Classification and diagnosis of E.coli using protein sub-cellular localization sites

Amino acids chains combine to make different type of proteins and protein is a molecular instrument. To make a protein, a cell must put a chain of amino acids together in the right order. Bacteria are small living microscopic organisms, due to its variation in functioning; the treatment and usage require their identification. There are two main groups of their classification i.e. gram negative and gram positive bacteria. Escherichia Coli (E. coli) is a gram-negative bacterium. Bacterial cells can be detected by the signaling data of proteins localization sites in them. These sites are detected by different type of methods like X-ray, ultrasound, nuclear magnetic resonance, electronic microscope and others. However, the data obtained by these methods tend to be complex and sometimes become difficult to classify any specie of bacteria. There are some classification systems available but sometime classification becomes difficult. To answer this problem, three different supervised Artificial Neural Networks (ANN) as follows Feed Forward Neural Network (FFNN), Probabilistic Neural Network (PNN) and Linear Vector Quantization Neural Network (LVQ) are used to classify Escherichia coli data. These algorithms are evaluated based on four criteria: Accuracy, Precision, Sensitivity and Specificity. Most efficient neural network architecture with University of California, Irvine (UCI) database for E. coli has been obtained and result comparison with earlier work is also shown.