Figure 8.

RBF network representation of the relations between effective and ineffective siRNA sequences. The input layer is the set of numerically represented siRNA sequences (A, G, C and T are converted to 1, 2, 3 and 4). Sense strands of siRNAs (cDNAs 5' to 3',19 nucleotides from positions 1 to 19) are described. The hidden layer classifies the input vectors into several clusters depending on the similarities or closeness of individual input vectors. The output layer indicates the effectiveness of individual siRNA sequences (1 : effective gene silencing, 0 : ineffective gene silencing). X_i: i-th input vector, TN: the maximum number of vectors, M_m: the location vector, m: the number of basis functions, σ: the standard deviation, w_i: i-th weighting variable, f(X): weighted sum function.