Plots summarising the estimated location of the direct target transcripts of hsa-miR-124 16 hours post-transfection. For each plot the x-axis represents the differential expression dataset organised by fold change and is ranked from most up-regulated (left) to most down-regulated (right). The central body of the plot represents how the algorithm traversed the dataset with green vertical lines highlighting a perfect match between the target 3' UTR and the RNAi seed sequence, and blue vertical lines the absence of perfect match. The characteristic triangle emphasises the broadly 'normal' distribution of the dataset (i.e. no overall bias towards up- or down-regulation). The analysis process is directed by the absolute ranking vector (see methods for further detail) and each data point is evaluated sequentially - from the outermost, and most dysregulated transcripts toward the central unaffected transcripts. The prominent black line indicates the location of the estimated optimal partitioning of the dataset with regard to the enrichment of (putative) direct RNAi targets. The uppermost plots of panels A and B trace the enrichment score and attempts to locate the most significant partitioning of the data throughout the analysis procedure. Note that the maximum enrichment score is indicated by "i*". The rightmost plots of panels A and B also describe the enrichment score, that is, but in this context summarise how the estimate of the enrichment score fluctuates as sequential data is processed. See methods section for further details. (Panel A) The data input was the differential expression profile as determined by the LIMMA statistical model. Note that SBSE estimates that the most significant grouping of hsa-miR-124 direct transcript targets are located to the right of the vertical line and are included amongst approximately 15% of the most down-regulated transcripts. (Panel B) The equivalent analysis to that described for panel A, but with the expression profile input shuffled. Note that the previous hsa-miR-124 signature has been abrogated and that there is now an absence of a significant estimate or partitioning of the data.
Wilson and Plucinski BMC Genomics 2011 12:250 doi:10.1186/1471-2164-12-250