Fig. 1

Overall pipeline for mirTarRnaSeq. Part 1: In order to assess miRNA-mRNA relationship across samples, count, transcript/count per million (T/CPM) or read per kilobase per million (RPKM) matrix for miRNA/mRNA sequencing are used as input. The user should do an initial modeling of the data to pick the best regression mode matching their dataset based on the Akaike information criterion (AIC) score. After choosing the appropriate regression model, and organism of interest (for miRanda comparison), the user can now run their model and get a report of the significant miRNA-mRNA relationships to observe if the dataset reflects statistical evidence for the latter relationship. Part 2 (correlation) investigates if there is miRNA-mRNA relationship across time points (T1, T2, T3, ...) or conditions (eg. miRNA-mRNA relationship in high temperature versus cold temperature, verus medium temperature) through correlation. After initial correlation on miRNA-mRNA fold change, a background distribution is estimated through sampling and P value is estimated by ranking of the miRNA-mRNA relationship correlation across the background distribution correlation. Part 3 (interrelation) investigates if there is a miRNA-mRNA relationship between two time points (T1 and/or a control and condition. For this we estimate the difference between the miRNA-mRNA fold change. We then form a background distribution for random differences in fold chance and then rank our difference values against the background distribution to get P value, FDR and test-statistics estimates