A comparison of computationally predicted functional metagenomes and microarray analysis for microbial P cycle genes in a unique basalt-soil forest [version 1; referees: 2 approved]

Abstract

Here we compared microbial results for the same Phosphorus (P) biogeochemical cycle genes from a GeoChip microarray and PICRUSt functional predictions from 16S rRNA data for 20 samples in the four spatially separated Gotjawal forests on Jeju Island in South Korea. The high homogeneity of microbial communities detected at each site allows sites to act as environmental replicates for comparing the two different functional analysis methods. We found that while both methods capture the homogeneity of the system, both differed greatly in the total abundance of genes detected, as well as the diversity of taxa detected. Additionally, we introduce a more comprehensive functional assay that again captures the homogeneity of the system but also captures more extensive community gene and taxonomic information and depth. While both methods have their advantages and limitations, PICRUSt appears better suited to asking questions specifically related to microbial community P as we did here. This comparison of methods makes important distinctions between both the results and the capabilities of each method and can help select the best tool for answering different scientific questions.