Initial Selection Model And Algorithms Of Drug Combinations

With the rapid improvement of technology and agent-to-target experiment data, it is possible for researchers to explore powerful drug combinations for certain complex diseases. This essay discusses how to search for effective com-bined drugs:complex drug agents selection is adapted to covering problems in graphics and combination from experiment data and is shown in bipartite; based on certain hypothesises and logic deduction, two models are constructed and related criterion are set up to analyze the bipartite; three new algorithms are adapted from original greedy methods.This essay is distinctive in several ways. Firstly, an initial selection is added according to current lack of certain agent-to-target experiment data. This would find out not only the necessary part of feasible solution but also needy experiments to improve it. Secondly, edge weighted model is designed by three supposes, target importance equality, same weight of edges incident to same target and drug weight proportional to sum of its incident edges, which a greedy algorithm is based on. Thirdly, in random graph model, from the perspective of possibility, two hypothesises are refreshed that edges are independent and agents selected into drug combination equals the union of its incident edges selected. These changes add randomness into bipartite, which random greedy algorithm is set up on. In addition, the complexness of these three algorithms remain the same as former greedy one.At last, NCI60database was chosen to test the three algorithms. The two new algorithms generated the different solution from former one in such a short time. These solutions are all optimal, and then possible reason was given by the description and analysis of processing and solution characteristics.