large language model

Expanding plant trait databases using large language model: A case study on flower color extraction

This paper uses a large language model to extract flower color data from Flora descriptions, expanding the TRY Plant Trait Database and linking it with environmental data to analyze flower color-environment relationships.

Spike sorting AI agent

SpikeAgent automates spike sorting with a multimodal LLM-based AI agent, achieving curation consistency equal to or exceeding human experts and drastically reducing expertise barriers.

Beyond the Hype: The Complexity of Automated Cell Type Annotations with GPT-4

This paper evaluates GPT-4's performance in automated cell type annotation on single-cell RNA-seq data, identifies its limitations, and introduces an improved approach using RAG and new evaluation metrics.

N-Power AI: A Specialized Agent Framework for Automated Sample Size and Power Analysis in Clinical Trial Design

N-Power AI automates sample size and power analysis in clinical trials using an LLM-based framework, achieving perfect accuracy compared to statistical software.

Discovering Symbolic Cognitive Models from Human and Animal Behavior

LLM-driven evolutionary program synthesis (CogFunSearch) discovers interpretable Python cognitive models that predict bandit-task choices of humans, rats and fruit flies better than leading hand-designed models.

Towards a universal foundation model for biobank-scale human genome variation

This paper introduces SNPBag, a foundation model for analyzing human genome variation, achieving state-of-the-art accuracy in genotype imputation and efficient ancestry inference.

Human Genome Book:Words,Sentences and Paragraphs

The paper introduces a method to structure the human genome as a book by mapping DNA sequences to words, sentences, and paragraphs using transfer learning from English language models.

A Domain-Specific Language Model Approach for Identifying Immunogenic Epitopes from Publicly Available Data: EpitopeMiner

EpitopeMiner identifies immunogenic epitopes using a domain-specific LLM with RAG, improving retrieval of relevant literature and supporting neoantigen validation.

Zero-Shot, Big-Shot, Active-Shot - How to estimate cell confluence, lazily

This paper shows that using the Segment Anything Model (SAM) for zero-shot confluence estimation in mesenchymal stem cells is effective without needing extensive dataset curation or fine-tuning.

Simulating 500 million years of evolution with a language model

ESM3, a 98-billion-parameter multimodal protein language model tokenized across sequence, structure, and function, designs a bright green fluorescent protein at 58% sequence identity to known fluorescent proteins.

The role of chromatin state in intron retention: a case study in leveraging large scale deep learning models

The paper uses large-scale deep learning models to study intron retention, showing that chromatin state plays a key role in its regulation.

Improving functional protein generation via foundation model-derived latent space likelihood optimization

The paper improves functional protein generation by optimizing likelihood in both amino acid sequence and latent space, outperforming existing models in AMP and MDH generation.

Pipeline to explore information on genome editing using large language models and genome editing meta-database

The paper develops a method to extract genome editing information using large language models and applies it to prioritize genes for future research.

Scaling unlocks broader generation and deeper functional understanding of proteins

Scaling sparse Mixture-of-Experts protein language models up to 46B parameters on 1.5T curated tokens broadens viable protein generation across families and amplifies gains from laboratory preference alignment.

Extraction of biological terms using large language models enhances the usability of metadata in the BioSample database

This paper evaluates using large language models to extract biological terms from BioSample metadata, improving data findability and reusability.

Integrating MHC Class I visibility targets into the ProteinMPNN protein design process

The paper integrates MHC Class I visibility targets into the ProteinMPNN protein design process to reduce immune visibility while preserving structure.

Cancer vs. Conversational Artificial Intelligence

The paper uses LLMs to analyze oncological papers, generate code for knowledge graphs, and update a cancer signaling pathway with new research insights.

Metadata Harmonization from Biological Datasets with Language Models

This paper uses large language models to standardize metadata across biomedical datasets, achieving high accuracy in ontology alignment and reducing standardization labor.

Evaluation of search-enabled Pre-trained Large Language Models on retrieval tasks for the PubChem Database

This study evaluates GPT-4o's ability to retrieve data from PubChem using natural language and programmatic access, showing programmatic retrieval is more accurate.

Enhancing Gene Set Overrepresentation Analysis with Large Language Models

This paper introduces llm2geneset, a framework using large language models to generate gene sets from input genes and context, improving overrepresentation analysis compared to human-curated databases.

FuncFetch: An LLM-assisted workflow enables mining thousands of enzyme-substrate interactions from published manuscripts

FuncFetch uses an LLM to mine enzyme-substrate interactions from manuscripts, achieving high precision and recall in extracting functional data.

Generation of antigen-specific paired chain antibody sequences using large language models

This paper introduces MAGE, a model that generates paired human antibody sequences against specific targets, validated against SARS-CoV-2, H5N1, and RSV-A.

A Large Language Model Guides the Affinity Maturation of Variant Antibodies Generated by Combinatorial Optimization

This paper introduces Ab-Affinity, a large language model that generates high-affinity antibodies against SARS-CoV-2 spike protein peptides using combinatorial optimization and AI.

A Framework for Autonomous AI-Driven Drug Discovery

This paper introduces an AI-driven drug discovery framework that integrates knowledge graphs and large language models to automate and scale drug discovery processes.

Deep functional profiling of gene sets using Large Language Models: A blueprint for tailored, context-aware functional annotation

This paper introduces a method using Large Language Models to perform deep functional profiling of gene sets, demonstrating its ability to uncover detailed immune functions and improve upon traditional annotation tools.

Automating Candidate Gene Prioritization with Large Language Models: Development and Benchmarking of an API-Driven Workflow Leveraging GPT-4

This paper develops an API-driven workflow using GPT-4 to automate candidate gene prioritization, demonstrating high consistency and scalability in identifying sepsis-related genes.

A novel HLA Class II presentation prediction algorithm deciphers immunogenic CD4 epitopes specific to KRAS G12C

EDGE-II predicts HLA class II epitopes and identifies immunogenic CD4+ T-cell responses specific to KRAS G12C mutations in cancer.

DDI-GPT: Explainable Prediction of Drug-Drug Interactions using Large Language Models enhanced with Knowledge Graphs

DDI-GPT predicts drug-drug interactions using a large language model enhanced with knowledge graphs, achieving high accuracy and providing explainable insights into interaction mechanisms.

Scouter: Predicting Transcriptional Responses to Genetic Perturbations with LLM embeddings

Scouter predicts transcriptional responses to genetic perturbations using LLM embeddings, achieving higher accuracy than existing methods.

CASSIA: a multi-agent large language model for reference free, interpretable, and automated cell annotation of single-cell RNA-sequencing data

CASSIA automates and improves cell type annotation in single-cell RNA-sequencing data with interpretability and accuracy.