Fig. 2

Similarity Assessment. a Dendrogram depicting cluster analysis shows that samples largely cluster by method and starting amount. b MA plot demonstrating the difference between the miRNA quantifications by the various library preparation methods. Individual points represent the miRNAs quantified; the y-axis of each plot shows the log ratio, which is the difference between the log2 transformed and DESeq2 normalized quantification estimates between the two methods, while the x-axis shows the average expression of each miRNA for the pair of methods compared (also log2 transformed and DESeq2 normalized). See the Similarity analysis section of the methods for more information. Thus the plot on the upper left corner shows the difference in quantification estimates between Clontech and Illumina for each miRNA quantified by both methods. We can see that some individual miRNAs greatly differ between the two methods across the full range of expression levels, as there is a difference of up to roughly 6 between the two methods for some miRNAs (with quantifications being much lower for the Clontech method relative to the Illumina method for these miRNAs), while the range of expression is roughly 6 to 18. As another example, the lower right plot shows the difference between the Deduped method and the Fivepercent method. Quantification estimates are quite similar, except for abundant miRNAs which show lower expression in the Deduped method, as one might expect. c The percent of variance explained by method, starting amount, batch, the number of reads mapped to miRNA, and the variance unaccounted for by these factors. Each point represents the variance explained by each factor for an individual miRNA sequence that was quantified by all of the tested methods