Multiple Target Optimal Intervention Using Swarm Intelligence Optimization In Molecular Network

With the changes of times, we have a deeper understanding for disease. We can not only use the network to represent the happening of the disease, but also study the principle of disease and find the cause of disease finally. To find pathology is looking for the molecular targets in a network, and prove that they are related to the disease, then research and development the corresponding drugs, this creates the network pharmacology. The network pharmacology is based on systems biology and network biology, and more conducive to understanding the behavior between cells and organizations at the molecular level, as well as their impact on the entire system. For disease-related molecular network, past single target selective drugs in the treatment of complex diseases has been unveiled its shortcomings, and is difficult to achieve ideal therapeutic effect. Now we adopt the way of multiple target drug treatment, meanwhile effect multiple links in the disease-related molecular network, predict and find the target who plays a recovery role in the system and effect the targets and try its best not to affect the balance of the network at the same time.The multiple target drug therapy can be divided into two steps. First, filtering the potential drug targets in network and understanding their pathogenicity, which also is the key to cure disease molecular network; Second, using the selected drug target to combine and intervention, then getting effective inhibition scheme against the diseases. However, the present algorithm takes too much time for such problems. In this paper, we use swarm intelligence algorithm to solve the molecular network problem and use the arachidonic acid inflammation metabolic network as a concrete example.In this paper, the innovation points will be introduced in detail in chapter3and chapter5. The third chapter puts forward some improvement to the existing particle swarm optimization algorithm. Proposed parallel calculation, initial population uniformity, constraint the update algebra, speed weight and location weight adaptive adjustment, avoid precocity and early termination worst strategy, which is much more conducive to the target selection. The fifth chapter uses the improved particle swarm algorithm for multiple targets screening in AA metabolic network. For the activity of enzymes, three kinds of perturbation scheme are given and the corresponding analysis of experimental results, constraint the objective function, which is much more conducive to the ideal combination scheme. The algorithm can not only filter related key targets, but also get effective combination of drug targets for the disease treatment. In the aspect of consumption on computing time, the algorithm according to the number of progress is several times efficient than the existing algorithms.