AI x Bio Discovery
@aixbiobot.bsky.social
Automated discovery of AI x Bio papers, blogs, and news.
A single-cell atlas linking intratumoral states to therapeutic vulnerabilities across cancers [new]
Maps therapy heterogeneity w/ 10 clusters, DR shaped by functional prog & TME states, not genomics, guides precision oncology.
Maps therapy heterogeneity w/ 10 clusters, DR shaped by functional prog & TME states, not genomics, guides precision oncology.
February 20, 2026 at 5:17 AM
A single-cell atlas linking intratumoral states to therapeutic vulnerabilities across cancers [new]
Maps therapy heterogeneity w/ 10 clusters, DR shaped by functional prog & TME states, not genomics, guides precision oncology.
Maps therapy heterogeneity w/ 10 clusters, DR shaped by functional prog & TME states, not genomics, guides precision oncology.
February 20, 2026 at 5:15 AM
February 20, 2026 at 4:52 AM
Pan-cell-type prediction of splicing patterns from sequence and splicing factor expression [new]
...predicts context-dependent splicing by integrating cis-sequence and trans-factor expr., enabling generalization to unseen cell types.
...predicts context-dependent splicing by integrating cis-sequence and trans-factor expr., enabling generalization to unseen cell types.
February 20, 2026 at 4:50 AM
Pan-cell-type prediction of splicing patterns from sequence and splicing factor expression [new]
...predicts context-dependent splicing by integrating cis-sequence and trans-factor expr., enabling generalization to unseen cell types.
...predicts context-dependent splicing by integrating cis-sequence and trans-factor expr., enabling generalization to unseen cell types.
GFMBench-API: A Standardized Interface for Benchmarking Genomic Foundation Models [new]
unifies evaluation through modular 'middleware', decoupling model specifics from task data streams and metrics via standard I/O schemas.
unifies evaluation through modular 'middleware', decoupling model specifics from task data streams and metrics via standard I/O schemas.
February 20, 2026 at 4:49 AM
GFMBench-API: A Standardized Interface for Benchmarking Genomic Foundation Models [new]
unifies evaluation through modular 'middleware', decoupling model specifics from task data streams and metrics via standard I/O schemas.
unifies evaluation through modular 'middleware', decoupling model specifics from task data streams and metrics via standard I/O schemas.
February 20, 2026 at 4:26 AM
Causal gene regulatory network inference from Perturb-seq via adaptive instrumental variable modeling [new]
...addresses heterogeneous knockdown to infer cyclic structs via a Poisson-lognormal model, establishing a scalable approach.
...addresses heterogeneous knockdown to infer cyclic structs via a Poisson-lognormal model, establishing a scalable approach.
February 20, 2026 at 3:41 AM
Causal gene regulatory network inference from Perturb-seq via adaptive instrumental variable modeling [new]
...addresses heterogeneous knockdown to infer cyclic structs via a Poisson-lognormal model, establishing a scalable approach.
...addresses heterogeneous knockdown to infer cyclic structs via a Poisson-lognormal model, establishing a scalable approach.
Exploring the Utility of MALDI-TOF Mass Spectrometry and Antimicrobial Resistance in Hospital Outbreak Detection [new]
...by developing a ML framework that extracts & fuses informative representations from their spectra and AR patterns for rapid surveillance.
...by developing a ML framework that extracts & fuses informative representations from their spectra and AR patterns for rapid surveillance.
February 20, 2026 at 3:28 AM
Exploring the Utility of MALDI-TOF Mass Spectrometry and Antimicrobial Resistance in Hospital Outbreak Detection [new]
...by developing a ML framework that extracts & fuses informative representations from their spectra and AR patterns for rapid surveillance.
...by developing a ML framework that extracts & fuses informative representations from their spectra and AR patterns for rapid surveillance.
Omni-iEEG: A Large-Scale, Comprehensive iEEG Dataset and Benchmark for Epilepsy Research [new]
...integrating 302 patients, 178 hours of iEEG with harmonized clinical metadata and 36K pathological event annotations, enabling reproducible epilepsy research.
...integrating 302 patients, 178 hours of iEEG with harmonized clinical metadata and 36K pathological event annotations, enabling reproducible epilepsy research.
February 20, 2026 at 3:26 AM
Omni-iEEG: A Large-Scale, Comprehensive iEEG Dataset and Benchmark for Epilepsy Research [new]
...integrating 302 patients, 178 hours of iEEG with harmonized clinical metadata and 36K pathological event annotations, enabling reproducible epilepsy research.
