An LC-MS-based lipidomics pre-processing framework underpins rapid hypothesis generation towards CHO systems...

Area of Science:

• Metabolomics
• Lipidomics
• Computational Biology

Background:

• Accurate lipid identification from LC-MS data is crucial for hypothesis generation.
• Traditional methods require time-consuming lipid verifications and can be biased by researcher expertise.
• Meta-analysis of multiple datasets necessitates efficient and unbiased lipid identification strategies.

Purpose of the Study:

• To develop a systematic computational framework for lipid identification from LC-MS data.
• To enable rapid hypothesis generation without relying on retention-time data or extensive experimental validation.
• To improve the accuracy and coverage of lipid identification in complex biological samples.

Main Methods:

• A hybrid technique is employed to deduce parent lipids from deconvoluted spectra by analyzing daughter ions.
• Parent lipid confirmation is achieved using established knowledge of preferred product ions.
• The framework leverages underlying mass features for identification, bypassing retention-time dependency.

Main Results:

• The developed framework achieved an accuracy of approximately 80%, correctly identifying 101 lipid species across 18 classes in Chinese hamster ovary (CHO) cells.
• Lipid inferences from the framework helped explain differences in recombinant-producing capabilities between CHO cell lines.
• A comparative XCMS-based analysis, guided by user knowledge, identified fewer species ( < 60) from a broader range of possibilities.

Conclusions:

• A systematic LC-MS-based framework for lipid identification has been established.
• This framework facilitates rapid hypothesis generation, essential for multi-omics studies and meta-analyses.
• The method offers unbiased and accurate lipid identification, overcoming limitations of traditional approaches.