Metabolomic and lipidomic signatures in autosomal dominant and late-onset Alzheimer disease brains

Scripts used to QC and analyze metabolomics data for Novotny et al. (2021) Metabolomic and lipidomic signatures in autosomal dominant and late-onset Alzheimer disease brains

---

Scripts

Knight ADRC dataset

01-QC_missingness.R (Knight ADRC Data)

• Check technical replicate correlation and average
• Remove metabolites missing >20% readings
• Recover metabolites with high missingness whose missingness between groups corresponds with differential abundance between groups
• Impute missing values in recovered metabolites with minimum
• Check missingness by individual

02-QC_IQR.R

• Log-transform data, remove outliers (>1.5xIQR), adjust mean

03-QC_PCA.R

• Average replicates in the scaled/imputed data provided by Metabolon
• Calculate PCA and identify outlier samples
• Remove outlier samples

04-main_analysis.R

• Main analysis including linear regressions for each status group (also matched by CDR, BraakTau, and BraakAbeta), association with age at death in sAD and CO, and association with APOE genotype

05-effect_comparison.R

• Compare effects of 16 common metabolites between ADAD, TREM2, and sAD. ANCOVA tests with and without matching individuals by CDR, BraakTau, and BraakAbeta

06-eigengene_analysis.R

• Calculate eigengene (PC1) for 16 metabolites in CO, sAD, ADAD, and TREM2 participants
• Test association of eigengene with phenotypic variables

07-heatmap_analysis.R

• Create heatmaps representing the 16 metabolite-profile and test association of early-stage AD (ESAD) group with phenotypic variables

08-pharmacological_analysis.R

• Check for effects of fluoxetine on beta-citrylglutamate associations and vitamin supplements on vitamin associations

ROSMAP and ADNI p180 datasets

09.1-ROSMAP_p180_analysis.R

• Full cleaning and analysis of ROSMAP serum samples

09.2-ROSMAP_p180_analysis.R

• Full cleaning and analysis of ROSMAP DLPFC samples (p180 only)

10-ADNI1_analysis.R

• Full cleaning and analysis of ADNI1 serum samples

11-ADNI2GO_analysis.R

• Full cleaning and analysis of ADNIGO/2 serum samples

12-serum_meta-analysis.R

• Meta-analysis of ROSMAP, ADNI1, and ADNI2/GO serum data

ROSMAP Metabolon dataset

13-ROSMAP_Metabolon_clean_phenotypic_data.R

• Create phenotype data frame to be used with ROSMAP Metabolon data

14.1-ROSMAP_Metabolon_QC_missingness_cogdx_recovery.R

• Recovery of metabolites using consensus clinical diagnosis for ROSMAP Metabolon data, same process as 01-QC_missingness.R

14.2-ROSMAP_Metabolon_QC_missingness.R

• Missingness QC for ROSMAP Metabolon data, same process as 01-QC_missingness.R

15-ROSMAP_Metabolon_QC_PCA.R

• Calculate and visualize PCA for individuals; no outliers removed

16-ROSMAP_Metabolon_main_analysis.R

• Differential abundance analysis for ROSMAP Metabolon data (linear regressions)

17-ROSMAP_Metabolon_eigengene_analysis.R

• Calculate eigengene for the 15 available metbolites in ROSMAP Metabolon data, check association with disease duration and Braak tau

18-EOAD_vs_LOAD.R

• Stratify sAD into EOAD (AAO<65 years) and LOAD (AAO>=65 years), check differential abundance and phenotypic variables

---

Metabolomics Browser

The browser can be accessed at http://ngi.pub/Metabolomics/.

app.R

• Main app file, load data and call modules

Modules Folder

boxplots.R

• Module for "Reading Distributions" tab, display results table, boxplots, and specific metabolite information

home.R

• Homepage, display phenotypic summary information

volcano.R

• Module for "Volcano Plots" tab, plot DE results, show boxplots, and specific metabolite information

Contact

Brenna Novotny: [email protected]