A simple Bayesian estimate of direct RNAi gene regulation events from differential gene expression profiles

Microarrays are commonly used to investigate both the therapeutic potential and functional effects of RNA interfering (RNAi) oligonucleotides such as microRNA (miRNA) and small interfering RNA (siRNA). However, the resulting datasets are often challenging to interpret as they include extensive information relating to both indirect transcription effects and off-target interference events.

In an attempt to refine the utility of microarray expression data when evaluating the direct transcriptional affects of an RNAi agent we have developed SBSE (Simple Bayesian Seed Estimate). The key assumption implemented in SBSE is that both direct regulation of transcription by miRNA, and siRNA off-target interference, can be estimated using the differential distribution of an RNAi sequence (seed) motif in a ranked 3' untranslated region (3' UTR) sequence repository. SBSE uses common microarray summary statistics (i.e. fold change) and a simple Bayesian analysis to estimate how the RNAi agent dictated the observed differential expression profile. On completion a trace of the estimate and the location of the optimal partitioning of the dataset are plotted within a simple graphical representation of the 3'UTR landscape. The combined estimates define the differential distribution of the query motif within the dataset and by inference are used to quantify the magnitude of the direct RNAi transcription effect.

SBSE has been evaluated using five diverse human RNAi microarray focused investigations. In each instance SBSE unambiguously identified the most likely location of the direct RNAi effects for each of the differential gene expression profiles.

These analyses indicate that miRNA with conserved seed regions may share minimal biological activity and that SBSE can be used to differentiate siRNAs of similar efficacy but with different off-target signalling potential.