Conserved Mechanisms of Plant Lipidome Remodeling under Heat and Cold Stresses Revealed through a Meta-Analysis

This repository contains R code for performing a comprehensive meta-analysis of plant lipidomic responses to temperature stress (heat and cold). The focus is on heterogeneity and sensitivity analysis, lipid class–specific and molecular species–level modeling. Very-long-chain fatty acids (VLCFAs) are also given special consideration.

Repository Structure

Main Features

Effect size calculation

• Loads data, followed by data cleaning
• Label the "heat" and "cold" treatments properly
• Calculates effect size for each observation

Heterogeneity Analysis

• Estimates of heterogeneity (τ²) and inconsistency (I²)

Sensitivity Analysis

• Leave-one-out influence diagnostics

Bayesian Hierarchical Modeling (Lipid class response and molecular species response)

• Conducted separately at each lipid class for "cold" and "heat" treatments
• Priors and diagnostics implemented using brms
• Split-platform analysis with platform being removed from the pooled model and fit per platform data (per stress)
• Cross-platform consistency check for split-platform results

VLCFA Analysis

• Special module to explore the role of "very-long-chain fatty acids (≥C22)"
• Distribution trends and treatment effects highlighted

AI (Claude and ChatGPT) has been used to help with codes. AI wasn't used for data extraction, analysis or generation of results.

---

Libraries

Install these R packages before running the scripts:

install.packages(c(
  "tidyverse", "brms", "metafor", "dplyr", "tibble", "readr", "ggplot2", "tidyr", "purrr", "stringr", "posterior"
))