8.8.6. Post-assignment statistics

Quantifying differences between two or more samples in the taxa to which reads or contigs map is the ultimate goal for many metagenomic experiments. While not the only option, MG-RAST (metagenomics.anl.gov/) provides an example of statistical comparisons using read mapping. Assuming read accounts are normalized by size of their target (various methods have been used for this), and then the count frequencies themselves can be used with a variety of standard statistics. Similarly, Qiime, mentioned above in section 8.8.3, provides an effective way for mapping reads to microbial taxa.