Evaluating Parallel Ward Algorithm for Drug Discovery | ||||
IJCI. International Journal of Computers and Information | ||||
Article 4, Volume 4, Issue 1, June 2015, Page 29-35 PDF (567.38 K) | ||||
Document Type: Original Article | ||||
DOI: 10.21608/ijci.2015.33959 | ||||
View on SCiNiTO | ||||
Authors | ||||
M. G. Malhat 1; Hamdy Mousa 2 | ||||
1Computer Science dept., Faculty of computers and Information, Menoufia University, Egypt | ||||
2Faculty of Computer and Information Menoufia University | ||||
Abstract | ||||
Millions of compounds are now available in chemical libraries and scientists have to test these compounds against biological targets in order to identify lead compounds. The identification of lead compounds is a key step in the drug discovery process. So, there are many hierarchical clustering algorithms are developed and modified for that purpose. Ward algorithm is one of the most popular hierarchical clustering algorithms that are used in many applications in the drug discovery process because of it is accuracy. But, it has limitation to handle large data sets within a reasonable time and memory resources. In this paper, we evaluate and compare two parallel approaches to run ward algorithm. The two approaches are parallel for loop and MapReduce framework. The results shows that parallel for loop failed to reduce computational time of ward algorithm due to overhead needed for data communications. But, MapReduce framework shows considerable reduction in computational time. The parallel ward algorithm saves 17% of time using three nodes and saves 58% of time using six nodes using MapReduce. | ||||
Keywords | ||||
Drug Discovery; Hierarchical Clustering; Ward Clustering; Parallel for; MapReduce | ||||
Statistics Article View: 214 PDF Download: 313 |
||||