Genomic Mappings and Spectral Analysis in The Frequency Domain

Departing from statistical methods, we examine large-scale genomic and proteomic data by applying spectral estimates and measures such as energy, power and cross spectral densities. The frequency analysis and spectral methods have significant advantages, guarantee robustness, enable consistent quantitative analysis and provide qualitative features. The symbolic and numeric approaches provide the overall coherency. The frequency-domain analysis necessitates one to use numeric mappings of finite sequences. Though additional studies and consistent evaluations are needed to assess the proposed methodology, we demonstrate promising consistency, data cohesiveness as well as the genomic and proteomic correlations. Regression analysis and classifications can be achieved under large uncertainties (gaps, errors, missing sites, inconsistency, etc.). The analysis of sequences and information complexity requires a great number of assumptions, hypotheses and postulates. We minimize the number of assumptions applied. The results are illustrated for HIV, cancer and other sequences.