189 | Fascicule 1 ANNALS of Faculty Engineering Hunedoara – International Journal of Engineering Tome XIV [2016] – Fascicule 1 [February] ISSN: 1584-2665 [print; online] ISSN: 1584-2673 [CD-Rom; online] a free-access multidisciplinary publication of the Faculty of Engineering Hunedoara 1. Tanusree ROY, 2. Soma BARMAN PREDICTION OF BREAST CANCER GENE USING ELECTRICAL NETWORK MODEL 1-2. Institute of Radio Physics & Electronics, University of Calcutta, INDIA ABSTACT: Genomic alterations cause cancer. Therefore, analysis of cancer genome sequence is very important for cancer genomics as it gives insights into the disease. Here, an electrical network modeling concept is realized based on amino acid structure and sequence to predict the cancer gene behavior associated with breast cell. The gene network phase and magnitude are investigated for prediction of breast cancer gene. Network phase provides better outcome with respect to magnitude response and achieves 90% accuracy with 97% sensitivity and 0.7 MCC. The proposed model achieves 60 % better accuracy in comparison with the existing nucleotide based methods. Therefore, the results suggest that the proposed modeling concept can be used as effective tool for accurate prediction of cancerous and healthy genes. Keywords: Genomics, Breast cancer, Network modeling, Prediction, Hydrophilic 1. INTRODUCTION Cancer is major genetic disease which causes half of the million deaths in the world [Vogelstein and Kinzler 2004], and breast cancer is one of the most common cancers especially in the female that takes first and second place according to the estimated new cases and deaths respectively [Siegel et. al 2015]. For this reason, early prediction of breast cancer gene is very important for female. Various methods like ultrasound [Stachs et. al 2013], mammography [Moschidis et. al 2014] and biopsy [Veronesi et. al 2010] are used to predict the breast cancer, and among these methods biopsy is best method. But it is very time consuming, requires complicated procedures and most of all invasive. To overcome this restriction of biopsy, network modeling concept is proposed. Over the past few years, network modeling of biological systems is gaining popularity in genomics research to investigate the roles and functions of genetic molecules. For example, an impedance model is developed to study the structural and electrical properties of protein [Alfinito et. al 2008]. A PSpice model of DNA (Deoxyribo nucleic acid) molecule is realized to study the electrical conductivity of DNA [Hodzic and Newcomb 2007]. Passive analog electrical circuits are implemented to model protein structure [Sampath 2006]. Marshall (2010) introduced a passive RC circuit to model DNA/RNA. However, these methods are very complex regarding circuitry and computational load as they are based on nucleotide level. But amino acid based methods are more reliable and simple in predicting genetic abnormalities [McClellan 2012]. Amino acids are the essential ‘building blocks’ of proteins [Vaidyanathan 2004]. Restriction of certain amino acid production may stop the progression of cancer and help to kill cancer cells [http://www.apjohncancerinstitute.org]. Recently, many studies based on amino acid statistics are suggested to predict the cancer gene behavior [Das and Mitra 2011; Barman et. al 2011; Roy et. al 2014]. But accurately predicting cancer genes is still a significant challenge. To overcome the challenge, in this paper a network modeling concept is made to develop amino acid equivalent model for the prediction of cancer associated genes. The main goals of this paper are: ≡ Realize electrical circuit model for individual amino acid based on their atomic structure using passive electrical components resistor, inductor and capacitor.