An Algorithm for Gene Prediction Based on the Z Curve

The methods based on digital signal processing have been applied to analyze and identify the genes in a DNA sequence. At first, we discussed the Voss mapping and the Z curve representation which can map the DNA alphabetic sequence into the digital sequences. The mathematic expression and biological meaning of the two mappings were studied. This scheme was based on the Z curve representation because of its clear biological meaning. According to period-3 property of protein coding regions, a kind of adaptive filter was proposed to predict the exons in a DNA sequence. The prediction curve of the exons was obtained with the Recursive Least Squares (RLS) adaptive algorithm. It is shown that the RLS algorithm is better than the existing multistage filter for gene prediction by comparing the simulation curves.