Fig. 2From: A scalable assembly-free variable selection algorithm for biomarker discovery from metagenomesAbundance distributions of synthetic bacterial communities. Abundance distribution of sampled species for various datasets of increasing complexity, ranging from 5 to 700 distinct genomes (see main Text and Methods). A four order of magnitude difference between the number of cells from the most abundant versus less abundant organisms is used to determine power law parameters for each dataset (see Methods)Back to article page