identifying biologically relevant shape-transcriptome relationships across diverse datasets.
This study introduces a split conformal prediction framework applied to the BeatAML 2.0 cohort that successfully generates statistically calibrated uncertainty intervals for patient-level AML drug sensitivity predictions, revealing distinct uncertainty patterns across drug classes while demonstrating that such uncertainty is independent of standard molecular risk classifications.
The paper introduces scCRAFT+, a robust semi-supervised scRNA-seq integration model that leverages Virtual Adversarial Training to incorporate marker gene information, thereby overcoming the limitations of existing methods in preserving fine-grained cell subtype distinctions and improving annotation accuracy even with noisy or incomplete marker sets.
Pillbox is a leakage-audited foundation-model predictor that integrates CpGPT, CLAMP, and gene-expression embeddings via FiLM-conditioned graph attention to achieve high accuracy in cancer drug response prediction while using cross-architecture residual correlation to diagnose feature-stack saturation and confirm that tissue lineage remains the dominant factor in drug response.
This paper introduces GermRL, a lightweight reinforcement learning framework that effectively alleviates germline bias in autoregressive antibody language models, enabling the one-shot generation of structurally plausible, diverse antibody candidates with high mutation thresholds from germline sequences for therapeutic discovery.
DivQuant is an optimization-based tool that improves the estimation of species richness and Shannon entropy from small samples by formulating the problem as a convex quadratic program with a Neyman objective, thereby achieving well-calibrated confidence intervals and superior accuracy compared to state-of-the-art methods across diverse biological datasets.
The T21 Research Assistant is a section-aware retrieval-augmented generation system built on NVIDIA Nemotron models that prioritizes experimental results from 1,789 open-access Down syndrome publications to provide rigorously cited, evidence-based answers to research queries.
EditorForge is a modular framework that enhances inverse-folding-based protein redesign by integrating explicit design masks, active-site shielding, and structural quality control to safely generate constrained, high-confidence sequence candidates for enzyme and genome-editor domains.