Genomic Signals Compression by Compressed Sensing and Its Application in Sequences Comparison

The analysis of gene sequences is fundamentally important for exploring biological functions. Recently, the cost of gene sequencing has dropped sharply, thereby resulting in the production of considerable genomic data. However, the costs of saving, processing, and transferring these data are rising. At present, processing this massive volume of information is done by character based method which is highly time - consuming. Alternative methods challenge these problems in the realm of signal processing. Accordingly, the signal outlook to the genome, signal processing of the genome and compression of the genome are presently hot issues which are practically in demand. Compression reduces the cost, memory space, bandwidth for exchange, and the time required for analysis. In this study, the character genes were firstly represented as signals. Then, these genomic signals were compressed by compressed sensing. Consequently, they were reconstructed by bayesian learning method. Adopted criteria for reconstruction were PRD and NMSE, respectively. Then, signals were selected with a compression rate of 75% for comparison. Meanwhile, the same cluster analysis was run with character based method. The results indicated that the time needed for signal based method was considerably lower than the character based method.