Abstract

Molecular modeling is a well-established tool used in a variety of scientific disciplines, including pharmaceutical chemistry (the development of new drugs) and bioinformatics (the application of information theory to biological sequence analysis) [1]. Molecular modeling is frequently used for planning experiments, analyzing the results of experiments, and for making predictions in situations where technical limitations make experiments difficult or impossible. Recently the Linux/GNU operating system, running on inexpensive desktop PCs using video cards with powerful hardware graphics acceleration, has grown very popular in molecular modeling due to its low cost, low barrier to use, and compatibility with existing UNIX workstations. The Computer Graphics Lab at the University of California San Francisco [2] is developing an advanced modeling application, Chimera [4], to succeed its well-regarded but aging application, MidasPlus [3]. Chimera is written in C++ and Python and emphasizes programmable extensibility as its primary feature. In recognition of the increasing interest in Linux as a molecular modeling platform, we undertook a port of Chimera to Linux. Porting Chimera to Linux represented a number of challenges which had to be overcome; we describe the problems we faced and our solutions.