Lipostar 2.0

Data processing pipeline for untargeted lipidomics

Description

Lipostar is a software developed by Molecular Discovery for LC-MS/MS-based lipidomics (DDA and DIA), which supports a large number of steps including lipid identification, quantification, statistical analysis, and biopathways analysis. Lipostar finds application either in untargeted, and semi-targeted lipidomics, including stable isotope labelling experiments. Within a Lipostar session, different modes of lipidomics analysis can be combined to increase the knowledge and obtain a more comprehensive analysis of lipid profiles. Lipid identification includes 1) a spectral matching approach, with the DB Manager module allowing to generate databases of fragmented lipids by applying fragmentation rules provided in the software or by importing experimental MS/MS data; 2) a high-throughput bottom-up approach, based on class-specific fragments recognition; 3) a high-throughput identification of oxidized species. Lipostar also includes unique features, such as the gap-filler to reduce the missing values and the trend analysis for global lipid profiling.

Technical Information

License:

Conditionally free (academic) / commercial license (commercial use) by request to Molecular Discovery (www.moldiscovery.com).

GUI:

Yes

CLI:

Desktop client:

Yes

Web platform:

Input formats:

.d(Agilent),

.d(Bruker),

.lcd(Shimadzu),

.raw(Thermo),

.raw(Waters),

.wiff(SCIEX)

Output formats:

Word,

CSV

Platforms:

Windows

Programming languages:

C++

Source code repository:

Download / Web-service link:

Yes

Documentation and user guide:

Yes

Training datasets:

Yes

Publications:

PMID:28471643

Tasks

4.1) Full MS (HRAM LC-MS)

Shotgun MS/MS:

LC-MS:

Yes

Algorithms:

MS1 accuracy to a library (generated through the DB Manager) followed by isotopic intensity pattern check. Use of CCS values and RT or retention time ranges if available.

Spectra annotation:

Yes

Support multiple adducts:

Yes

Correction for in source fragments:

Yes

Lipid identification:

Matching to the Library (including isotopic pattern and, when defined, RT and CCS values)

Batch processing:

Yes

Other features:

See Lipid identification using DDA dataset (Data dependent acquisition 4.2).

4.2) Data dependent acquisition (DDA)

Shotgun MS/MS:

LC-MS:

Yes

Spectra matching:

Yes

Decision rules:

Yes

Requirements:

Experimental spectra library, and/or in silico generated spectra from LIPID MAPS and customizable structural databases (via lipid builder module). Fragmentation rules are available with the software.

Algorithms:

MS1 accuracy, MS2 library matching + rule-based assignment

Scores:

Global scoring that takes into account retention time, mass accuracy, isotopic pattern and MS2 fragmentation

Spectra annotation:

Yes

Support multiple adducts:

Yes

Correction for in source fragments:

Yes

Batch processing:

Yes

Other features:

Retention time mapping via Kendrick mass defect (KMD) plots, trend analysis, noise and frequency filters

4.3) Data independent acquisition (DIA)

DIA type and input formats:

.d(Agilent),

.d(Bruker),

.lcd(Shimadzu),

.raw(Thermo),

.raw(Waters),

.wiff(SCIEX)

AIF like:

Yes

SWATH like:

Yes

MS-MS2 Deconvolution algorithm:

Yes

Additional requirements:

Other features:

See Lipid identification using DDA dataset (Data dependent acquisition 4.2)

4.4) Tools considering ion mobility separation

4.5) Identification of oxidized lipids

Lipid coverage:

8 out of 8 LIPID MAPS categories

Spectra matching:

Decision rules:

Yes

Requirements:

Experimental spectra library, and/or in silico generated spectra from LIPID MAPS and customizable structural databases (via lipid builder module).

Algorithms:

Reinspecting the unknown features to verify whether they are compatible with oxidized species whose parent lipid is included in the database. A match based on mass value and on the presence of oxidized MS/MS fragments (if available; fragmentation rules defined) will results in a flag for potentially oxidized species. The list can be exported to be explored with LPPTiger.

Scores:

Global scoring that takes into account mass accuracy, isotopic patter and MS2 fragmentation.

Spectra annotation:

Yes

Other features:

Corresponding MS2 spectra exported to LPPtiger,

RPC RT filter (RT oxidized lipid < RT unmodified lipid)

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

Peak processings:

Deconvolution,

Peak alignment,

Peak picking

Adjustable feature detection:

Peak boundaries, S/N, peak smoothing, gap filter

Quantified using:

Peak area

Normalization methods:

LOESS,

Quality control,

Samples sum/average,

Normalization factor,

Single/multi standard

Relative quantification:

Yes

Semi absolute quantification using internal standards:

Yes

Calibration curve:

Yes

6) Analysis and visualization of lipidomics data

Data pre-treatments:

Scaling (autoscaling, pareto),

Normalization (single/multi standard, normalization factor, samples sum/average, quality control, LOESS),

Filtering

Missing data handling:

Yes

Univariate statistical analysis methods:

Volcano plot analysis,

Fold-change analysis,

ANOVA

Unsupervised multivariate statistical analysis methods:

CPCA,

PCA

Supervised multivariate statistical analysis methods:

O-PLS-DA,

O-PLS,

PLS-DA,

PLS,

DPCA,

LDA

Feature identification:

Clustering and correlation analysis methods:

Trend analysis,

Bisecting K-means,

K-means,

Heatmap

Classification and feature selection:

LDA, PLS, PLS-DA, O-PLS, O-PLS-DA

Direct connection to identification results:

Yes

7.3) Pathway and network solutions

Covered biopathways:

13 maps available, mainly covering GL, GP, FA (including eicosanoids), SL, ST (including bile acids)

Direct projection of identification results:

Yes

Information on species up down regulation displayed on the maps:

Yes

Level of connection:

Mostly subclass level, not lipid species level

Link with enzyme codes:

Yes

Export of the visualized information:

Text file or/and image