Investigation of genetic-environment interaction in smoking addiction

WHO defines tobacco use as one of the most common public health problems and preventable causes of premature death. Smoking addiction, like many other addictions, is a complex form of behavior with both genetic and environmental aspects. Genetic transmission in smoking addiction has been investigated by family studies, twin studies and molecular genetic studies. Smoking addiction was caused by 56% genetic, 24% familial, and 29% environmental factors. Currently, most of the published molecular genetic studies related to smoking addiction have been done using the functional candidate gene approach. The selection of candidate genes has mostly focused on genes encoding nicotinic acetylcholine receptors, genes affecting nicotine metabolism, dopamine-related genes, and to a lesser extent genes encoding proteins associated with the serotonergic and noradrenergic systems. In this study, it was tried to obtain a smoking addiction model that reveals the genetic and environmental factors affecting smoking addiction and the interactions of these factors with each other. Structural equation modeling analysis and deep learning methods have also investigated the relationship of genes and genetic variants with environmental variables, which have already been discovered to be related to smoking addiction. The polygenic risk score was calculated and included in the model with both ready-made software and software that can be run for certain variants, representing the genetic aspect of smoking addiction. The addiction model obtained supports some of the previously shown relationships between smoking and other substance addictions and socioeconomic status, while also presenting some new findings. In addition, a multi-layer perceptron based decision support system has been developed and a classification performance of 85% has been achieved. In order to increase the performance of the decision support system, SEM analysis was used for the first time to reduce the size of the dataset. With this method, it has been shown that a successful feature selection can be achieved at a competitive level with the methods in the literature. Thus, a model was obtained that reveals the infrastructure of smoking addiction with both genetic and environmental aspects, while contributing to the literature with developed PRS software and SEM analysis based feature selection method.