Abstract:

Promoter recognition is one of the important research directions in bioinformatics. Due to the characteristics of the promoters, there have been many kinds of identification algorithms based on the signal, the content and the CpG islands. Gene sequences are large scale nonlinear data with high dimensionality, so a new promoter recognition method based on manifold reconstruction is proposed, using nonlinear dimensionality reduction method to compress the data before promoter recognition. The experimental results show that this method can obain better results.

Key words: dimensionality reduction;manifold learning;promoter recognition