A Multi-Threaded Accelerated Drug Combination Pathway Network Implementation

A Drug Combination Pathway Network is a new approach to predict the most efficient concentration ratio between two or more compounds. The state of the art implementation of the pathway network was introduced by Jiangyong and Wei, et al. in 2016. However, the time complexity of this implementation is O(CV3), where $V$ is the number of nodes in the network and $C$ is the number of concentrations. In this paper, we describes an approach that takes advantage of a multi-core CPU for calculating the result of the pathway network. The time complexity of our implementation is O(EV + $V$ 2ZogV) under PRAM model assuming C processors, where $E$ is the number of edges in the network. Experimental results on an AMD Ryzen™ 3900X CPU show performance improvement of more than 156 times faster than the original C++ implementation.