I am trying to analyse metabolomic data with FDR using R software.
False Discovery Rate is multiple comparisons procedure, so you should perform test for statistical significance first and then multiple comparisons. For example, you can perform ANOVA (if your data have required characteristics) to find out, if you had significant result and then simply perform pairwise testing i. e. via t-tests for every pair of "treatments" in your dataset. You will obtain several p-values, which you have to "adjust" by selected method (FDR). For p-value adjusting, you can use simple R command (see below).
https://stat.ethz.ch/R-manual/R-devel/library/stats/html/p.adjust.html
We use the R package "multtest". Here, you can perform various statistical tests, e.g. the Welch t-test and perform various multiple testing corrections, e.g. Benjamini-Hochberg adjustment in order to control the FDR.
False Discovery Rate is multiple comparisons procedure, so you should perform test for statistical significance first and then multiple comparisons. For example, you can perform ANOVA (if your data have required characteristics) to find out, if you had significant result and then simply perform pairwise testing i. e. via t-tests for every pair of "treatments" in your dataset. You will obtain several p-values, which you have to "adjust" by selected method (FDR). For p-value adjusting, you can use simple R command (see below).
https://stat.ethz.ch/R-manual/R-devel/library/stats/html/p.adjust.html
You might check the link collection of Korbinian Strimmer on this topic:
http://strimmerlab.org/notes/fdr.html
http://strimmerlab.org/notes/fdr.html
You may simply use metaboanalyst (google it), which interfaced the appropriate R packages and scripts into a very handsome windows clickable interface.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R