Integration of ANN and BLAST (IAB). (a) Calculation of the density of unique subsequence: the solid triangles mark the starting position of the unique subsequences in an N-mer oligo, and the count of the solid triangles is the number of unique subsequences. Thus, the M-mer U
can be calculated from the number of unique subsequences. (b) ANN training: there were 17 input nodes in the ANN for the input vector (10-mer ~ 26-mer U
) that is calculated in (a). In addition, the cross homology identified by WU-BLAST was as the desired output. The monitor object represents the central point that contains all of the parameters needed for other components to work properly. (c) IAB algorithm architecture: for each sliding N-mer oligo, the input vector (10-mer ~ 26-mer U
) calculated by the unique maker database (UMD) was delivered to the ANN for cross homology prediction. The selected oligos were checked by BLAST after filtering by ANN scores.