Pre-processing of the untargeted metabolomics data from CHRIS

This repository contains all data analysis files for the pre-processing and normalization of the (HILIC-based) untargeted metabolomics data from CHRIS.

Contact persons

• Marilyn De Graeve
• Johannes Rainer

Previous contributors

• Vinicius Verri Hernandes
• Mar Garcia-Aloy

Files and structure of the analysis

The MS data is provided as a self-contained SQL database that also contains sample annotations (batch IDs etc). The data can be loaded as a MsExperiment object through the MsBackendSql package.

Analysis scripts are supposed to be run on the calculation cluster using the respective shell script (e.g. peak_detection.sh for peak_detection.Rmd). The script should then be executed with sbatch --mem-per-cpu=24000 -c 10 --partition=batch ./peak_detection.sh.

General data overview

• general_data_overview.Rmd: general overview and summary statistics as well as initial quality assessment of the whole data set.

Preprocessing

-- NEEDS TO BE UPDATED AND FIXED --

• peak_detection.Rmd: peak detection and peak post-processing of both positive and negative polarity data.
• peak_detection_qa.Rmd: quality assessment and summaries for the peak detection step.

TODO:

• evaluate alignment on QC samples:
  • check EICs before/after: are we able to alleviate the retention time shifts caused by the LC maintenance in November 2021?

Normalization

• normalization.Rmd: implementation of various data normalisation approaches (between-sample, within-batch and between-batch) for 1 specific polarity.

Required packages and setup

The analysis requires a recent version of R (version >= 3.6.0) and a set of R packages that can be installed with the code below.

install.packages("BiocManager")
library(BiocManager)
BiocManager::install(c("BiocStyle",
                       "xcms",
                       "RColorBrewer",
                       "pander",
                       "UpSetR",
                       "pheatmap",
                       "SummarizedExperiment",
                       "MsExperiment",
                       "Spectra",
                       "MetaboCoreUtils",
                       "MsBackendSql",
                       "MsQuality",
                       "writexl",
                       "ProtGenerics",
                       "MsCoreUtils",
                       "MetaboAnnotation"))

install.packages(c("tidyverse",
                   "readxl",
                   "readr",
                   "RSQLite",
                   "rmarkdown",
                   "kableExtra",
                   "magick",
                   "vioplot",
                   "pandoc",
                   "pander",
                   "car"))

                 
                 

Raw mzML data

mzML files for this data set are accessible from the calculation clusters at the base path /data/massspec/mzML/.

MS data was recorded in profile mode. Sciex wiff files were converted to mzML file format using proteowizard. The profile mode mzML files were centroided in R to generate the centroided mzML files. The scripts to perform the wiff-to-mzML conversion and centroiding can be found here.