Figure 3 - Performance of RNAxs on a set of 360 siRNAs targeting the four genes firefly luciferase, PPIB, ALPPL2 and DBI.

siRNAs were grouped into functionality classes of less than 50% mRNA repression <F50, repression of at least 50% F50, 70% F70, 80% F80 or 90% F90. The random distribution is depicted in black. (a–d) Functional class enrichments (light gray) for asymmetry (a), accessibility (b), the combination of asymmetry with self-folding plus free-end (c) and all parameters including accessibility (RNAxs) (d). The three top-ranked siRNAs are all contained in F50 (dark gray). (e) Comparison of RNAxs to other design tools. OptiRNA²⁰, Ambion (siRNA Target Finder), Qiagen (siRNA Design Tool), Invitrogen (Block-iT RNAi Designer), oligowalk²¹ and Sirna (using total score threshold; score > 12) were compared to RNAxs for the four functional classes (<F50, F50, F80, F90). All tools were used with default parameters using the available web servers. For each tool, the repression efficiency of the three best-ranked siRNAs was assessed. RNAxs performed better than the other design tools for all functional classes. (f) Western blot analysis of extracts prepared from EpH4 cells, transiently transfected with scrambled siRNA, Dharmacon mmLEF1 SMARTpool (a combination of four siRNAs) or the single top-ranked siRNA designed with RNAxs. Relative LEF1 expression levels are indicated. Actin protein levels show equal loading. Full-length blots/gels are presented in Supplementary Figure 7 online.