RIF (regulatory impact factor)

A new application of the RIF algorithm¹⁶ was adopted to predict the regulatory impact of the genes and miRNAs with expression values associated to a given mineral’s amount GEBVs. The algorithm was originally developed to determine the regulatory impact of TFs over selected genes (targets) related to a given trait through their expression values analysis between contrasting groups for the same trait¹⁶. To this end, the impact factor is calculated two scores (RIF 1 and RIF 2). RIF 1 gives high scores to TFs that are most differentially co-expressed, highly abundant, and with more expression difference between the groups. RIF 2 gives a high score to TFs whose expression can predict better the abundance of DEGs¹⁶. In our approach we used the genes and miRNAs with expression values correlated to each mineral, from the previous PCIT analyses, as elements to be tested as regulators and the mineral GEBV as the target, in the place of the TFs in the original application. We used an analogy to the concepts used originally to calculate RIF1 and RIF2, i.e., RIF1 gave high score to mRNAs and miRNAs that were most differentially co-expressed, highly abundant, and with more expression difference between the groups of minerals. RIF 2 gave a high score to mRNAs and miRNAs whose expression can predict better the GEBV for mineral amount.

We carried out 10 different analyses with the RIF algorithm¹⁶, being one for each mineral. As input, we used the GEBVs for the 30 contrasting samples for each mineral as targets (15 representing samples with high mineral mass fraction and 15 with low mineral mass fraction) and the expression values for the genes and miRNAs correlated to the same mineral as elements to be tested. To select these contrasting groups we expanded the sample selection based on GEBVs previously made^9,10. Genes and miRNAs with RIF I or II scores higher than |1.96| were considered as significant, as authors suggest¹⁶.