Skip to contents

MSMICA

Mass Spectrometry Metabolomics Identification through Connections of Untargeted Metabolomics Data

Code Availability

MSMICA is available for non-commercial use only. All commercial use requires a business license. Please contact James Zhan at for more information.

License

MSMICA is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0). See the LICENSE file for details.

Overview

MSMICA is an R package that uses experimentally optimized parameters — including adduct formations, adduct correlations, enzyme-based precursor-product correlations, and isotope correlations — to quantify the confidence of putative metabolite identification using a rigorous Bayesian probability framework.

The algorithm assigns each LC-MS feature a posterior identification probability and outputs a unique list of identified metabolites with one adduct form, m/z, and retention time per entry.

Citation

If you use MSMICA in your research, please cite our original work:

  • Citation TBD — manuscript under review

How MSMICA Works

MSMICA performs identification in three main stages:

  1. m/z annotation — Each observed feature is matched against the KEGG/HMDB metabolite database (combined or KEGG-only) at the user-specified ppm tolerance. All user-specified adduct forms are considered simultaneously.

  2. Clustering of adducts and isotopes — For metabolites whose candidate features form adduct/isotopologue clusters (i.e. two or more adduct forms of the same metabolite co-elute and co-vary in intensity), a Bayesian model incorporating adduct formation, adduct correlation (Spearman r), and isotopologue abundance ratio is used to reduce the identity redundancies of features.

  3. Local optimization per monoisotopic mass — For all remaining features, MSMICA uses a local Bayesian optimization scores each candidate feature using retention-time prediction, precursor-product/transporter Spearman correlation, and biospecimen-specific HMDB concentration priors. This stage iterates until no new metabolite identification can be made.

What MSMICA Can Do

Input: - A LC-MS feature table with m/z, retention time, and per-sample intensities as columns (wide format). - Optional: a list of pre-identified metabolites (e.g. from targeted analysis or authentic standards) with KEGG IDs, m/z, and retention times to serve as network anchors. - Optional: a sample class file to restrict correlation analysis to study samples only.

Output: - A unique list of identified metabolites with assigned adduct, m/z, retention time, identification method, and posterior probability (0–100 %). - When applied to human plasma, urine, tissue, and mouse serum with combined HILIC positive / C18 negative data, MSMICA typically identifies ~3,000 unique metabolites per sample type.

Dependency Installation

MSMICA is not yet on CRAN. Install from GitHub using remotes after installing the package dependencies below.

Package dependencies: dplyr, readr, tidyr, data.table, mgcv, pracma, preprocessCore, imputeLCMD, MetaboCoreUtilsAdduct.

Installation helpers: remotes for GitHub packages and BiocManager for Bioconductor packages.

# Install only missing R package dependencies from CRAN
required_pkgs <- c("dplyr", "readr", "tidyr",
                   "data.table", "mgcv", "pracma", "remotes")
missing_pkgs <- required_pkgs[!required_pkgs %in% rownames(installed.packages())]
if (length(missing_pkgs) > 0) install.packages(missing_pkgs)

# Install Bioconductor dependencies
if (!requireNamespace("BiocManager", quietly = TRUE))
    install.packages("BiocManager")
BiocManager::install(c("preprocessCore", "pcaMethods", "impute"))

# Install imputeLCMD from CRAN
if (!requireNamespace("imputeLCMD", quietly = TRUE))
    install.packages("imputeLCMD")

MSMICA Installation 

# Install MSMICA from GitHub
remotes::install_github("jamesjiadazhan/MSMICA")

To install from a local copy, install the dependencies above first. Then download and unzip the MSMICA repository, copy the local path, replace "/path/to/MSMICA" below, and run:

install.packages("/path/to/MSMICA", repos = NULL, type = "source")

Getting started

Check the “Get started” header on the top of the MSMICA website

Contact