Comments on results presentation

[segondsa]

Summary plots:

• Can't wait to see how the ratios plus 95%CI will look like!

Summary table:

• I don't think showing the FC is a good idea as it could be misleading. The log2 FC is probably enough/better.
• Mean: on raw or log?

Individual lipids: Table:

• Same as before for FC and mean

Graphs comment:

• the p-value and adjusted p-values should be presented with emphasis on the latter.
• 'Paired/Independent (as per experimental design) Welch's test was performed on log/raw data'. Data were log transformed to meet the assumptions for parametric tests.
• NA: The data did not meet the criteria for performing a statistical test. There may have been too few non-zero values. In that case the graph should still be consistent with the design (paired or not).

[segondsa]

I was not too sure about the committing to changes in the R script so I copy below what I think we should do.

switch(stat_test, linear_t_test = "Independent t-test performed on linear data. The samples were not paired /matched.", log2_t_test = "Independent t-test performed on log2 transformed data. The samples were not paired/matched.", linear_paired_t_test = "Paired t-test performed on linear data. The samples were paired/matched.", log2_paired_t_test = "Paired t-test performed on log2 transformed data. The samples were paired/matched.", one_sample_t_test = "The data did not meet the criteria for performing a two sample t-test. One of the conditions had almost all 0 values, so a one-sample t-test was performed for the other condition. The results of this test may not be very robust and the plot should be viewed carefully to understand the data.", none = "The data did not meet the criteria for performing a statistical test. There may have been too few non-zero values.", paired_none = "The data did not meet the criteria for performing a statistical test. There may have been too few non-zero values or the variance of a condition may have been too high." )