large language model

Autonomous multimodal agents enable transparent, spatiotemporal reconstruction of immune dynamics in pancreatic cancer progression

The paper introduces an agentic computational framework to reconstruct immune dynamics in pancreatic cancer using H&E histology, identifying three temporally ordered immune-stromal states in mouse models.

yakRNA Design: A semantic multimodal RNA composer

The paper introduces yakRNA Design, a language model that generates functional synthetic RNAs without 3D structural data, achieving efficient frame-shifting capabilities in ribosomes.

Leveraging large language models for metabolic engineering design

This paper develops a large language model-based method to extract metabolic engineering strategies from research articles and trains a hybrid model to predict engineering targets, outperforming traditional approaches.

An autonomous LLM agent platform for computational binder design and conjugation-aware prioritization of antibody drug conjugates

This paper introduces an autonomous LLM agent platform for computational binder design and ADC prioritization, achieving accurate classification across oncology targets with minimal human intervention.

A Context-Aware Target Engagement and Pharmacodynamic Biomarker Resource to Accelerate Drug Discovery and Development

The paper develops TargetTrace, a resource linking drug targets to pharmacodynamic biomarkers, compiling 71,323 target-biomarker relationships for oncology drug development.

TACO: TabPFN Augmented Causal Outcomes for Early Detection of Long COVID

TACO detects presymptomatic Long COVID by combining causal inference with a foundation model, achieving higher precision with fewer features and no hyperparameter tuning.

HARVEST: Unlocking the Dark Bioactivity Data of Pharmaceutical Patents via Agentic AI

HARVEST extracts structured bioactivity data from pharmaceutical patents using AI, producing millions of records and revealing gaps in existing databases.

PathwaySeeker: Evidence-Grounded AI Reasoning over Organism-Specific Metabolic Networks

PathwaySeeker uses AI to reason about organism-specific metabolic networks using proteomic and metabolomic data, distinguishing confirmed reactions from hypotheses.

ORION: An agentic reasoning construct for the analysis of complex human immune profiling

ORION uses AI to analyze immune profiling data, reducing analysis time from months to hours and generating testable hypotheses.

CROssBARv2: A Unified Computational Framework for Heterogeneous Biomedical Data Representation and LLM-Driven Exploration

CROssBARv2 integrates heterogeneous biomedical data into a knowledge graph with embeddings and LLM-driven tools for predictive modeling and question-answering.

LLM-autonomous development of deep learning models for quantitative microscopy

This paper presents a framework where a large language model autonomously develops deep learning models for microscopy tasks, achieving high accuracy across multiple modalities and problem types.

Negative design enables cell-free expression and folding of designed transmembrane β-Barrels

The paper demonstrates that negative design, which locally destabilizes β-strands, improves folding efficiency of synthetic transmembrane β-barrels in a cell-free system.

A Context-Aware Single-Cell Proteomics Analysis pipeline.

The paper introduces a pipeline for single-cell proteomics analysis that addresses challenges in data processing and annotation using context-aware methods and large language models.

Transcriptome-based cell type assignment for kidney cell culture models

The paper develops a transcriptome-based method to match bulk RNA-seq data from kidney cell lines to single-cell reference datasets, identifying accurate cell type identities and evaluating method performance.

Cirrina: LLM-driven pharmacological reasoning agent enables preclinical CNS drug evaluation

Cirrina, an LLM-driven pharmacological reasoning agent, improves preclinical CNS drug evaluation by integrating mechanistic tools and providing better decision accuracy than deterministic pipelines.

Co-designing sequence and structure of functional de novo enzymes with EnzyGen2

EnzyGen2 co-designs protein sequences and structures with ligand-guided functionality, outperforming existing methods in in silico benchmarks and generating active enzymes with high novelty.

EthoClaw: An Integrated AI Workflow Platform for Automated Analysis in Neuroethology

EthoClaw provides an AI-driven platform for automated behavioral analysis, integrating video processing, tracking, and reporting to enhance reproducibility and efficiency in neuroethology.

Decoding Immunomodulatory Hydrogels for Arthritis: Comparative Insights from Predictive Machine Learning and Large Language Models

The paper uses machine learning to analyze hydrogel formulations for arthritis, identifying key design factors and showing improved predictive accuracy over classical methods.

Integrating 730,947 exome sequences with clinical literature improves gene discovery

gnomAD v4 (730,947 exomes) brings LOFTEE-2 loss-of-function annotation (0.90 precision, strict mode), missense-aware constraint, and LLM-mined literature evidence combined with constraint to rank disease genes.

Leveraging AI-powered interactive playbacks to decipher rules of communication in zebra finches

This study uses an AI model to analyze zebra finch vocal interactions and identifies key features of their communication rules.

PlantScience.ai: An LLM-Powered Virtual Scientist for Plant Science

PlantScience.ai couples an LLM-built, continuously updated plant science knowledge graph (4.0 million entities, 10.8 million edges) with graph-based retrieval for citation-traceable plant biology answers.

Leveraging Large Language Models to Extract Prognostic Pathology Features in Ewing Sarcoma

This paper uses large language models to extract prognostic histologic features from Ewing sarcoma pathology reports, identifying NSE and S100 as significant biomarkers.

Large language models possess some ecologicalknowledge, but how much?

Benchmarking GPT-4o and Gemini 1.5 Pro across five closed-book ecological tasks shows solid coarse species presence and trait recall but poor fine-scale range mapping and threat classification.

CoNVict: An Agentic AI System for Copy Number Variation Prioritization in Rare Disease Diagnosis

CoNVict is an agentic AI system that prioritizes copy number variants in rare disease diagnosis by ranking candidates through structured reasoning, outperforming existing methods in identifying causal variants.

CellVoyager: AI CompBio Agent Generates New Insights by Autonomously Analyzing Biological Data

CellVoyager autonomously analyzes scRNA-seq data to generate new hypotheses and findings, outperforming existing models and producing scientifically interesting results.

Automated extraction and optimization of protein purification protocols using multi-agent large language models

This paper introduces a multi-agent LLM system to automate and optimize protein purification protocols, improving efficiency and accuracy in protocol creation.

Improving Causal Gene Identification Using Large Language Models

This study evaluates using Large Language Models for causal gene identification, showing improvements with Retrieval-Augmented Generation and genomic distance information, but with diminishing returns when combined.

NeuroNarrator: A Generalist EEG-to-Text Foundation Model for Clinical Interpretation via Spectro-Spatial Grounding and Temporal State-Space Reasoning

NeuroNarrator translates EEG segments into clinical narratives using spectro-spatial grounding and temporal state-space reasoning, enabling interpretable clinical interpretation.

Phenotypic reversion and target prioritization for cellular inflammation via representation learning with foundation models

The paper uses single-cell foundation models to identify genetic targets that reverse inflammation in endothelial cells by analyzing Perturb-seq data with and without proinflammatory stimulation.

Grounding olfactory perception in language: Benchmarks and models for generating natural language odor descriptions

ODIEU pairs human odor texts for 10,124 molecules with olfaction-tuned Sentence-BERT evaluators; CIRANO, a frozen GPT-2 with a molecular prefix, generates descriptions approaching human agreement.