How is shRNA knockdown score calculated?

VectorBuilder applies rules similar to that used by the RNAi consortium (TRC) to design and score shRNAs. For each given RefSeq transcript, we search for all possible 21mers that are considered as candidate target sites. Candidates are excluded if they contain features thought to reduce knockdown efficiency/specificity or cloneability, including a run of ≥4 of the same base, a run of ≥7 G or C, GC content <25% or >60%, and AA at the 5’ end. Knockdown scores are penalized for candidates that contain internal stem-loop, high GC content toward the 3’ end, known miRNA seed sequences, or off-target matches to other genes. For genes with alternative transcripts, target sites that exist in all transcripts are given higher scores.

All scores are ≥0, with mean at ~5, standard deviation at ~5, and 95% of scores ≤15. An shRNA with a knockdown score about 15 is considered to have the best knockdown performance and cloneability, while an shRNA with a knockdown score of 0 has the worst knockdown performance or is hard to be cloned.

Please note that knockdown scores are only a rough guide. Actual knockdown efficiency could depart significantly from what the scores predict. Target sites with low scores may still work well. Also, please note that targeting 3’ UTR can be as effective as targeting coding region.