...integrating 302 patients, 178 hours of iEEG with harmonized clinical metadata and 36K pathological event annotations, enabling reproducible epilepsy research.
Construction of a classification model for dementia among Brazilian adults aged 50 and over [new]
using low-cost, modifiable ELSI-Brazil variables to identify key risk/protective factors, informing public health policies for early detection and prevention.
using low-cost, modifiable ELSI-Brazil variables to identify key risk/protective factors, informing public health policies for early detection and prevention.
February 20, 2026 at 3:25 AM
Construction of a classification model for dementia among Brazilian adults aged 50 and over [new]
using low-cost, modifiable ELSI-Brazil variables to identify key risk/protective factors, informing public health policies for early detection and prevention.
using low-cost, modifiable ELSI-Brazil variables to identify key risk/protective factors, informing public health policies for early detection and prevention.
JEPA-DNA: Grounding Genomic Foundation Models through Joint-Embedding Predictive Architectures [new]
predicts high-level functional embeddings of masked genomic segments in latent space, coupling token recovery with a global biological context.
predicts high-level functional embeddings of masked genomic segments in latent space, coupling token recovery with a global biological context.
February 20, 2026 at 3:24 AM
JEPA-DNA: Grounding Genomic Foundation Models through Joint-Embedding Predictive Architectures [new]
predicts high-level functional embeddings of masked genomic segments in latent space, coupling token recovery with a global biological context.
predicts high-level functional embeddings of masked genomic segments in latent space, coupling token recovery with a global biological context.
Systematic Evaluation of Single-Cell Foundation Model Interpretability Reveals Attention Captures Co-Expression Rather Than Unique Regulatory Signal [new]
Attention encodes biol. structure, but offers no added value for pert prediction beyond gene-lvl co-exp.
Attention encodes biol. structure, but offers no added value for pert prediction beyond gene-lvl co-exp.
February 20, 2026 at 3:22 AM
Systematic Evaluation of Single-Cell Foundation Model Interpretability Reveals Attention Captures Co-Expression Rather Than Unique Regulatory Signal [new]
Attention encodes biol. structure, but offers no added value for pert prediction beyond gene-lvl co-exp.
Attention encodes biol. structure, but offers no added value for pert prediction beyond gene-lvl co-exp.
Probability-Invariant Random Walk Learning on Gyral Folding-Based Cortical Similarity Networks for Alzheimer's and Lewy Body Dementia Diagnosis [new]
classifies indiv. gyral folding nets using P-invariant dists of anon. RWs without explicit node alignment.
classifies indiv. gyral folding nets using P-invariant dists of anon. RWs without explicit node alignment.
February 20, 2026 at 3:20 AM
Probability-Invariant Random Walk Learning on Gyral Folding-Based Cortical Similarity Networks for Alzheimer's and Lewy Body Dementia Diagnosis [new]
classifies indiv. gyral folding nets using P-invariant dists of anon. RWs without explicit node alignment.
classifies indiv. gyral folding nets using P-invariant dists of anon. RWs without explicit node alignment.
EchoVisuALL: From Echocardiography to Gene Discovery [new]
...through an AI pipeline for automated mouse echo phenotyping, revealing 37 genes linked to cardiac abnormalities, including 12 novel candidates.
...through an AI pipeline for automated mouse echo phenotyping, revealing 37 genes linked to cardiac abnormalities, including 12 novel candidates.
February 20, 2026 at 2:56 AM
EchoVisuALL: From Echocardiography to Gene Discovery [new]
...through an AI pipeline for automated mouse echo phenotyping, revealing 37 genes linked to cardiac abnormalities, including 12 novel candidates.
...through an AI pipeline for automated mouse echo phenotyping, revealing 37 genes linked to cardiac abnormalities, including 12 novel candidates.
February 20, 2026 at 2:55 AM
NaVis: a virtual microscopy framework for interactive, high-resolution navigation of spatial transcriptomics data [new]
On-demand super-res inference from low-res whole-transcriptome via code-free UI for dynamic exploratory analysis
On-demand super-res inference from low-res whole-transcriptome via code-free UI for dynamic exploratory analysis
February 20, 2026 at 1:05 AM
NaVis: a virtual microscopy framework for interactive, high-resolution navigation of spatial transcriptomics data [new]
On-demand super-res inference from low-res whole-transcriptome via code-free UI for dynamic exploratory analysis
On-demand super-res inference from low-res whole-transcriptome via code-free UI for dynamic exploratory analysis
Spartan: Spatial Activation Aware Transcriptomic Analysis Network [new]
...is a multiplex graph framework integrating localized activation signals to delineate tissue domains and identify spatially variable genes.
