Recently we have done a targeted lipidomic analysis and measured all quantifiable phospholipids in the brain of a disease model mouse at different time points. Since the dataset is very big, I wonder if we can use some bioinformatics tool to get meaningful understanding. It will be helpful, if anybody can suggest some bioinformatics tool that specifically used for lipid analysis to understand their biological functions.
Dear Abdullah Sheikh ,
I recommend the work of Kai Simons and his colleagues, see for example: https://f1000research.com/posters/5-2144 and https://www.nature.com/articles/s41598-021-98702-5
Best regards.
Dear Dr. Tole S.B., Thank you for your answer. Actually, we have done lipidomic analysis of our samples using mass spectrometry. To understand the meaning of the data in respect of biological and molecular processes, we want to analyze them using bioinformatics tools. We are now exploring the methods that Rob suggested in the previous answer.
I am going to assume you have the lipid concentrations already quantified from the mass spec and you just need to do the statistics and are looking for a systems biology approach to get some biological meaning.
1. Have a look at this playlist from lipid maps in 2021 if you're looking to learn more about this. The first video will go over some of the issues of statistical analysis: https://www.youtube.com/playlist?list=PLftrKvk5gjt4gJmoKG2SxCL_xtSxtRirO
From this play list, I recommend considering LION. It is annotating your lipids so LION recognises them that's the tricky bit, but there are videos on that too.
Also, you can try metaboanalyst pathway analysis (https://www.metaboanalyst.ca/MetaboAnalyst/ModuleView.xhtml) if you have annotated your lipids. This depends on the affected pathway being in this library, which may not necessarily be the case especially for lipids and in mice. I have used it for human plasma lipidomics.
2. I am using WGCNA for human plasma lipidomic data, as a clustering and dimensionality reduction technique, and have found some biologically relevant clusters. The clustering is agnostic and doesn't depend on known pathways being in a library, like pathway analysis does. But what clusters you get is greatly affected by your dataset. This is typically used for genomic data so proceed with caution, especially for brain lipids. You can google 'WGCNA lipidomics' for lipidomics papers using it. A lecture on WGCNA: https://www.youtube.com/watch?v=rRIRMW_RRS4 A diagram to summarise WGCNA: https://wikis.utexas.edu/display/bioiteam/Clustering+using+WGCNA
3. Good old fashioned fold changes and regressions, with Benjamini-Hochberg Procedure to correct for the false discovery rate. Use R to deal with the large amount of data. Check the literature for what pathways your top lipids are involved in. Consider if there is an enzyme involved and if there is a potential lipid ratio that represents enzyme activity in the brain.
Things I haven't tried myself, but am considering:
1. Rob's suggestion, thanks.
2. Lipidome Wide Association Study (may not be appropriate for brain phospholipids though? LION may do similar), this paper uses it for plasma lipidome: https://www.nature.com/articles/s41591-023-02271-1
3. LipidSuite: interactive web server for lipidomics differential and enrichment analysis Article LipidSuite: Interactive web server for lipidomics differenti...
- Badges
- Science method