Chemoinformatics is an interface science aimed primarily at discovering novel chemical entities that will ultimately result in the development of novel treatments for diseases. Most of Chemoinformatics methods depend on the generation of chemical spaces into which molecular descriptor data sets are projected and where analysis or design is carried out. It is necessary to reduce such huge chemical space to a small manageable subspace so that a chemist or a druggist can accelerate his research effectively. So the problem of reducing the size of a chemical space of millions of molecules is an important task for virtual screening in drug discovery. Usually chemical compounds are described by three main categories of chemical descriptors, namely; 2Dimensional features, 3Dimensional features, Electro topological state (Estate) attributes. The total number of such descriptors is very large; of the order of 769 descriptors. The present study deals with three challenges, the first challenge is to reduce the descriptor space significantly but at the same time retain the features of chemical space such as diversity, coverage, etc. The second challenge is to design meaningful low-dimensional