GAMol: Genetic Algorithm based De novo molecule generator
R. Vasundhara Devi, S. Siva Sathya, Mohane Selvaraj Coumar, Mohane Selvaraj · 2014
Abstract. Design & synthesis of chemical molecules for drug discovery, agrochemicals or for polymer industry consumes lot of time and cost. Computer aided molecule design (CAMD) could be used to virtually develop novel molecules at a faster rate and lesser cost. GAMol is a Genetic algorithm based de novo molecule generator from molecular frag-ments. As the chemical search space is enormous and the novel molecule design is a NP-hard problem, Genetic algorithm has been used here to virtually design chemical molecules similar to a reference molecule. A set of twenty fragments derived from ten alkanes and one alcohol molecule has been chosen to carry out de novo molecule design using appropriate fitness function and genetic operators. The virtually designed molecules are evaluated using two fitness functions viz., molecular weight and Tanimoto similarity to a reference mol-ecule. The ability of the de novo program, GAMol to generate virtual molecules was tested for different population size and varying number of generations. The program could be used to design a variety of other virtual molecules by varying the input fragments and reference molecule based on the requirement of the chemical industry.