The article describes the query parallelization method that takes into account the dependencies between operations in the data query. The method is based on the representation of the query as a directed graph with vertices as operations and edges as data connections. The graph is processed as an adjacency list, saving more memory than during processing a sparse adjacency matrix. The graph is modified only by operations, which do not change the elements of the adjacency list. Therefore it is possible to achieve intra-query parallelism by consideration of a request structure and implementation of mathematical methods of parallel calculations for its equivalent transformation. This article also presents an example of complex query parallelisation and describes applicability of the graph theory and methods of parallel computing both for query parallelisation and optimisation.