Oxylipins Coffee Antioxidant

This repository includes a code and the database for statistical analyses of oxylipins in the foam cell model as supplementary material for the manuscript:

Oscar J. Lara-Guzmán, Sonia Medina, Rafael Álvarez, Camille Oger, Thierry Durand, Jean-Marie Galano, Natalia Zuluaga, Ángel Gil-Izquierdo and Katalina Muñoz-Durango. 2020. Oxylipin regulation by phenolic compounds from coffee beverage: positive outcomes from a randomized controlled trial in healthy adults and macrophage derived foam cells. The manuscript is currently submitted to Free Radical Biology & Medicine journal as a research article.

The file Oxylipins_CGAs_FoamCells.R is a commented R script that requires the provided database file Oxylipins_CGAs_FoamCells.csv to properly run.

RStudio and R software used: R i386 3.6.1

Summary step by step:

1. Select the database and set up:

• Encoding: automatic
• Heading: YES
• Row names: Use first column
• Separator: Semicolon
• Decimal: Period
• Quote: Double quote (")
• na.strings: NA

2. Load the required libraries: factoextra, utils, stats, FactoMineR, ggfortify, ggplot2, "corrplot", magrittr, cluster, ggpubr, NbClust, REdaS, DiscriMiner, corrr, tidyverse, ggraph, devtools, dplyr, igraph.
3. Attach the database.
4. Two data frames derived from the database are required for analysis:

• All_treatments, which contain all data, except categorical variables.
• Treatments, which contain all data, except observations for MIX treatment and positive control. This database is used for the correlation analysis.

5. Scale and normalize the databases All_treatments and Treatments to:

• All_Treatments_Scale. For KMO (Kaiser-Meyer-Olkin) test and PCAs.
• Treatments_Scale. For clustering and distance analysis.

6. Create correlations from treatments using the command “cor”.

• Use the command “corrplot” to generate graphics.
• Create the function cor.mtest and matrix of the p-value of the correlation. Execute the command “corrplot” for visualization.

7. Create a heatmap of distances measure or (dis)similarity between observations using the commands “get_dist” and “fviz_dist”.
8. Estimate the KMO value (Kaiser-Meyer-Olkin), this test measures how suited your dataset is for Factor Analysis.
9. For the PCA analysis is required a cluster analysis. “fviz_nbclust” and “fviz_cluster” functions are used to estimate the number of the clusters and visualization.
10. Calculate the eigenvalues with “get_eigenvalue” command and visualize this with “fviz_eig”.
11. Use “fviz_pca_ind” and “fviz_pca_var” to visualize the PCA score and loading plots.
12. Use “dimdesc” to estimate the correlations and P-value for each variable in PC1 and PC2.