Prediction of antisense oligonucleotide efficacy using aggregate motifs.

Antisense oligonucleotide technology allows the targeted reduction of mRNA expression through the in vitro application of short (approximately 20 nt) DNA molecules. Oligonucleotides are valuable both in the study of gene regulation and for having potential therapeutic effects. In theory, a base sequence complementary to a region of the transcript would hybridize to its mRNA target. Nevertheless, in practice some complementary antisense oligonucleotides are more active and more potent than others in suppressing specific gene expression. We present a novel computational approach to modeling oligonucleotide efficacy that uses aggregate motifs, which are flexible tetramotifs that expand the predictive ability of the data descriptors and the attribute space. We also demonstrate our findings on the largest dataset yet reported in the literature. It was shown that the prediction accuracy was significantly enhanced, offering more than eightfold improvement compared to the traditional methods.