(75f) Designing Highly Active siRNAs From Asymmetry-Based Selection Algorithm Predictions

Authors: 
Malefyt, A. P., Michigan State University
Wu, M., Michigan State University
Chan, C., Michigan State Uiversity
Walton, S. P., Michigan State University


In the development of RNA interference (RNAi) therapeutics, selecting siRNA sequences that complement the messenger RNA (mRNA) target does not guarantee silencing. Factors such as 5’-end stability are known to be critical for ensuring the correct strand is preferentially incorporated into the RNA induced silencing complex (RISC). Two methods for determining this asymmetry between strands are terminal sequence and relative terminal thermodynamic stability. Through the analysis of large siRNA databases, we have shown that highly active siRNA sequences are more likely to have large asymmetry between the sense and antisense 5’-ends in both end sequence nucleotides as well as thermodynamic stability.

We used this information to create an algorithm for predicting highly active siRNA sequences against desired proteins using only the mRNA sequence of the target. The algorithm uses end sequence and thermodynamic stability parameters, trained from existing siRNA activity databases, to rank the probability that an siRNA sequence has high, medium, and low activity for its target gene. We will discuss the applicability of the algorithm for predicting highly active sequences for enhanced green fluorescent protein, EGFP. Additionally, we will highlight comparisons between our technique and other selection approaches.