Fig. 1

The overall structure of DiSNPindel. Haar wavelet transformation [11] was used to filter out the high-frequency noisy sub-signals of a base peak (wave). Three features were extracted from the primary and secondary peaks at a base position, namely, horizontal distance, vertical height ratio and half-wave width ratio, which were then inputted into LM-BPNN for intra-individual SNP diagnosis. The LM-BPNN contained three, ten and one neurons (vectors) in the input, hidden and output layers, respectively. Intra-individual InDel diagnosis was conducted using a stepwise allelic base alignment algorithm