Certain investigations on optimization techniques for gene selection in cancer microarray gene expression datasets

Abstract

In the cosmic field of the bioinformatics research large volume of biological data has been generated This is due to the digitization of high throughput device at lower cost which made outstanding growth in data everywhere Handling such numerous data has become extremely challenging technique for selecting the relevant disease causing gene The microarray technology enables us to measure the expression levels of thousands of genes simultaneously providing great chance for cancer diagnosis The primary objective of this thesis work is to make efficient gene selection from the numerous gene by heuristic evaluation approach to identify the cancer causing gene from the given datasets In this work, various novel algorithms like Hybrid Particle Swarm Optimization with Simulated Annealing algorithm HPSOSA Cuckoo Search with Cross Over CSC Modified Honey Bee Mating Optimization HBMO and Parallel Lion Optimization Algorithm PLOA are proposed newly and analyzed using K Nearest Neighbour Support Vector Machine and Naïve Bayes classifiers It is inferred that result produced by the various newly proposed optimization technique shows the optimal result In the first work a new population based stochastic optimization technique called Hybrid Particle Swarm Optimization with Simulated Annealing algorithm HPSOSA is proposed The PSO algorithm is easy to implement in comparison to different optimization techniques and there are a few parameters to adjust PSO is the quicker algorithm in locating global optimal solutions However the weakness of PSO algorithm is that it is simple to fall into local optimum in high dimensional space and has a low convergence rate within the iterative process To overcome these risks a new HPSOSA set of rules is proposed newline