Skip to main content
Fig. 2 | BMC Bioinformatics

Fig. 2

From: Boolean implication analysis unveils candidate universal relationships in microbiome data

Fig. 2

Boolean implication relationships represent diversity in microbiome data. All four types of Boolean relationships were found in our dataset. a Describes the StepMiner algorithm that creates thresholds for each microbe. Among all samples, the normalized counts (log2 scale) of a particular microbe are sorted, and a step function is fitted where the sharpest change between low microbe count and high microbe count takes place. The midpoint of the step position that minimizes the square error is chosen as the threshold (red line) for each respective microbe. A noise margin of ± 0.5 is considered around the threshold which is  ignored for Boolean analysis. b Depicts a log–log plot of the number of each type of relationship in the main dataset and the corresponding number of microbes that exhibit that specific relationship. Each of the four types of relationships found in our datasets are shown. Each point in the scatter plot corresponds to a sample, where the two axes represent the counts of each microbe. c high → low d low → low e high → high f equivalent. The remaining diagrams (gj) show the correlation distributions according to the four Boolean implication relationships found in our datasets, with correlation on the x-axis and number of microbe pairs on the y-axis

Back to article page