Application of numerical ecology methods to microarray data reveals obscured patterns in the muosa-associated microbial community of the human colorectum

Abstract

The mucosa-associated microbial community of the human colorectum has been implicated in the pathogenesis of several diseases. Previous studies have largely employed methodologies that failed to reveal patterns generally obscured by inter-subject variability. To overcome this problem we describe the use of numerical ecology methods to analyse data produced using a gut microbe-specific phylogenetic microarray. DNA samples were prepared from biopsy tissue collected from the caecum, transverse, sigmoid and rectum of 10 patients following normal colonoscopy (5 males and 5 females, mean age 56 years). Fluorescently labeled cRNA was prepared for each sample and hybridized to a DNA microarray consisting of 766 unique probes for gut bacteria. We then used analysis with respect to instrumental variables applied to correspondence analysis, which is a method often employed in numerical ecology; but only recently applied to microarray data. This type of analysis allows for the “subtraction” of effects (e.g. subject), which then allows the remaining effects to be effectively compared in a statistically valid fashion. Consistent with previous studies, the diversity profiles generated possessed a marked inter-subject variability. However, we were also able to identify significant differences in the profiles on the basis of sex. Streptococcus and Ruminococcus genera were more abundant in females; and males possessed a greater abundance of Bacteroides and Faecalibacterium. Furthermore, when the subject effect was subtracted, we observed for the first time evidence of a longitudinal gradient for specific microbes with respect to biopsy site. The application of numerical ecology methods to the analysis of data generated with a phylogenetic microarray has proven to be a powerful new methodology in the study of the mucosa-associated microbial community of the human colorectum.