MZmine

Processing and annotation of direct infusion, LC-MS/MS, LC-IM-MS, imaging, and GC-MS datasets

Description

MZmine is a modular framework featuring various algorithms for different tasks. Lipid identification is performed by class specific fragmentation rules, or via spectra matching using most common database formats (e.g. json, msp, mgf and jdx), enabling the utilization of in-silico lipid libraries such as LipidBlast for LC-MS/MS data sets, as well as NIST and custom libraries. MZmine is also able to annotate lipids from GC-MS, LC-IM-MS, and imaging datasets. MZmine features a set of visualization tools, including Kendrick mass plots as a graphical screening tool for lipids. Dedicated exports of processed and annotated datasets for MetaboAnalyst and GNPS enable quick access to statistical analysis tools as well as feature-based molecular networking. MZmine includes a generic tool to create lipid subclass databases (can further be modified to search for reaction products e.g. to determine double bond positions by means of Paternò-Büchi reaction).

Technical Information

Documentation and user guide:

Mzmine documentation ,

Mzmine user manual

Download / Web-service link:

Yes

Source code repository:

Yes

Programming languages:

Java

Platforms:

Windows,

Linux,

MacOS

Output formats:

CSV,

MetaboAnalyst,

mzTab,

SQL,

XML,

GNPS,

SIRIUS

Input formats:

mzML,

imzML,

mzXML,

mzData,

netCDF,

.tdf(Bruker),

.raw(Thermo),

.raw(Waters)

Web platform:

Desktop client:

Yes

CLI:

Yes

GUI:

Yes

License:

GPL

Tasks

6) Analysis and visualization of lipidomics data

Direct connection to identification results:

Yes

Classification and feature selection:

Clustering and correlation analysis methods:

Hierarchical,

Expectation-maximization,

Farthest-first,

K-means

Feature identification:

Supervised multivariate statistical analysis methods:

Unsupervised multivariate statistical analysis methods:

PCA

Univariate statistical analysis methods:

ANOVA

Missing data handling:

Yes

Data pre-treatments:

Various filtering options,

Normalization (reference feature)

4.2) Data dependent acquisition (DDA)

Other features:

Retention time mapping via Kendrick mass defect (KMD) plots,

Connected to online databases (KEGG, PubChem, HMDB, YMDB, LipidMaps, MassBank.eu, ChemSpider, MetaCyc)

Batch processing:

Yes

Correction for in source fragments:

Support multiple adducts:

Yes

Spectra annotation:

Yes

Scores:

Spectra matching cosine similarity, rule-based score: presence of fragments

Algorithms:

Spectra matching based on cosine similarity, MS1 accuracy + MS2 rule-based assignment.

Requirements:

Spectra library and fragmentation rules (fragmentation rules are provided)

Decision rules:

Yes

Spectra matching:

Yes

LC-MS:

Yes

Shotgun MS/MS:

Yes

5) Lipid quantification from untargeted lipidomics datasets (HRAM MS, DDA, DIA)

Calibration curve:

Relative quantification:

Yes

Normalization methods:

By assigning features as internal standards

Quantification by:

Peak area

Adjustable feature detection:

Yes

Peak processings:

Peak picking,

Peak alignment,

Deconvolution,

Deisotoping