A systematic evaluation of high-dimensional, ensemble-based regression for exploring large model spaces in microbiome analyses

Table 1 Characteristics of data from the mouse microbiome study

Measurement	Original variable	Pre-processing	Dimension	Role in models
16S	Sequence counts	log (Relative abundances)	344	Independent
ITS	Sequence counts	log (Relative abundances)	109	Independent
C mRNA	Expression levels	log (GAPDH normalized expression levels)	6	Independent
CFU	Colony counts	\(log \left (\frac {\text {Colony forming units (CFU)}}{\text {grams of fecal matter}}\right)\)	1	Dependent (Response)
Antibiotic treatment
±C. albicans challenge
+ Length of treatment	Categorical	Dummy encoding	5 levels	Independent

C. albicans colonization level was measured by counting the Colony forming units (CFU) following quantitative culture of the fecal pellets. Cytokine mRNA expression (C mRNA) was measured in the terminal ileum samples. Bacterial 16S and Fungal ITS were amplified from both the terminal ileum sections and the fecal pellets. These amplicons were then sequenced by 454 pyrosequencing and taxonomically classified with a tailored bioinformatic pipeline [54].

ISSN: 1471-2105