topic Re: Resources or tutorials for metabolomics data analysis using JMP? in Discussions

Resources or tutorials for metabolomics data analysis using JMP?

Mcc99 — Thu, 25 Apr 2024 18:21:36 GMT

Does anyone have recommendations for resources (tutorials, websites, books, paper, etc) dealing with common metabolomics data analysis in JMP?

Re: Resources or tutorials for metabolomics data analysis using JMP?

Thierry_S — Thu, 25 Apr 2024 20:18:15 GMT

Hi,

Do you need help analyzing Mass Spectrometry spectra (highly specialized, likely not practical in JMP) or with large tables of metabolite relative or absolute quantities? In the latter case, the metabolomic data can be treated as any other -omic (transcriptomic, proteomic) data and used in regression models, correlation, or more advanced models.

Let us know what you need.

Best,

Re: Resources or tutorials for metabolomics data analysis using JMP?

Mcc99 — Mon, 29 Apr 2024 19:42:00 GMT

Yes, with large tables of matabolites with peak intensities or peak areas. I'm more interested in the statistical methods after data cleanup, fragment identification, normalization, etc. Thanks!

Re: Resources or tutorials for metabolomics data analysis using JMP?

Thierry_S — Tue, 30 Apr 2024 01:39:25 GMT

Hi,

Metabolomic data can (and should) be analyzed as any other omic data. Articles about proteomic and transcriptomic data analyses should be applicable (sorry, I do not have specific examples to share). Depending on your level of comfort with JMP, the following may be too simple or too complex.

To start, you may need to transform your intensities/Area under the Peak in Log format.
When working with log-transformed data, the relative fold changes between conditions and a reference condition can be calculated as the difference between those. Note that a tall data format (all data stacked) can speed up this basic calculation.
The Graph Builder Heat Map display can be useful for showing overall trends, but be aware that very large data sets (e.g., primary data) can stall the Graph Builder app.
You may want to explore the data using Principal Component Analysis to assess if there are obvious groups in the data that may indicate batch effects or other dominant trends.
In terms of analyses, I tend to use the Fit model with some JSL coding to iterate across the different analytes. In this case, a wide data format is preferable where the cases/conditions are in rows, and the analytes are in columns.

Considering the breadth of possibilities regarding analyses, proposing direct solutions to your problem is rather challenging. I recommend writing an analysis plan to define your questions (e.g., which metabolites are differentially modulated between Condition A and B) and then seek specific help with those analyses.

Best,