Because the short interfering RNA (siRNA) widely used for studying gene functions in mammalian cells varies markedly in its gene-silencing efficacy, several siRNA design rules/guidelines have been reported recently. Analyzing the reported siRNA design guidelines from qualitative and quantitative points of views, we found that they were not always effective selection rules for many other mammalian genes. Though some rules from the guidelines are suitable for extracting effective sequences for specific genes, they might sometimes be unsuitable for selecting sequences for other genes. Since the gene-silencing efficacy depends very much on the target sequence positions selected from the target gene, we examined 860 effective siRNA sequences from 503 different mammalian cDNAs in the literature. As a result, we got many preferred and unpreferred nucleotides different from the ones used in the previous guidelines. These sequence-dependent nucleotides could be used as a more general guideline for selecting new siRNA sequences in target genes. We proposed a measure (score) for selecting effective siRNA candidates based on the positional features of specific significant nucleotides and demonstrated the effectiveness of the proposed measure compared with the recently reported other selection methods. In this paper we also discussed the elimination of ineffective siRNA sequences from target candidates and optimal GC content in siRNA sequences.

View full abstract