Protein Is Compressible

Abstract

We present a lossless compression algorithm, dubbed here ProtComp, for sequences of amino acids. The algorithm is adaptive, such that it can become efficient for compressing a proteome, i.e. the ensemble of all the protein sequences of a certain organism. The results indicate that protein sequences possess statistically relevant regularities, which can be exploited in order to compress a given proteome, despite previous claims of the contrary. We also show the ability of the algorithm to compare proteomes and based on that to construct evolutionary trees which are biologically plausible.