Sequence yields for 12 different sample preparations.

12 cecal and colon derived samples were prepared from four different NOD mice using a variety of RNA extraction protocols. Samples were multiplexed on a single Illumina sequencing run to generate 21.7 million sequences. Of these ∼1.5 million (∼10%) could be mapped to a known bacterial transcript either via BWA against bacterial genomes (nt) or via BLASTX against the protein non-redundant database (peptide). Use of the Invitrogen extraction kit resulted in the most consistent generation of a high proportion of bacterial transcripts. ‘*’indicates the additional use of the RiboMinus kit.