www.Find-a-Drug.org   17 April 2003 Issue 3
 

Thinking on Linux

In response to requests we have now implemented the software on the Linux operating system. This should allow more members to donate their unused CPU cycles to the search for new drugs for diseases such as Cancer and HIV, as well as to combat bioterrorism.

"This initial implementation does not include any graphics or GUI. Command line options are provided for registration and for managing the Job Queue Server." commented Keith Davies, Scientific Director of Find-a-Drug. "In the future we hope to add the option for OpenGL graphics and an X-window GUI for the Find-a-Drug control panel and Job Queue Server."

The software can be downloaded from our UK, European and USA servers. The implementation has been validated on a number of Linux systems including Red Hat Linux 7.1, SuSE 7.2 and Mandrake 8.0. It is expected to work with any version of Linux which has a 2.2.10 kernel (or later) and libc version 2.1.1 (or later).

The computational algorithms used by the THINK software are written in FORTRAN 95. The Linux version uses the Lahey FORTRAN (http://www.lahey.com) compiler and Curl network communication library (http://curl.haxx.se/libcurl/). The performance is comparable to the Windows version. The minimum specification PC has 64Mb of memory, a Pentium processor and 10 Mb of available disk space. We are grateful for assistance from members porting and testing the software.

News about the OU Cancer Project

The Oxford University and United Devices cancer project moved on to its second phase about 9 months ago. At this time it ceased to use the THINK software. Ed Moses resigned as CEO of the Oxford University spin-off company after 9 months and Paul Finn has recently taken over. Paul has significant computational and medicinal chemistry experience. Other experimental validations of the hits predicted by THINK have been in-line with expectations and we wait for similiar results from this project.
 

Progress Monitor

This month is the anniversary of the start of this project. In the weeks since the last newsletter the number of members has grown significantly. We now have over 5,000 registered participants and the number of jobs we are processing exceeds 1000 per day or 300 million molecules per month.

From a scientific perspective progress can be measured in terms of the number of jobs or molecules processed and the number of hits identified. However, the time taken for each job varies quite significantly with the jobs giving the most hits often taking longer. The time spent on a job obviously depends on the speed of the computer and other activities which use CPU time. In response to members' suggestions we have now implemented a points system. This awards 100 points per CPU hour for an average PC. A simple benchmark is used to estimate the speed of the CPU and the average load. While this approach is approximate, it aims to provide the best measure of a members contribution.

We have almost completed our first Bioterrorism Antidote project and our first large scale Cancer target. In this work we screened approximately 0.5 billion molecules and some 3.7 million hits were predicted to bind to the protein target. The analysis of these hits is incomplete at present but it is apparent that they include some significant families of promising molecules. Preparatory work is progressing well for the start of the HIV and other projects. Later this year we hope to be able to announce our first partnership agreement to follow-up the hits.

Most recently we have revised the web-site to give a "cleaner" look and feel. The statistics pages have also been upgraded to implement most of the suggestions for improvements we have received from members. We continue to publish information about the low level of calculation redundancy and histograms of the time and numbers of hits for jobs.


Linux Penguin courtesy of lewing@isc.tamu.edu and The GIMP
Find-a-Drug is a tradename of Treweren Consultants Ltd Copyright © Treweren Consultants Ltd 2003