Page 205 - Macromolecular Crystallography

194 MACROMOLECULAR CRYS TALLOGRAPHY

despite technological advances, covered in Chapter black background. The system displayed the elec-
11, methods are still very labour intensive. tron density as a mesh of vectors and the molecular
model as a stick figure that could be interactively
manipulated to fit the density and ‘regularized’
13.6 Model building and molecular (Hermans and McQueen, 1974) to maintain proper
graphics geometry. Most importantly, the adjusted coordi-
nates could be easily saved to disk for another round
Structural genomics has been made possible by
of the iterative process of map calculation and model
many concurrent advances in technology besides
adjustment. The first widely used software of this
those in molecular biology. In particular, advances in
type was FRODO, developed by the crystallogra-
computing power and algorithms go hand-in-hand
pher Alwyn Jones (Jones, 1981) after this proof of
with advances in crystallography. Software to per-
concept.
form most crystallographic tasks, in particular the
The majority of current display representations
CCP4 suite of programs (Collaborative Computa-
and interactive techniques for computer-assisted
tional Project, 1994) and the CNS package (Brünger
molecular design were developed in the 1980s
et al., 1998), is freely available from academic groups.
(Olson and Goodsell, 1992). The colour Evans and
Avery important aspect of crystallographic comput-
Sutherland PS300, a calligraphic vector-drawing
ing is the role of computer graphics.
device driven by a DEC VAX minicomputer, became
the standard platform for macromolecular crystallo-
graphers and molecular modellers. A major draw-
13.6.1 A brief history of molecular graphics
back of these specialized display devices was a price
Physical models have enjoyed a long and success- over $50,000. Many of the techniques developed
ful history in chemistry. Watson and Crick used in universities were incorporated by newly-formed
cardboard cutouts of the DNA bases and observed companies into commercial software for general
the similarity in shape between an AT and a GC molecular modelling.
base pair. This insight led them eventually to the UNIXworkstationsbegantomakeinroadstoward
double helix structure of DNA and a molecular the end of the 1980s, dominating in the 1990s.
mechanism for genetics in 1953. Physical models Machines created by Silicon Graphics (SGI) became
were vital in the first 3D structural determination the preferred tool of the crystallographic and mod-
of a protein in 1958, using X-ray crystallography elling communities. These workstations integrated
(Kendrew et al., 1958). Their myoglobin structure a powerful general-purpose computer and a custom
was modelled by brass Tinkertoy-like (or a Meccano- chip optimized for graphical display on a raster CRT
like) representations for the atoms and bonds, sup- monitor. While less expensive than a VAX/PS300,
ported by thousands of vertical rods. At a scale these workstations remained a costly investment.
of 5 cm/Å, the model occupied roughly a cube of The emerging standards based on the X window
nearly 2 m per side. system led to a system for ‘push-button crystal-
Interactive molecular graphics was pioneered by lography’ by the crystallographer Duncan McRee
Leventhal in the 1960s (Levinthal, 1966). Several (McRee, 1992).
research groups in the late 1970s and early 1980s At the same time, the power and capabilities
began efforts to replace the physical models used by of standard PCs increased dramatically. The
crystallographers with computer graphics. In 1978, Richardsons (Richardson and Richardson, 1992) at
the first protein structure was solved with com- Duke University developed molecular graphics pro-
puter graphics using the GRIP system developed grams to run on the Macintosh. In a dramatic
by Fred Brooks’ computer science group at UNC development, the World Wide Web emerged. Roger
(Tsernoglou et al., 1977). An IBM mainframe per- Sayle created RasMol as part of his PhD disserta-
formed the electron density map calculations while tion at Oxford, and made the source code freely
a DEC minicomputer drove the custom graphics ter- available on the web. This program allowed raster
minal. All drawing consisted of white lines on a display of molecules on UNIX or PCs. The RasMol