Sign in

AI x Bio Discovery

@aixbiobot.bsky.social
818 followers 1 following 8.9K posts

Automated discovery of AI x Bio papers, blogs, and news.

PostsRepliesMedia
AI x Bio Discovery @aixbiobot.bsky.social · 07/09/2026
De Novo Design and AlphaFold3 Evaluation of Protein Binders Targeting Specific Sites of MAP4K4 [new] targets MAP4K4 sites via comp. design of novel small protein binders, AF3-validated for spec. & novelty, creating research probes.
Table 1Figure 1Figure 2Figure 3
011
AI x Bio Discovery @aixbiobot.bsky.social · 07/09/2026
Learning from tandem mass spectra at scale with a self-supervised foundation model for proteomics [new] by reconstructing masked regions to create universal peptide fragmentation embeddings for diverse, label-free applications.
Figure 1Figure 2Figure 3Figure 4
032
AI x Bio Discovery @aixbiobot.bsky.social · 07/09/2026
Predicting Endometriosis Status and Menstrual Cycle Phase Using DNA Methylation [new] ...reveals distinct genomic endometrial methylation patterns that classify disease and cycle phase, emphasizing cycle-related epigenetic variation.
Abstract 1Table 1Figure 1Figure 2
000
AI x Bio Discovery @aixbiobot.bsky.social · 07/09/2026
Benchmark validity in graph neural network scoring of metabolic reaction activity on Recon3D: detecting label leakage, memorized noise and input-invariant models [new]
Figure 1Figure 2Figure 3Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 07/09/2026
PharmCast: rapid generation of three-dimensional pharmacophore fingerprints from two-dimensional structure without conformer generation [new]
Figure 1Figure 2Figure 3Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 07/09/2026
Rapid robust high-fidelity 3D neuronal extraction from multiview calcium imaging datasets [new] This pipeline processes volumetric calcium imaging from diverse one/two-photon microscopy modalities.
screenshot
000
AI x Bio Discovery @aixbiobot.bsky.social · 07/09/2026
ReTIF: Granularity-Aware Multitask Interaction Routing for RNA-Compound Interaction Prediction and Binding-Site Localization [new] Gen. distinct int. reps for DTI/BS, maint. task granularity via dist. aggr. & prior-guid. prop.
ReTIF: Granularity-Aware Multitask Interaction Routing for RNA-Compound Interaction Prediction and Binding-Site LocalizationFigure 1Figure 2Figure 3
000
AI x Bio Discovery @aixbiobot.bsky.social · 07/09/2026
PromptBio: An Agentic Platform for End-to-End Computational Biomedical Research [new] ...translates natural language research questions into adaptive, traceable computational workflows for complex biomedical analysis.
Figure 1Figure 2Figure 3Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 07/09/2026
OmniSyn unifies target-aware molecular generation and optimization within a synthesis-native LLM framework across the human proteome [new]
Figure 1Figure 2Figure 3Figure 4
010
AI x Bio Discovery @aixbiobot.bsky.social · 07/09/2026
A Generalizable Feature Extractor for Alzheimer's-Related Brain MRI Tasks [new] ...is a compact brain-age pretrained model that adapts to diverse AD-related tasks with minimal new parameters, transferring to unseen cohorts without retraining.
Figure 1Figure 2.1Figure 2.2Figure 2.4
000
AI x Bio Discovery @aixbiobot.bsky.social · 07/09/2026
Simulation-free Unbalanced Dynamic Optimal Transport with General Growth Penalty [new] extends UDOT for cellular dynamics by allowing general convex growth penalties without relying on simulations.
Simulation-free Unbalanced Dynamic Optimal Transport with General Growth PenaltyFigure 1Figure 2Figure 3
000
AI x Bio Discovery @aixbiobot.bsky.social · 06/09/2026
High-throughput genomic feature extraction reveals environmental adaptations of prokaryotes [new] ...by using ML on 13k+ genomes to link genes/ncRNAs with salinity, temp, oxygen, and pH preferences.
