Fig. 7

a Flowchart of the steps taken in our study to acquire 11 datasets of miRNA expression in serum measured by qPCR and to identify the most suitable single miRNA or a set of miRNAs to use as reference. b Flowchart of our approach to analysis of single miRNAs. Each dataset was analyzed by the same four algorithms implemented in the Python programming language. Algorithms independently assigned a stability value to each miRNA. We changed algorithms to assign a ranking from 0 to 1 based on the stability value (the lower the ranking value, the better the reference), thus each miRNA had 4 ranking values. We averaged the four values for each miRNA, which resulted in a single measure of stability and aggregated the results from 11 datasets. c The outline of the two-pronged approach of our analysis. We first analyzed all single miRNAs and then created all possible average expressions of two or three miRNA-combinations and analyzed the suitability of single miRNAs and their combinations as a good qPCR reference using the algorithms shown in Fig. 1b