HolobiomicsLab 
xcms-ramclustR-object-integration
Summary
Integrate processed LC–MS all-ion fragmentation chromatograms from xcms (feature detection and retention-time alignment) with RamClustR spectral clustering objects to enable downstream metabolite annotation. This skill bridges feature-level and spectrum-level representations needed for compound identification via fragment library matching.
When to use
You have centroid-mode LC–MS all-ion fragmentation (AIF) data already processed through xcms for feature detection and retention-time correction, and a corresponding RamClustR object that groups co-eluting fragment ions into putative spectral clusters. You need to annotate these clustered features against ion fragment databases (e.g., LipidPos) using tools like MetaboAnnotatoR's annotateRC function.
When NOT to use
- LC–MS data is in profile (non-centroid) mode — xcms and RamClustR require centroid mode input
- You only have raw chromatographic data without prior feature detection — use xcms independently first
- Target features are already annotated — integration is for de novo identification, not refinement
Inputs
- xcms xset object (processed LC–MS features with retention times)
- RamClustR RC object (spectral clusters and co-eluting ion assignments)
- target feature table (CSV format, e.g., targetTable.csv)
- ion fragment database library (e.g., LipidPos for positive mode)
Outputs
- annotations object containing ranked candidate metabolite matches
- rankedResult data frame with rank-1 and alternative annotations per feature
- optional: visualization and summary reports of annotation results
How to apply
Load the xcms xset object (containing processed features and retention times) and the RamClustR RC object (containing spectral clustering results and co-eluting ion assignments) into the same R environment. Prepare a target feature table (CSV format, e.g., targetTable.csv) listing the features to annotate, with columns matching the xset feature identifiers. Pass both objects together to the annotateRC function along with the target table and a fragment library (e.g., LipidPos for lipid-positive mode). The function will match experimental spectra against the database using the spectral information from the RC object and chromatographic alignment from the xset. Extract the ranked candidate annotations from the resulting annotations object, prioritizing rank-1 matches. Verify success by checking that annotations are returned and that matched ion pairs are consistent with the fragmentation patterns.
Related tools
- xcms (Feature detection, alignment, and retention-time correction of LC–MS chromatograms)
- RamClustR (Spectral clustering and co-eluting ion grouping for all-ion fragmentation data)
- MetaboAnnotatoR (Main annotation tool; executes annotateRC function to match xset and RC objects against fragment databases) — https://github.com/gggraca/MetaboAnnotatoR
- R (Runtime environment (version 4.5.0 or higher required))
Examples
# Load objects and annotate
library(MetaboAnnotatoR)
load('xset.RData'); load('RC.RData')
targets <- read.csv('targetTable.csv')
annotations <- annotateRC(xset, RC, targets, db='LipidPos')
head(annotations$rankedResult)
Evaluation signals
- Annotations object is successfully created and contains ranked candidate entries for input features
- At least one feature receives a rank-1 annotation (as in the reference study, 3 of 6 features were annotated)
- Matched ion pairs in each annotation correspond to fragment peaks present in both experimental and database spectra
- No missing values or NA entries in critical rankedResult columns (feature ID, candidate name, match score)
- Visualization via plotResultSpec shows visual agreement between experimental spectra and matched ion positions
Limitations
- Requires raw LC–MS data in centroid mode; profile data must be converted beforehand
- Annotation success is library-dependent — features absent from the fragment database will not be annotated
- In the reference study, only 3 of 6 example features received successful lipid annotations, indicating inherent false-negative rates
- Performance depends on quality of xcms feature detection and RamClustR clustering; poor alignment or clustering reduces annotation reliability
- Installation may require manual fixes for mzR/Rcpp version mismatches or multiarch compilation issues (see README)
Evidence
- [intro] xcms feature detection and RamClustR spectral clustering prerequisite: "An example of feature annotation using LC-MS AIF chromatograms processed using xcms and RamClustR packages"
- [readme] centroid mode requirement: "It requires raw LC-MS AIF chromatograms acquired/transformed in centroid mode."
- [intro] annotateRC function usage with xset and RC objects: "Execute the annotateRC function with the xset, RC object, and LipidPos fragment libraries to match experimental spectra against the database."
- [intro] ranked annotation extraction: "This information can be accessed from the rankedResult object stored in the annotations"
- [intro] annotation success rate in reference study: "Three out of the six features were annotated with to a lipid"
- [intro] downstream visualization and export: "It is possible to visualise the spectra containing the matched ions to each candidate"