PRIMARY STRUCTURES OF PROTEINS AS SPACE-DEPENDENT SIGNALS

Abstract

This minireview is devoted to a scarcely populated but potentially quite interesting field of computational biochemistry: the use of signal analysis methods typical of engineering and physics to describe protein sequences as mono-dimensional series. The protein sequences are described by means of a vector of numerical variables that summarize the autocorrelation structures of the analyzed series. In this way, the "atomic level" of protein sequences description shifts from the pairwise alignment of structures to a self-consistent numerical description of the SINGLE sequence. The main steps of the signal analysis approach can be summarized as follows: (a) use of hydrophobic code for primary structures; (b) treatment of the hydrophobicity distribution along the sequence like a time series, with the corresponding use of nonlinear signal analysis techniques to underpin fine position-dependent properties of the hydrophobicity profiles; these position-dependent properties are summarized by means of self-consistent numerical descriptors at the level of single sequences; (c) adoption of a local approach for both inter-sequence (within homologous series of proteins) comparisons and intra-sequence (among short patches along the same sequence) analyses as a starting point for periodicity detection.