High-throughput genomic feature extraction reveals environmental adaptations of prokaryotesFigure 1Figure 2Figure 3
000
AI x Bio Discovery @aixbiobot.bsky.social · 06/09/2026
PanScreen: A Comprehensive Approach to Off-Target Liability Assessment [updated] ...automating early safety evaluation via structure-based modeling & deep learning to predict binding affinities & modes of action.
Figure 1Figure 3Figure 2Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 06/09/2026
Unlocking Sensitive Data with SPHERE in the Age of AI [new] by generating synthetic twins for AI & open science, preserving original data privacy and scientific utility for crucial discoveries.
Figure 1Figure 2Figure 3Figure 4
011
AI x Bio Discovery @aixbiobot.bsky.social · 05/09/2026
Relational Graph Convolutional Networks for Glioblastoma Biomarker Discovery via ceRNA and Copy Number Variation Analysis [updated] combine ceRNA and CNV analyses via a late-fusion ensemble, IDing five novel glio biomarker candidates.
Figure 1Figure 2Figure 3Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 05/09/2026
Forecasting viral evolution from phylogenetic trees [new] by a machine learning model that learns past evolutionary paths from trees to anticipate future viral mutations across diverse viruses.
Figure 1Figure 2Figure 3Figure 4
010
AI x Bio Discovery @aixbiobot.bsky.social · 05/09/2026
Turning Domain Expertise into Multi-Dimensional Evaluation of Biomedical AI with Karenina [new]
Figure 1Figure 2Figure Supplementary Figure 1Figure Supplementary Figure 2
000
AI x Bio Discovery @aixbiobot.bsky.social · 05/09/2026
Ensemble tests mask missing dynamics in protein conformational generators [new] ...by showing they retain ensemble fidelity despite destroyed dynamics, thus requiring time-resolved validation to confirm actual learned dynamics.
Ensemble tests mask missing dynamics in protein conformational generatorsFigure 1Figure 2Figure 3
000
AI x Bio Discovery @aixbiobot.bsky.social · 05/09/2026
PredIDR3: A new output-encoding scheme and abundant negative source provide more information for deep learning-based protein intrinsic disorder prediction [new]
PredIDR3: A new output-encoding scheme and abundant negative source provide more information for deep learning-based protein intrinsic disorder predictionFigure 1Figure 2Figure 3
010
AI x Bio Discovery @aixbiobot.bsky.social · 05/09/2026
BAGEL-CAR: Reflections on the Bits to Binders Competition [new] Successful AI CAR binder design lessons: critically re-evaluate common filtering steps, assay intermediate outputs, and consider target geometry.
BAGEL-CAR: Reflections on the Bits to Binders CompetitionFigure 1Figure 2Figure 3
000
AI x Bio Discovery @aixbiobot.bsky.social · 05/09/2026
Cross-domain confidence reliability and remappability of frozen single-cell representations [new] is unreliable via raw estimates. Errors/drift cause failures, necessitating target-specific recalibration with local labels.
Cross-domain confidence reliability and remappability of frozen single-cell representationsFigure 1Figure 2Figure 3
000
AI x Bio Discovery @aixbiobot.bsky.social · 05/09/2026
Evaluating performance bias in face-to-BMI vision transformer models across diverse human populations [new] Diversity in training data is essential for developing machine learning health tools to generalize across human populations.
Figure 1Figure 2Figure 3Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 05/09/2026
When DL-Based Prescreening Meets Synthon-Based Docking: Target-Adapting PharmacoNet via MEL-Steered Correction [new] adapts prescreen method using MEL frag docking to steer hotspots and fine-tune int. weights for target spec.
Figure 1Figure 2Figure 3Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
siProGenA: Generative siRNA Candidate Construction via Position Proposal and Guide Generation [new] ...by decomposing siRNA design into mRNA-conditioned position proposal and constrained guide variant generation.
siProGenA: Generative siRNA Candidate Construction via Position Proposal and Guide GenerationFigure 1Figure 3Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
Gene duplication shaped the origin and evolution of the vertebrate olfactory combinatorial code [new] ...initiating code at gnathostome anc.'s 1st receptor dupl., driving asymmetric chemical spec. & persistent combin. coding.
