Hi,
I just started using XCMS online. The package has identified numerous features that are up or down regulated in my different lc-ms data sets. However, I am not sure which statistical output is most important for prioritizing identified features.
How would you compare p-value vs. q value, vs. t-statistics vs. feature score?
Separately, does does the statistics take into account the total intensity of the peaks to come up with the p-score?
Thank you in advance,
Yevgeniy
The most important outputs for prioritizing features are p-value and fold change.
Typically, you would order by p-value, and filter the feature table based on thresholds like p-value <= 0.001, fold change >= 3.
The p-value is calculated from the t-statistics (Welch t-test, unequal variances).
The q-value (http://http://cran.r-project.org/web/packages/qvalue/index.html) tries to give you an estimate of the false discovery rate
and is calculated based on the distribution of the p-values.
The feature score takes all statistics and the intensity distribution of the feature into account.
However, it is still experimental.