FGF21 regulates sex-specific metabolic adaptation to juvenile protein restriction

# FGF21 Liver-Specific Knockout Study This repository contains all data, analysis scripts, and figures used in the manuscript: **"FGF21 regulates sex-specific metabolic adaptation to juvenile protein restriction"** ## Contents FGF21_study/├── data/│ ├── raw/ # Raw input data files (e.g., body weight, qPCR, puberty onset)│ │ └── transcriptomics_counts/ # TSV files with raw counts of liver transcriptomics│ ├── processed/ # Output files from each script (e.g., Excel summary tables)│ ├── external/ # Data from an external source (e.g., Green et al., Cell Metab, 2022)├── metadata/│ ├── diet_composition.xlsx # Decription of the experimental diets│ ├── qpcr_primers.xlsw # Primer sequences for qPCR│ ├── statistical_models_overview.xlsx # Summary table of statistical models used to analyze the data├── scripts/│ ├── 01_...R # Numbered analysis scripts │ └── ...├── plots/│ ├── BPS_by_Weight_...png # Final figures (PNG format)│ └── ...├── FGF21_study.Rproj # RStudio project file└── README.md # This file ## How to reproduce All scripts are written in R and use a modular, FAIR-compliant pipeline: 1. Install required R packages: - `tidyverse`, `car`, `lme4`, `emmeans`, `DHARMa`, `openxlsx`, `broom`, etc.2. Open the project in RStudio via `FGF21_study.Rproj`.3. Run each script in order (see numbered filenames). Each script:- Declares root via `here::i_am()`- Outputs results to `data/processed/`- Saves figures to `plots/` ## List of Scripts (51 Total) Scripts are ordered to match the flow of figures and tables in the manuscript. ├──01–20: Juvenile Growth and Reproduction01_length_lmm_spline.R02_body_weight_lmm_spline.R03_femur_lm.R04_igf1_lmm.R05_kcal_lmm_by_interval.R06_feed_efficiency_lmm_by_interval.R07_lean_mass_permutation.R08_fat_mass_parametric.R09_BPS_timing.R10_sperm_count.R11_seminal_vesicle_ANCOVA.R12_testis_weight_t_test.R13_VO_timing.R14_first_estrus.R15_estral_cycles_P56_to_P70.R16_uterus_weight_t_test.R17_VO_to_estrus.R18_logistic_regression_puberty_males.R19_logistic_regression_puberty_females.R20_growth_puberty_reciprocal_model_females.R├──21–24: Serum and Liver Amino Acids21_serum_amino_acids_postprandial.R22_serum_amino_acids_postabsorptive.R23_individual_AA_log_lmm_and_heatmap.R24_liver_individual_AA_lmm_and_heatmap.R├──25–31: Transcriptomics & Gene Expression25_rlog_vst_pca_sample_qc.R26_DESeq2_global_model.R27_GSEA_liver.R28_Volcano_plot_liver.R29_liver_heatmaps_gene_sets.R30_sPLSDA_liver_transcriptome_metabolome.R31_liver_qPCR_models.R├──32–39: Longitudinal Models & Strain Comparisons32_FGF21_lmm_longitudinal.R33_length_lmm_spline_FGF21_liverKO.R34_body_weight_lmm_spline_FGF21_liverKO.R35_body_weight_lmm_spline_B6J.R36_femur_lm_FGF21_liverKO.R37_BPS_timing_B6J.R38_VO_timing_B6J.R39_FGF21_B6J.R├──40–48: Liver KO Specific Models40_BPS_FGF21_liverKO.R41_sperm_count_FGF21_liverKO.R42_seminal_vesicle_ANCOVA_FGF21_liverKO.R43_testis_weight_lm_FGF21_liverKO.R44_VO_timing_FGF21_liverKO.R45_first_estrus_FGF21_liverKO.R46_uterus_weight_lm_FGF21_liverKO.R47_FGF21_AFT_FGF21_liverKO.R48_liver_qPCR_models_FGF21_liverKO.R├──49–51: Additional Models for Figure S449_IGF1_longitudinal_lmm_FGF21_liverKO.R50_logistic_regression_growth_puberty_males_FGF21_liverKO.R51_logistic_regression_puberty_females_FGF21_liverKO.R ## Model Diagnostics To determine whether transformations or non-parametric methods were necessary, we used a standalone helper script: `scripts/52_diagnostics_DHARMa_helper.R` This script defines a reusable function `check_model_diagnostics()` that performs the following steps: 1. Fits a linear (LM or LMM) model on the raw outcome.2. Runs residual diagnostics using the `DHARMa` package.3. If the model fails to meet assumptions: - Applies a log-transformation and re-runs diagnostics. - If the log-transformed model passes, the log-transformation is retained. - If both fail, the script recommends using non-parametric alternatives such as `ARTool` or `permuco`. This diagnostic pipeline was applied to each outcome variable prior to formal modeling, guiding decisions on:- When to apply log-transformations (e.g., for lean mass, amino acids)- When to use permutation-based ANOVA (e.g., if assumptions were violated even after transformation) **Note**: This diagnostic procedure is not embedded in the main scripts. It serves as documentation of the decision-making process applied to the raw data before modeling, ensuring statistical validity and transparency. ## Output Files Each script saves: - Model results as `.xlsx` files in `data/processed/`- Figures as `.png` files in `plots/`, named to match figure panels in the manuscript (e.g., `BPS_by_Weight_FGF21_liverKO.png`) ## Session Information A full list of all packages and their versions is provided in the `sessionInfo.txt` file at the root of the repository. ## Citation If you use this code or dataset, please cite: > Joly, A. et al. (2025) *Liver FGF21 mediates sex-specific adaptation to juvenile protein malnutrition.* bioRxiv. https://doi.org/10.1101/2025.07.09.663702 ## External Data This repository includes transcriptomic data from: > Green, C. L., et al. (2022). *Sex and genetic background define the metabolic, physiologic, and molecular response to protein restriction*. Cell Metabolism, 34(2), 209-226.e5. https://doi.org/10.1016/j.cmet.2021.12.018 Data used under the terms of the **Elsevier Open Access license (CC BY-NC-ND 4.0)**. ## License This dataset and code are shared under a [CC BY 4.0 License](https://creativecommons.org/licenses/by/4.0/).