iSLS10 Workshop Lipidomics Data Analysis

Sphingolipids and T2DM Risk

Summary

This repository contains datasets and R codes used for 9th International Singapore Lipid Symposium iSLS10 workshop held Wednesday, March 9, 2022. We will be using the data published in Chew et al., 2019, JCI Insight 5(13) 10.1172/jci.insight.126925 as an example dataset for this workshop.

In the first part, we will import and clean a lipidomics dataset and de-identified metadata into R, look into quality control aspects, convert lipid names to current nomenclature, and re-structure the data for subsequent statistical analysis covered in the second part. We will also explore how to efficiently conduct descriptive statistics annd make plots for each analyte in the dataset using R.

In the second part, we will inspect the overall data trends in both sample meta data and lipidomics data via visualization. We will study the correlation structure between sphingolipids and other competing risk factors collected at baseline, and test associations of individual lipids with the risk of DM incidence, using logistic regression (binary outcome) and Cox regression analysis (time-to-event analysis). We will also explore options for multi-variable modelling, such as multi-variable Cox regression with group structured LASSO penalty and popular machine learning methods (e.g. random forest or boosting).

Download the R scripts

Download the R Project containing the scripts and data used in this workshop from this repository. Alternatively you can clone this repository in RStudio/git.

Preparations to run the scripts

Following R packages and external software should be installed on your R system before running the scripts (see also comments in the R scripts of Part 1 and 2):

Part 1:
- CRAN packages: here, tidyverse, ggrepel
- Optional: rgoslin (used to convert lipid nomenclatures). Only available via github (https://github.com/lifs-tools/rgoslin). Note that on Windows, you will need an installation of rtools, and on macOS of XCode in order to install rgoslin.
Part 2:
- CRAN packages: scales, gplots, survival, glmnet, grpreg, randomForest, huge
- Bioconductor packages: mixOmics, qvalue , impute
- Cytoscape (https://cytoscape.org/download.html)

Authors

Bo Burla - Singapore Lipidomics Incubator @ National University of Singapore
Hyungwon Choi - Computational & Statistical Systems Biology Laboratory @ National University of Singapore

License

This project is licensed under the MIT License - see the LICENSE.md file for details

Acknowledgments

Wee Siong Chew and Shanshan Ji

SLINGhub / iSLS10