DOI 10.1007/s00726-005-0225-6 Using cellular automata images and pseudo amino acid composition to predict protein subcellular location

Summary. The avalanche of newly found protein sequences in the post-genomic era has motivated and challenged us to develop an automated method that can rapidly and accurately predict the localization of an un-characterized protein in cells because the knowledge thus obtained can greatly speed up the process in finding its biological functions. However, it is very difficult to establish such a desired predictor by acquiring the key statistical information buried in a pile of extremely complicated and high-ly variable sequences. In this paper, based on the concept of the pseudo amino acid composition (Chou, K. C. PROTEINS: Structure, Function, and Genetics, 2001, 43: 246–255), the approach of cellular automata image is introduced to cope with this problem. Many important features, which are originally hidden in the long amino acid sequences, can be clearly dis-played through their cellular automata images. One of the remarkable merits by doing so is that many image recognition tools can be straight-forwardly applied to the target aimed here. High success rates were ob-served through the self-consistency, jackknife, and independent dataset tests, respectively