1 Topics

Expected Time to Completion:

30 - 45 minutes

  • DESeq2
  • edgeR / edgeR-robust
  • metagenomeSeq

 


2 Questions

  • Use closedps in the included example dataset.
  • Differential Abundance - Test for taxa/OTUs that are differentially abundant between the Fast and the Control sample classes. Use phyloseq::mt, DESeq2, edgeR, edgeR-robust, and metagenomeSeq.
  • Plot - Plot your results from the differential abundance testing in an informative way that illustrates the confidence of the test result as well as the taxonomic classification. This plot should be more easily interpretable than a table of values.
  • Compare - Compare the results from each method. Which taxa/OTUs were consistently rejected (significant)? Which methods tended to reject more taxa/OTUs? Which method would you prefer to use on your own data?
  • Bonus - Combine the results of each test method into a single plot.
  • Variance Stabilization - Time-allowing, explore ordination results (plot_ordination) after transforming the count data using DESeq2’s getVarianceStabilizedData and rlog functions.

 


3 Hints

  • There are lots of package-specific idiosyncracies. I have provided example functions to help you quickly convert phyloseq data into the relevant data format. In addition, phyloseq includes a conversion function for DESeq2.
  • Try to remember to unload one package before you switch to the next, to avoid namespace conflicts. e.g. unloadNamespace("metagenomeSeq")
  • Don’t get stuck on plotting. I’m most interested that you are able to successfully run and compare these methods. Organize your results in a data.frame, and then find a ggplot2 plot example online.
  • In case I haven’t mentioned you can replace the otu_table in a phyloseq object using otu_table(physeq) <- x, where physeq is a copy of the phyloseq object you want to modify and x is the new otu_table. This will be very helpful for ordinations on the transformed count values in the variance stabilization question.

 


4 Bonus

  • Try an alternative method that isn’t provided (yet) in this lab / tutorial.