ARISA data from the human gut microbiome can detect individual differences observed by 454 sequencing regardless of binning strategy

ARISA (Automated Ribosomal Intergenic Spacer Analysis) is a low-cost technique that allows for the rapid comparison of different microbial environments. In this study, we asked if a set of ARISA profiles can distinguish human microbial environments from one another with the same accuracy as results generated from 454 high throughput DNA sequencing. Using a set of human microbial communities where the sequencing results cluster by subject, we tested how choices made during ARISA data processing influence clustering. We found that choice of clustering methods had a profound effect with Ward’s clustering generating profiles the most similar to 454 sequencing. Factors such as bin size, using presence or absence calls and technical replicate manipulation had a negligible effect on clustering. In fact, no established bin sizing method reported in the literature performed significantly different results than simply picking bin intervals at random. We conclude that in an analysis of ARISA data from an ecosystem of sufficient complexity to saturate bins, a careful choice of clustering algorithm is essential whereas differing strategies for choosing bins are likely to have a much less pronounced effect on the outcome of the analysis. As a tool for distinguishing complex microbial communities, ARISA closely approximates the results obtained from DNA sequencing at a fraction of the cost; however ARISA fails to reproduce the sequencing results perfectly.