I wrote a post on the Seagrass Microbiome website yesterday about my struggles with fungal ITS sequencing data which I thought I’d share here as well in case anyone is looking to jump into the fungal fray. To summarize: changing the default method of the QIIME assign_taxonomy.py script from “UCLUST” to “blast” dramatically increased the number of ITS sequences reads that were classified in my dataset. Blast is not the best (or even a good) method for classifying sequences, but for our purposes we just need kingdom level classification (i.e. we just want to know that we are actually analyzing fungi and not, for example, seagrass) so its likely alright. However, if anyone knows of a way to perform phylogenetic analysis on sequence reads that cannot be aligned (like ITS), please speak up!
After posting about my struggles (and their subsequent solution) on twitter, a real mycologist pointed me towards the methods section of Smith and Peay 2014. In Smith and Peay 2014 the authors use UPARSE (vs. UCLUST) to pick OTUs; something I’ve been thinking about trying for 16S analysis already. Additionally, the article references Peay et al 2013 when discussing taxonomic assignment and Peay et al 2013 uses the QIIME assign_taxonomy.py script, with you guessed it, the “blast” method. Thus, it appears that there is at least some precedence in the literature for using “blast” for ITS sequence classification in QIIME which is reassuring; I just wish I had figured this out earlier!
The full post can be found here: http://seagrassmicrobiome.org/2015/02/23/fungal-its-taxonomy-problem-solved-for-now/
Cassie Ettinger (@casettron) is a PhD student in Jonathan Eisen’s lab and is interested in plant-microbe interactions and host-microbe coevolution.