Drug Discovery using Virtual High Throughput Screening on 32-bit Beowulf Cluster Ng Yan Hong 1 , Stephen Wong 2 , Zhu Weiliang 1 , Chen Gang 1 , Ye Min-Htun 3 , Simon See 3 , Puah Chum Mok 1 1 Singapore Polytechnic,500 Dover Road ,Singapore 139651 y.ng@sp.edu.sg 2 Bioinformatics Institute, 30 Biopolis Street,#07-01, Matrix, Singapore 138671 stephen@bii.a-star.edu.sg 3 Sun Microsystems Inc, Asia Pacific Science & Technology Center, #07-01/13, Central Mall, Singapore 059567 Simon.See@Sun.COM Abstract: Virtual High Throughput Screening requires not only more processors, but also high speed processors. The study conducted to compare mapping of DOCK program on SGI Fuel, Sun Fire V240 and Intel Pentium III, has revealed that Dock programs were written for 32 bit platform, and run well on the 32 bits based Intel processors. Attempts to convert DOCK programs onto the 64-bits processors have neither improve accuracy, nor speed-up the screening process. 1 INTRODUCTION 1.1 Virtual High Throughput Screening (VHTS) Generally speaking, there are four main techniques used for drug development: (1) random screening in bioassay system, (2) active principles of natural products, (3) preparation of structural ana- logues of existing medicinals or other biologically active com- pounds, (4) specific design of drugs to interact with an enzyme or a receptor site. High Throughput Screening (HTS) follows the random screening approach to identify pharmaceutical lead compounds from chemical databases that might contain hundreds of thousands or even millions of molecules. It works on the relationship between ligand structures and their bioactivities. Small ligands targeting on proteins is shown in Figure 1. This process is known as dock- ing. Fig. 1. The concept of docking is like finding keys which best fitted into the lock. Figure 2 shows the result of one of our virtual high throughput screening projects. This protein is associated with a disease. Ex- perimental results suggested that if we were able to stop it from functioning, we could have cured or control the disease. Fig. 2. The diagram shows the target protein and its active site Bioassay Drug Candidates For further study Ligand Database Target Protein Ligand Binding Site