Water is one of the most interesting chemical systems to study. Investigation of water clusters can help to understand unique properties of condensed phase and particulate H2O. One of the main problems encountered while studying these systems is the global minimum problem. The potential energy landscape of water clusters becomes more and more complicated with growing number of water molecules. In this work we use a combined approach to search of minima of water clusters. It is based on a combination of three different computational techniques. The first is based on classical molecular dynamics. The second algorithm is aimed at improving orientational structure of water molecules within a given cluster, using a Monte Carlo approach. The third algorithm is based on a Diffusion Monte Carlo method (DMC) combined with local minimization (i.e. PES deformation). The proposed approach is tested on TIP4P water cluster systems. The low energy structures obtained from our optimization scheme are used for analysis of the tendency of transition from amorphous (small clusters) toward ordered, crystal-like structures (big clusters).