Dealing with Near “outlier” in 3 group NTA Set

[JMS]

I have a set of data from solid phase microextraction GC-MS (single quad) analysis of the headspace of three types of cheese samples. I used MS DIAL to do alignment and identification using the NIST library and RI matching. I used a set of method blanks to test for significance in the features (testing vs. the maximum of the 99% confidence interval), and ended up with 139 features. The cheese types are a ‘control’ cheese, and then the control cheese soaked in two types of wine- ‘alicante’ wine and a ‘cabernet’ wine. A primary question is the differences in the volatile profile between the cabernet and alicante soaked wine.

I did a volcano plot from the abundances (4 sample replicates for each cheese) and it looks like there are a lot of high-fold change relationships that are not statistically significant (unpaired t-test, p =0.05). I think this is because of high variability in the cabernet cheese samples. This is backed up, I think, by the PCA (z-score normalized), which shows not much grouping between the cabernet replicates as compared with the alicante and control. Both are attached.

I am curious if anyone has ideas about different tests that might be able to tease out differences that look like they are there but are not statistically significant by the test I did. Or maybe a better way to preprocess the abundance data prior to testing or PCA?

Attachments:

You must be logged in to view attached files.

[Author]

Posts