Comprehensive Lipidomics Automation Workflow for Multiple Reaction Monitoring Using Large Language Models
- Automates lipid profiling—from parsing raw binary (.d) files into organized pandas DataFrames to annotation using custom MRM transition databases and robust statistical analysis.
- Applies trimmed mean of M-values (TMM) normalization for lipid load correction and generalized linear models (GLMs) in edgeR for overdispersed count data analysis, providing a robust statistical framework for lipidomics.
- Integrates LIGER (lipidome gene enrichment reactions) to connect lipid expression with gene activation and suppression patterns, highlighting biologically relevant pathways.
- Incorporates a language user interface (LUI) with interactive artificially intelligent (AI) agents, providing a conversational way to perform statistical and bioinformatics workflows.
- Applied this pipeline to Alzheimer’s disease mouse models, profiling nearly 1,500 MRM transitions across 11 lipid classes, and revealing metabolic pathways enriched in differentially expressed lipids.
Paper · Code