Table 1Figure 1Figure 2Figure 3
000
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
AltraFlowSOM: A Semi-Supervised Framework for Imaging Mass Cytometry Phenotyping [new] ...embeds expert labels into SOM training, blending guided topology & unsupervised discovery to overcome IMC phenotyping bottlenecks.
Figure 1Figure 2Figure 3Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
Modelling interpretable patient-level representationsfrom structured and simple multimodal data [new] jointly modeling structured and simple data, linking pt-level factors to sub-obs clusters and proportions for direct interpret.
Modelling interpretable patient-level representationsfrom structured and simple multimodal dataFigure 1Figure 2Figure 3
000
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
ContrasTED: contrastive domain embeddings for scalable remote homology classification [new] using CATH-sup. center-contrastive learning to map structure-aware embeddings for nearest-centroid superfamily assign., extending annotation.
ContrasTED: contrastive domain embeddings for scalable remote homology classificationFigure 1Figure 2Figure 3
000
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
Retrieval of binding sites across the AlphaFold human proteome using protein language model representations [new] ...by using frozen ESM-C 600M embeddings of cavity-lining residues for exhaustive MaxSim retrieval across the proteome.
Retrieval of binding sites across the AlphaFold human proteome using protein language model representationsFigure 2Figure 3Figure 5
010
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction [new] ...assesses PLM impact on DTA, comparing them to modified 1D convolution for improved protein representation.
Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity PredictionFigure 1Figure 2Figure 3
010
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
Ageas enables time-agnostic cell fate inference from single-cell and spatial multi-omics data [new] ...by learning fate memory from terminal cells to infer fate bias in static progenitors, uncovering spatially organized fate priming.
Figure 1Figure 2Figure 3Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
Hi-cGAN: Prediction of Hi-C interaction matrices with conditional generative adversarial networks [new] by computationally predicting genome cont. structures using chromatin factor occupancy, offering alternative to wet-lab methods.
Hi-cGAN: Prediction of Hi-C interaction matrices with conditional generative adversarial networksFigure 1Figure 2Figure 3
000
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
scRep: A Latent-Space Self-Distilled Foundation Model for Single-Cell Representation Learning [new] learns stable bio reps aligning pert cell views in latent space via self-distillation, avoids raw expr reconstr.
Figure 1Figure 2Figure 3Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
De novo design of ligand binding proteins using large language models alone [new] ...leverages capacity to generate sequences for desired folds and bind targets, while explaining design principles, enhancing accessibility.
De novo design of ligand binding proteins using large language models aloneFigure 1Figure 2Figure 3
010
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
The Identification of Biological Stains at Crime Scenes: A Promising Role for Proteomics and Machine Learning [new] ...uses three proteomic methods to identify multiple body fluids and their complex mixtures via specific peptide analysis and machine learning.
The Identification of Biological Stains at Crime Scenes: A Promising Role for Proteomics and Machine LearningFigure 1Figure 2Figure 3
000
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
Multi-parametric NIR-II fluorescence perfusion imaging towards quantitative stroke evaluation [new] uses an analytic frame w/ deep learning processes images, generating multi-par. data for precise, anat.-aligned stroke severity map.
Figure 1Figure 2Figure 3Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
SurfSpec: Enhancing Off-Target-Agnostic Specificity by Bounding Pocket-Ligand Geometric Mismatch [new] ...it establishes a conservative specificity lower bound for unknown off-targets by analyzing target-ligand geometry, guiding iterative ligand growth.
SurfSpec: Enhancing Off-Target-Agnostic Specificity by Bounding Pocket-Ligand Geometric MismatchFigure 1Figure 2Figure 3
000
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
Sparse concept attribution for histomorphological hypothesis generation from whole-slide classifiers [new] ...automates slide interp. by sparsely attributing image features to a generalist concept bank, enabling scaled hypothesis generation for expert val.
