bio-workflows-metabolomics-pipeline

name	bio-workflows-metabolomics-pipeline
description	End-to-end metabolomics workflow from raw MS data to pathway analysis. Orchestrates XCMS preprocessing, annotation, normalization, statistical analysis, and pathway mapping. Use when processing LC-MS metabolomics data.
tool_type	r
primary_tool	XCMS
workflow	true
depends_on	["metabolomics/xcms-preprocessing","metabolomics/metabolite-annotation","metabolomics/normalization-qc","metabolomics/statistical-analysis","metabolomics/pathway-mapping","metabolomics/lipidomics","metabolomics/targeted-analysis","metabolomics/msdial-preprocessing"]

Metabolomics Pipeline

Pipeline Overview

Raw MS Data (mzML/mzXML) ──> Peak Detection ──> Feature Matrix
                                                     │
                                                     ▼
                   ┌─────────────────────────────────────────────┐
                   │          metabolomics-pipeline              │
                   ├─────────────────────────────────────────────┤
                   │  1. Peak Detection (XCMS)                   │
                   │  2. Retention Time Alignment                │
                   │  3. Feature Grouping & Gap Filling          │
                   │  4. QC & Normalization                      │
                   │  5. Statistical Analysis                    │
                   │  6. Metabolite Annotation                   │
                   │  7. Pathway Mapping                         │
                   └─────────────────────────────────────────────┘
                                                     │
                                                     ▼
                       Differential Metabolites + Enriched Pathways

Complete R Workflow

library(xcms)
library(MSnbase)
library(MetaboAnalystR)
library(ggplot2)

# === 1. LOAD DATA ===
mzml_files <- list.files('data/', pattern = '\\.mzML$', full.names = TRUE)
sample_data <- read.csv('sample_metadata.csv')

raw_data <- readMSData(mzml_files, mode = 'onDisk')

# Add sample metadata
pData(raw_data) <- sample_data

cat('Loaded', length(mzml_files), 'samples\n')

# === 2. PEAK DETECTION ===
cwp <- CentWaveParam(
    peakwidth = c(5, 30),
    ppm = 25
    snthresh  
    prefilter   
    mzdiff  
    noise  


xdata  findChromPeaksraw_data param  cwp
cat nrowchromPeaksxdata 


xdata  adjustRtimexdata param  ObiwarpParambinSize  
cat


pdp  PeakDensityParam
    sampleGroups  pDataxdatacondition
    minFraction  
    bw  
    binSize  


xdata  groupChromPeaksxdata param  pdp
cat nrowfeatureDefinitionsxdata 


xdata  fillChromPeaksxdata param  ChromPeakAreaParam


feature_matrix  featureValuesxdata value   method  
feature_info  featureDefinitionsxdata



feature_matrixfeature_matrix    
log_matrix  log2feature_matrix


valid_features  rowSumslog_matrix  ncollog_matrix  
filtered_matrix  log_matrixvalid_features 
cat nrowfiltered_matrix 


sample_medians  applyfiltered_matrix  median na.rm  
global_median  mediansample_medians
normalized  sweepfiltered_matrix  sample_medians  global_median



pca  prcomptnormalized scale.  
pca_df  data.framePC1  pcax  PC2  pcax 
                     Sample  rownamespcax
                     Condition  pDataxdatacondition

ggplotpca_df aesPC1 PC2 color  Condition 
    geom_pointsize   
    theme_bw 
    labstitle  
ggsave width   height  


librarylimma

design  model.matrix   condition data  pDataxdata
colnamesdesign  levelsfactorpDataxdatacondition


imputed  normalized
imputedimputed  imputed na.rm    

fit  lmFitimputed design
contrast  makeContrastsTreatment  Control levels  design
fit2  contrasts.fitfit contrast
fit2  eBayesfit2

results  topTablefit2 number   adjust.method  
resultsfeature_id  rownamesresults
resultssignificant  resultslogFC    resultsadj.P.Val  

cat resultssignificant 



resultsmz  feature_inforesultsfeature_id 
resultsrt  feature_inforesultsfeature_id 


libraryKEGGREST

annotate_mz  mz ppm   
    
    
    mz_range  mz    ppm mz    ppm
      



ggplotresults aesx  logFC y  log10adj.P.Val color  significant 
    geom_pointalpha   
    geom_hlineyintercept  log10 linetype   
    geom_vlinexintercept    linetype   
    scale_color_manualvalues    
    theme_bw 
    labstitle   x   y  
ggsave width   height  


write.csvresults  row.names  
write.csvnormalized 
cat

MetaboAnalystR Pathway Analysis

library(MetaboAnalystR)

# Initialize
mSet <- InitDataObjects('conc', 'pathora', FALSE)

# Load compound list (HMDB IDs)
sig_features <- results[results$significant, ]
compound_list <- sig_features$hmdb_id  # Requires annotation

mSet <- Setup.MapData(mSet, compound_list)
mSet <- CrossReferencing(mSet, 'hmdb')
mSet <- CreateMappingResultTable(mSet)

# Pathway analysis
mSet <- SetKEGG.PathLib(mSet, 'hsa')
mSet <- SetMetabolomeFilter(mSet, FALSE)
mSet <- CalculateOraScore(mSet, 'rbc', 'hyperg')

# View results
pathway_results <- mSet$analSetora.mat
headpathway_results


mSet  PlotPathSummarymSet     

Alternative: MS-DIAL Preprocessing

# Load MS-DIAL exported data
msdial_export <- read.csv('msdial_alignment.csv')

# MS-DIAL already provides:
# - Peak detection
# - Alignment
# - Gap filling
# - Annotation attempts

# Continue with normalization and statistics
feature_matrix <- as.matrix(msdial_export[, grep('Area', colnames(msdial_export))])
rownames(feature_matrix) <- msdial_export$`Alignment.ID`

# Proceed with normalization and limma as above

QC Checkpoints

Stage	Check	Action if Failed
Peak detection	>1000 features	Adjust parameters
Alignment	RT deviation <30s	Check QC samples
Grouping	>60% features grouped	Adjust bw/minFraction
Missing values	<30% per sample	Check injection
QC RSD	<30% for QC features	Check instrument
PCA	Groups separate	Check batch effects

Workflow Variants

Lipidomics

# Adjust peak width for lipids
cwp_lipid <- CentWaveParam(
    peakwidth = c(10, 60),  # Broader peaks
    ppm = 15,
    snthresh = 5
)

# Use LipidMaps for annotation

Targeted Analysis

# Define target compounds
targets <- data.frame(
    name = c('Glucose', 'Lactate', 'Citrate'),
    mz = c(179.0561, 89.0244, 191.0197),
    rt = c(120, 90, 180)
)

# Extract targeted features
extractTargets <- function(xdata, targets, mz_ppm = 10, rt_tol = 30) {
    lapply(1:nrow(targets), function(i) {
        chromPeaks(xdata, mz  targetsmzi ppm  mz_ppm
                   rt  targetsrti  rt_tol targetsrti  rt_tol
    

Related Skills

• metabolomics/xcms-preprocessing - XCMS parameters
• metabolomics/metabolite-annotation - Compound identification
• metabolomics/normalization-qc - QC and normalization methods
• metabolomics/statistical-analysis - Statistical testing
• metabolomics/pathway-mapping - KEGG/MetaboAnalyst
• metabolomics/lipidomics - Lipid-specific analysis
• metabolomics/targeted-analysis - Absolute quantification
• metabolomics/msdial-preprocessing - MS-DIAL export processing
• multi-omics-integration/mofa-integration - Integrate with other omics