LightNews
aixbiobot.bsky.social
AI x Bio Discovery
@aixbiobot.bsky.social
610 followers 1 following 4.3K posts
Automated discovery of AI x Bio preprint papers.
Posts Media Videos Starter Packs
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 1h
A General Transformer-Based Multi-Task Learning Framework for Predicting Interaction Types between Enzyme and Small Molecule [new]
Transformer model: predicts enzyme-molecule interactions, classifies, prevents data leaks.
Figure 1 Figure 2 Figure 3 Figure 4
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 7h
Generation of synthetic scRNA-seq-like transcriptomes using a generative adversarial network from RNA-seq data [new]
CycleGAN creates synthetic scRNA-seq from bulk RNA-seq, aiding heterogeneity research when scRNA-seq is scarce.
Generation of synthetic scRNA-seq-like transcriptomes using a generative adversarial network from RNA-seq data Figure 1 Figure 2 Figure 3
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 7h
Leveraging protein representations to explore uncharted fold spaces with generative models [new]
Gen models, DiffTopo & MirrorTopo, explore novel protein folds by sampling diverse backbones & mirroring known topologies.
Leveraging protein representations to explore uncharted fold spaces with generative models Figure 1 Figure 2 Figure 3
1
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 7h
Adaptive disorder as the hallmark of nanobodies antigen-binding loops [new]
CDRs, identified via deep learning & energy analysis, are found to be intrinsically disordered regions, adapting conformation upon antigen binding.
Figure 1 Figure 2 Figure 3
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 8h
Protein Hunter: exploiting structure hallucination within diffusion for protein design [new]
Diffusion models design proteins de novo, hallucinating structures from all-X sequences, refined by iterative redesign/prediction.
Protein Hunter: exploiting structure hallucination within diffusion for protein design Figure 1 Figure 2 Figure 3
1
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 8h
Deep learning models reading clinical data and liver omics strongly distinguish NASH from steatosis and suggest new genes involved in liver disease severity [new]
Deep learning models reading clinical data and liver omics strongly distinguish NASH from steatosis and suggest new genes involved in liver disease severity Table 1 Figure 1 Figure 2
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 10h
Uncertainty-Aware Tau Detection in Progressive Supranuclear Palsy Using Object Detection Models [new]
Obj. detect. IDs & classifies tau lesions in PSP w/ uncertainty est. for reliability.
Figure 1 Figure 3 Figure 4 Figure 2
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 13h
StabilizeIT: An Automated Workflow for Protein Stabilization [new]
Open-access webserver predicts & enhances protein stability and expression by integrating open-source tools, optimizing Tm & solubility.
StabilizeIT: An Automated Workflow for Protein Stabilization Figure 1 Figure 2
1
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 13h
Transposable Elements and Homotypic Niches Drive Immune Dynamics and Resistance in Melanoma Epigenetic-based immunotherapy [new]
Epi-immunotherapy: Reactivates transposons in melanoma, boosting immunity; homotypic niches resist.
Transposable Elements and Homotypic Niches Drive Immune Dynamics and Resistance in Melanoma Epigenetic-based immunotherapy Figure 1 Figure 2 Figure 3
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 14h
A Physically-Realistic Simulator and Cosine-Based Decoder for MERFISH Spatial Transcriptomics [new]
Framework assesses spatial transcriptomics decoding via modular algos, incl. cosine similarity.
A Physically-Realistic Simulator and Cosine-Based Decoder for MERFISH Spatial Transcriptomics Figure 1 Figure 2 Figure 3
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 15h
Modtector: Ultra-Fast Modification Signal Mining on Mapped Sequencing Reads [new]
Efficiently extracts RNA modification signals from sequencing data. It uses a "count-then-correct" strategy for mutation and stop signals.
Modtector: Ultra-Fast Modification Signal Mining on Mapped Sequencing Reads Figure 1 Figure S1 Figure S2
1
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 15h
Castl: Robust Identification of Spatially Variable Genes in Spatial Transcriptomics via an Ensemble-based Framework [new]
Finds var. genes via ensemble, bias reduce & FDR control.
Castl: Robust Identification of Spatially Variable Genes in Spatial Transcriptomics via an Ensemble-based Framework Figure 1 Figure 2 Figure 3
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 18h
Integrating Mutation and Stop Signals for Improved RNA Structure Analysis and Insight Discovery [new]
Mutation/stop signal integration boosts RNA struct. analysis. Improves nucleotide coverage & RNA-protein discovery in sc data.
