Fig. 1. Classifications of effective siRNA sequences derived by the SOM technique. Two hundred effective siRNA sequences were classified into four groups: SOM-C1–SOM-C4. The ratios of the nucleotides A, G, C and T at positions 1–19 for individual groups are shown in four colors: A, purple; G, brown; C, yellow and T, blue. (A) The SOM-C1 class includes 54 effective sequences. (B) The SOM-C2 class includes 51 effective sequences. (C) The SOM-C3 class includes 48 effective sequences. (D) The SOM-C4 class includes 47 effective sequences.

Fig. 2. Significant nucleotides and their corresponding z values for individual classes. UN_i and LN_i are upper and lower nucleotides for class i (i.e., 1: SOM-C1, 2: SOM-C2, 3: SOM-C3 and 4: SOM-C4) and Z is the corresponding statistic value. (A) SOM-C1 (54 sequences), (B) SOM-C2 (51 sequences), (C) SOM-C3 (48 sequences) and (D) SOM-C4 (47 sequences).

Fig. 3. Upper and lower level significances and their integrated scores. (A) The upper and lower level significances were, respectively, integrated at positions 1–19 for individual classes by using Eqs. (2) and (3). As a result, the integrated upper and lower significances were obtained for the nucleotides A, G, C and T at positions 1–19. (B) The calculation of individual nucleotide scores was based on the upper and lower level significances obtained using Eq. (4). The blanks show no score. The nucleotides C and T at position 2, for example, have no score.

Fig. 4. Relations between effective and ineffective sequence scores of the recently reported siRNAs. After individual scores of effective and ineffective siRNA sequences for MG1, MG2 and MG3 were calculated using Eq. (8), the averages of the individual scores were obtained.

Fig. 5. Score comparisons between the proposed method and other methods. Scores of the effective and ineffective siRNA sequences were computed on the basis of positional scores of the individual reported guidelines shown in Saetrom and Snove (2004). The individual guideline scores are average values of the effective and ineffective siRNAs for MG1, MG2 and MG3. Reynolds et al. (purple), Ui-Tei et al. (brown), Amarzguioui et al. (yellow), Hsieh (light blue), Takasaki et al. (dark brown).

Fig. 6. The proposed score evaluation using human cyclin B1. (A) Gene-silencing efficacy of effective and ineffective sequences for human cyclin B1. The ratio is the number of cyclin B1 gene expressions over the number of tubulin gene expressions in the experiment. The average ratios for seven effective and ineffective sequences were, respectively, 0.9 and 26.9, and the standard deviations were, respectively, 1.1 and 8.2. (B) Effective and ineffective sequences for human cyclin B1. The seven sequences 1–7 are effective siRNAs, and the seven sequences 8–14 are ineffective siRNAs. (C) Average scores for the effective and ineffective sequences. The average for effective sequences was 33.4, and the average for ineffective sequences was −11.8.

Fig. 7. Comparisons of the scores yielded by the previous and present methods. (A) Scores for human cyclin B1. For effective and ineffective sequences the average scores yielded by the present method were, respectively, 33.4 and −11.8, whereas those yielded by the previous method were, respectively, 13.2 and −6. (B) Score comparisons for MG1, MG2 and MG3. For MG1, MG2 and MG3 the individual average scores of effective and ineffective sequences were, respectively, 10.4 and −3.4, 12 and −5.5, and 13.9 and 2 when the present method was used, whereas they were, respectively, 4.2 and −2.2, 6 and −2.9, and 3.8 and 0.2 when the previous method was used.

Fig. 8. Score comparisons between the proposed method and other methods in the case of the 860 effective siRNAs. Scores of the effective and ineffective siRNA sequences were computed on the basis of positional scores of the individual reported guidelines shown in Saetrom and Snove (2004). The individual guideline scores are average values of the effective and ineffective siRNAs for MG1, MG2 and MG3. Reynolds et al. (purple), Ui-Tei et al. (brown), Amarzguioui et al. (yellow), Hsieh (light blue), Takasaki et al. (dark brown).

Preferred and unpreferred nucleotides in the individual classes

C1–C4: SOM-C1–SOM-C4, P: preferred nucleotide, U: unpreferred nucleotide.