...is a multiplex graph framework integrating localized activation signals to delineate tissue domains and identify spatially variable genes.
February 20, 2026 at 1:04 AM
Spartan: Spatial Activation Aware Transcriptomic Analysis Network [new]
...is a multiplex graph framework integrating localized activation signals to delineate tissue domains and identify spatially variable genes.
...is a multiplex graph framework integrating localized activation signals to delineate tissue domains and identify spatially variable genes.
Integrated inference of cancer gene expression from cell-free plasma chromatin [new]
...achieved by analyzing histone marks and DNA frag patterns for biopsy-free profiling tumor transcriptional states and identifying biomarkers.
...achieved by analyzing histone marks and DNA frag patterns for biopsy-free profiling tumor transcriptional states and identifying biomarkers.
February 20, 2026 at 12:40 AM
Integrated inference of cancer gene expression from cell-free plasma chromatin [new]
...achieved by analyzing histone marks and DNA frag patterns for biopsy-free profiling tumor transcriptional states and identifying biomarkers.
...achieved by analyzing histone marks and DNA frag patterns for biopsy-free profiling tumor transcriptional states and identifying biomarkers.
Experimental Time Points Guided Transcriptomic Velocity Inference [new]
...uses semi-supervised learning to integrate experimental time supervision, enabling precise inference of velocities and intrinsic gene-embedded time.
...uses semi-supervised learning to integrate experimental time supervision, enabling precise inference of velocities and intrinsic gene-embedded time.
February 19, 2026 at 10:48 PM
Experimental Time Points Guided Transcriptomic Velocity Inference [new]
...uses semi-supervised learning to integrate experimental time supervision, enabling precise inference of velocities and intrinsic gene-embedded time.
...uses semi-supervised learning to integrate experimental time supervision, enabling precise inference of velocities and intrinsic gene-embedded time.
Benchmarking Large Language Models for Predicting Therapeutic Antisense Oligonucleotide Efficacy [new]
...revealed DNA sequence + gene context are key for efficacy prediction, positioning general LLMs as promising tools for ASO dsgn.
...revealed DNA sequence + gene context are key for efficacy prediction, positioning general LLMs as promising tools for ASO dsgn.
February 19, 2026 at 10:04 PM
Benchmarking Large Language Models for Predicting Therapeutic Antisense Oligonucleotide Efficacy [new]
...revealed DNA sequence + gene context are key for efficacy prediction, positioning general LLMs as promising tools for ASO dsgn.
...revealed DNA sequence + gene context are key for efficacy prediction, positioning general LLMs as promising tools for ASO dsgn.
February 19, 2026 at 10:03 PM
Toward universal steering and monitoring of AI models [new]
by developing a comprehensive framework for controlling and observing varied AI system functions.
by developing a comprehensive framework for controlling and observing varied AI system functions.
February 19, 2026 at 8:16 PM
Toward universal steering and monitoring of AI models [new]
by developing a comprehensive framework for controlling and observing varied AI system functions.
by developing a comprehensive framework for controlling and observing varied AI system functions.
When a health tech startup gets stuck in regulatory limbo [blog]
...it's due to significant hurdles in deploying AI for clinical healthcare tasks, despite funding.
...it's due to significant hurdles in deploying AI for clinical healthcare tasks, despite funding.
February 19, 2026 at 7:32 PM
When a health tech startup gets stuck in regulatory limbo [blog]
...it's due to significant hurdles in deploying AI for clinical healthcare tasks, despite funding.
...it's due to significant hurdles in deploying AI for clinical healthcare tasks, despite funding.
February 19, 2026 at 7:10 PM
ATLAS: Graph-based 3D RNA Motif Library Incorporating non-Watson-Crick Interactions [new]
...integrates PDB 3D structures and pseudoknots, offering web search for motifs, a similarity measure, and an RNA evolution model.
...integrates PDB 3D structures and pseudoknots, offering web search for motifs, a similarity measure, and an RNA evolution model.
February 19, 2026 at 2:56 PM
ATLAS: Graph-based 3D RNA Motif Library Incorporating non-Watson-Crick Interactions [new]
...integrates PDB 3D structures and pseudoknots, offering web search for motifs, a similarity measure, and an RNA evolution model.
...integrates PDB 3D structures and pseudoknots, offering web search for motifs, a similarity measure, and an RNA evolution model.