The front page of AI-driven biology
Friday, October 16, 2026Fri, Oct 16
Updated 12:01 AM PT
Recent original research from bioRxiv and arXiv. Preprints have not been peer reviewed.
A computational protein-design system combined sparse experimental measurements with evolutionary and structural data to simultaneously engineer catalytic efficiency, expression yield, and thermal stability in the industrial ketoreductase Gre2.
Combining a metastatic colonization score called MetScore with standard classical-basal tumor subtypes better predicts patient survival and chemotherapy response in pancreatic ductal adenocarcinoma than the single-subtype model alone.
MicroClaw is an artificial intelligence assistant designed to plan and execute automated imaging experiments across multiple microscope platforms using natural language prompts.
Researchers built a mechanistic computational model combining chromatin dynamics, in vitro response data, and drug distribution kinetics to simulate how triple-negative breast cancer responds to combination epigenetic therapy.
A computational team introduced AmbientMapper, a generative model designed to decontaminate high-throughput single-cell combinatorial indexing experiments by learning ambient background profiles from empty droplets.
A wet-lab validation study comparing generative adversarial networks and variational autoencoders found that sampling protocols, rather than model architecture alone, dictate the experimental success of computer-designed antibodies.
A computational study demonstrates that combining machine-learned interatomic potentials with electrostatic embedding can accurately calculate catalytic reaction barriers in designed enzymes without costly quantum mechanical simulations.
Researchers built IGNITE, a semi-supervised machine learning framework that ranks therapeutic targets by combining human genetic evidence with Perturb-seq profiles from 22 million primary human CD4+ T cells.
A multi-agent framework called scACORN coordinates specialized small language models rather than a single monolithic model to interpret single-cell transcriptomic profiles.
A retrospective benchmark using Pfam updates revealed that structural homology searches using Foldseek recovered correct functions for only 15.9% of previously uncharacterized protein domains, compared with 30.1% for difficulty-matched known domains.
A statistical model named PRIME uses a hierarchical Bayesian framework to estimate regional missense intolerance across genes by sharing evolutionary information across sub-genic segments.
An automated spatial computing pipeline analyzed nanoscale expansion microscopy images of brain tissue from the somatosensory cortex of 5xFAD Alzheimer's disease mice.
Researchers introduced Dynamo, an open-source Python tool designed to reconstruct, register, and track changes in branching neuronal dendrites and axons across time-lapse imaging series.
Researchers released POME, a self-supervised model that creates low-dimensional numerical representations for messy patient records characterized by mixed data types and missing values.
DiffGSP combines Fick's laws of diffusion with graph signal processing to model and reverse the spread of mRNA molecules between adjacent spots during tissue processing.
Researchers built REN-former by fine-tuning the Geneformer foundation model on single-cell RNA sequencing from 45 kidney patient samples to predict directional cell shifts from acute kidney injury to chronic kidney disease.
A new generative method called high-dimensional adversarial random forests (h-ARF) synthesizes tabular molecular profiles to supplement small biological sample cohorts from hard-to-access tissues.
Using the reinforcement-learning molecular design tools Link-INVENT and AutoMolDesigner, researchers designed Y-11, a small-molecule antibiotic targeting MsbA, an essential transporter protein in Gram-negative bacterial outer-membrane assembly.
A computational study showed that splitting single-cell datasets randomly between training and test sets leaks labels through related clonal sisters, patient backgrounds, or experimental batches.
A hierarchical transformer architecture called HTT combines report-level language embeddings with continuous-time visit intervals to analyze sequential cancer pathology records.
ProtScape combines global protein interaction data, single-cell gene expression, and protein language models to reconstruct context-specific protein interactomes across more than 200 cell types.
Computational biologists developed a multistate design framework that optimizes enzyme active sites across an entire reaction pathway rather than modeling a single rigid transition state.
Researchers tested multi-eGO, an atomistic simulation model developed for small molecules, on binding between the human PDZ2 protein domain and an experimental peptide.
LucaCell is a single-cell foundation model that encodes transcriptomic features through pre-trained mRNA nucleotide sequence embeddings rather than fixed gene names, allowing it to translate across species without manual ortholog mapping.
Therna Biosciences has released public datasets from its Chronos platform profiling roughly 60,000 synthetic regulatory elements across approximately 50 pooled cell lines.
Computational biologists introduced PANDA, an end-to-end generative architecture that designs proteins by denoising atomic coordinates in unified 3D space and reading sequence directly from atomic positions.
Researchers introduced ICARus, a computational framework that uses protein language model predictions to reduce off-target noise in RNA interference functional screens.
Researchers released HARIBOSS++, an updated structural database and web platform cataloging small molecules that bind RNA.
Profiling 384 plasma proteins across 62 untreated HIV elite controllers revealed that a distinct subset of 9 individuals has inflammatory profiles matching people without HIV.
Researchers adapted AgroNT, a transformer-based genomic foundation model, by adding stress-conditioned prompt tokens to pinpoint regulatory regions tied to heat and drought resilience in maize.