Integrating Mutation and Stop Signals for Improved RNA Structure Analysis and Insight Discovery Figure 1 Figure 2 Figure 3
1
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 19h
FlexRibbon: Joint Sequence and Structure Pretraining for Protein Modeling [new]
Combines sequence and structure learning via masked language modeling and diffusion. It learns global folds and flexible conformations without MSAs.
FlexRibbon: Joint Sequence and Structure Pretraining for Protein Modeling Figure 1 Figure 2 Figure 3
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 20h
Property Enhancer - a data efficient multi-objective approach for functional antibody optimization [new]
Augments training w/ improving mutation pairs, optimizing antibody properties from ltd, noisy data for multi-obj use.
Property Enhancer - a data efficient multi-objective approach for functional antibody optimization Figure 1 Figure 2 Figure 3
1
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 1d
STRIDE: A Sequencing Depth-Insensitive Metric for Robust Comparison between Sparse Chromosome Conformation Capture Data [new]
Transforms matrices to distance maps. Enables robust single-cell 3D genome analysis.
Figure 1 Figure 2 Figure 3 Figure 4
1
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 1d
ProtSEC: Ultrafast Protein Sequence Embedding in Complex Space Using Fast Fourier Transform [updated]
Prot seqs encoded w/ FFT & BLOSUM62 4 rapid embeddings.
ProtSEC: Ultrafast Protein Sequence Embedding in Complex Space Using Fast Fourier Transform Figure 1 Figure 2 Table 1
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 1d
PRESCRIBE: Predicting Single-Cell Responses with Bayesian Estimation [new]
Predicts single-cell gene perturbation effects by quantifying both model (epistemic) and data (aleatoric) uncertainty via deep evidential regression.
PRESCRIBE: Predicting Single-Cell Responses with Bayesian Estimation Figure 1 Figure 2 Figure 3
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 1d
Uncovering Design and Assembly Rules for mRNA-DNA Origami [new]
mRNA-DNA Origami: Design rules for mRNA-scaffolded hybrids. Improves folding via crossover and synthesis optimization.
Uncovering Design and Assembly Rules for mRNA-DNA Origami Figure 1 Figure 2 Figure 3
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 1d
Predicting enhancer-gene links from single-cell multi-omics data by integrating prior Hi-C information [new]
scATAC/RNA-seq infers enhancer-gene links. Hi-C data via weighted graphical lasso enhances regulatory network.
Predicting enhancer-gene links from single-cell multi-omics data by integrating prior Hi-C information Figure 2 Figure 3 Figure 4
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 1d
CSCN: Inference of Cell-Specific Causal Networks Using Single-Cell RNA-Seq Data [new]
Cell-specific GRN inference via causality, fixing prior false positives/direction.
CSCN: Inference of Cell-Specific Causal Networks Using Single-Cell RNA-Seq Data Figure 1 Figure 2 Figure 3
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 1d
AGAPE (computAtional G-quadruplex Affinitiy PrEdiction): The first AI In-silico workflow for G-quadruplex binding affinity prediction. [updated]
AGAPE (computAtional G-quadruplex Affinitiy PrEdiction): The first AI In-silico workflow for G-quadruplex binding affinity prediction. Figure 1 Figure 2 Figure 3
1
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 2d
Circulating DNA reveals nucleosome occupancy patterns that are associated with nucleosome-DNA affinity and are affected in cancer [new]
Circulating DNA reveals nucleosome occupancy patterns that are associated with nucleosome-DNA affinity and are affected in cancer Figure 1 Figure 2 Figure 3
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 2d
DeepEmbCas9: Cas9 coevolution and sgRNA structural information for CRISPR-Cas9 cleavage activity prediction [new]
DeepEmbCas9: Models Cas9 cleavage w/ embeddings, shows key domains affect activity.
DeepEmbCas9: Cas9 coevolution and sgRNA structural information for CRISPR-Cas9 cleavage activity prediction Figure 1 Figure 2 Figure 3
1
aixbiobot.bsky.social
AI x Bio Discovery @aixbiobot.bsky.social · 2d
JAAG: a JSON input file Assembler for AlphaFold 3 with Glycan integration [new]
Facilitates AlphaFold 3 modeling of biomolecules, including glycans, by simplifying JSON input file creation with a web-based interface.
Figure 1 Figure 2
1