LLR ranking of true Kmers. Each kmer is ranked by its LLR score for multiple coverages and levels of "noise" (colors). The x-axis represents the coverage after downsampling (where "all" represents the entire dataset) and the y-axis the rank of the associated LLR value. Each point represents the median of 10 simulations, with upper and lower quartile bounds given by error-bars. Vertical black dashed lines correspond to the lowest coverage in which a majority of MEME runs (5 iterations) detect any true motif for the corresponding genome. The modified position per kmer is highlighted in each subplot title. Children of "GATC" have been removed from E. coli plots and children of "GGATC" and "GATCC" have been removed from G. metallireducens plots.