Primer tests for Fungal ITS regions...plus, statistics!

Reading a good paper is so inherently satisfying--and if you want to share my satisfaction, I recommend this recent piece of literature:

Bazzicalupo AL, Bálint M, Schmitt I. (2013) Comparison of ITS1 and ITS2 rDNA in 454 sequencing of hyperdiverse fungal communities. Fungal Ecology, 6(1):102–9. 

I only wish this paper wasn't paywalled, because it contains quite a bit of useful information that is extremely relevant for the environmental sequencing community.

Firstly, the authors carried out a comparison of ITS primer sets and assessed their ability (and overlap) in recovering different fungal Orders, Families, Genera, and Species. I'm a big fan--these type of primer comparisons are important for figuring out what we might be missing in any given PCR-based approach.

Our results suggest that ITS2 may be more variable and recovers more of the molecular diversity. We confirm an earlier in silico study showing that ITS1 and ITS2 yielded somewhat different taxonomic community compositions when blasted against public databases. However, we demonstrate that both ITS1 and ITS2 reveal similar patterns in community structure when analyzed in a community ecology context. [Bazzicalupo et al. 2013]

Secondly, I feel like I learned some statistics by reading this paper! Or at least, I understood why authors chose the methods they did. I really liked that this paper includes detailed explanation of the statistical tests used to assess the ITS regions and make OTU comparisons. For example:

We compared OTU abundance distributions between the ITS1 and ITS2 datasets at all similarity levels with the KolmogoroveSmirnov (KS) test to see whether the ITS1 or ITS2 would project higher OTU rich- ness in the samples. KS tests are often used to test the distribution of datasets against other distributions, so one may use it to test if a dataset is e.g. normally distributed (Conover 1999). However, the KS test may also be used to compare the shapes of two empirical distributions. Species abundance distributions contain information about both the richness and evenness, thus the comparison of distributions is more meaningful than comparing the means of distributions with e.g. t-tests (Phillips et al. 2012). [Bazzicalupo et al. 2013]

I don't have a strong statistics background (but I'm very aware that I need to become more competent in this area), and this paper helped me understand what types of statistical tests I could apply to environmental sequence data in future analyses. In this regard, the Bazzicalupo et al. methods section was a great change of narrative, compared to the stats-name-dropping-without-explanation I see so often in other papers.