Sparse concept attribution for histomorphological hypothesis generation from whole-slide classifiersFigure 1Figure 3Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
An Integrative Computational Approach to Predict Viral Epitopes by Targeting the MHC-TCR Complexation [new] unravels T-cell activation and identifies viral epitopes for vaccine development through integrative computational analysis of MHC-TCR complexation.
Figure Scheme 1Figure 1Figure 2Figure 3
000
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
SimpleDesign: A Joint Model for Protein Sequence and Structure Codesign [new] unifies sequence and structure generation by training directly in data space with a single-stage objective and a Mixture-of-Transformer.
Figure 1Figure 2Figure 3Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
Neural-Network Maxent: a general extension with learned nonlinearity, applied to time-series for Desert Locust distribution modelling [new] replacing fixed features with NNet to capture nonlinear temp. relations in env. TS for Desert Locust habitat modeling.
Figure 1Figure 2Figure 3Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 04/09/2026
Joint ancestry inference reveals the landscape of archaic introgression in admixed populations [new] by simultaneous continental and archaic ancestry inference, mapping archaic segs in continental blocks, tracing ancestral source.
Figure 1Figure 2Figure 3Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 03/09/2026
Multimodal Protein Retrieval via Joint Representation Learning from Sequences and Cryo-EM Density Maps [new] connects protein seqs & cryo-EM maps via shared latent space for bidir. retrieval & biologically meaningful struct. repr.
Multimodal Protein Retrieval via Joint Representation Learning from Sequences and Cryo-EM Density MapsFigure 1Figure 3Figure 5
000
AI x Bio Discovery @aixbiobot.bsky.social · 03/09/2026
Towards Sparse Causal Features for Zero-shot Mutation Effect Prediction in a Protein Language Model [new] Finds latent features causally mediating pLM preds, revealing bio meaning & reuse in mutations, often linked to 3D contacts.
Towards Sparse Causal Features for Zero-shot Mutation Effect Prediction in a Protein Language ModelFigure 1Figure 2Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 03/09/2026
Morphologic intratumoral heterogeneity from routine whole-slide histopathology is prognostic for survival in primary central nervous system lymphoma: development in the LOC Network and international e... [new]
Figure 1Figure 2Figure 3Figure 4
000
AI x Bio Discovery @aixbiobot.bsky.social · 03/09/2026
DAG-HEART: Directed Acyclic Graph-Guided Health Equity-Aware Representation Transfer Learning Framework for Breast Cancer [new] integrates directed multi-omics with TL & DA for eq. breast cancer outcome pred in D-minority groups.
DAG-HEART: Directed Acyclic Graph-Guided Health Equity-Aware Representation Transfer Learning Framework for Breast CancerFigure 1Figure 2Figure 3
000
AI x Bio Discovery @aixbiobot.bsky.social · 03/09/2026
BioByte 169 [WE’RE BACK]: And In the Darkness Bind Them [blog] ...with an AI autonomously designing protein binders end-to-end from target selection to final candidates.
000
AI x Bio Discovery @aixbiobot.bsky.social · 03/09/2026
From neuropeptide and receptor annotation to ligand-receptor pairing: a sequence- and structure-based framework for mapping the neuropeptide-receptor interactome in Gryllus bimaculatus [new]
From neuropeptide and receptor annotation to ligand-receptor pairing: a sequence- and structure-based framework for mapping the neuropeptide-receptor interactome in Gryllus bimaculatusFigure 1Figure 2Figure 3
000
AI x Bio Discovery @aixbiobot.bsky.social · 03/09/2026
Beyond benchmark accuracy: machine-learning turnover-number predictors require system-level validation [new] As benchmark results don't predict downstream model behavior, where localized errors can misrepresent bio system limits.
Figure 1Figure 2Figure 4Figure 5
000
AI x Bio Discovery @aixbiobot.bsky.social · 03/09/2026
Macrophage signature-based prediction of cancer treatment response using MIL-attention [new] ...predicts immunotherapy response from macrophage embeddings, leveraging attention pooling and custom loss to interpret immune states.
Macrophage signature-based prediction of cancer treatment response using MIL-attentionFigure 1Figure 2Figure 3
000