Interesting that these questionaire responses are very different from
the numbers obained from looking at downloads counts for other cross-platform
crystallography software. I wonder if they are more reflective of the currently
available ARP-wARP platforms and perhaps the role of administrator-
programmer types as the main responders.
The open source MIFit distribution (recent data since this was available from
Jan 2009) from http://code.google.com/p/mifit/downloads/list
logs today at
84 Windows, zipped
84 Linux, gzipped
85 Windows, installer
I recall a similar disconnect between CCP4 downloads and BB replies to a
questions on optimal hardware platform for crystallographic computing -
basically, an antipathy from specific posters towards the Windows OS but a
lot of downloads logged.
I have come to prefer the laptop crystallography in a Windows enviroment
because it is simpler and more efficient for my current work (early stage drug
discovery and development using fragment screening and protein structure
data) that requires chemistry 'bench' software and tight integration with
project reporting and management functions.
For the record, I don't see any decision-making issues with crystallographic
computing performance on Windows. For example, On a moderately high-end
laptop I will typically be able to line-up and auto-process ~8 image data
sets/hour (either MOSFLM or d*TREK). A basic structure solution task
(MOLREP, 3 rounds of refinement with REFMAC with some water-picking and
validation calculations) is usually performed in well under 5 minutes. Cost is
a
slippery argument since the cost of much of the crystallographic software for
commercial use is often several times that of the computer itself and order of
magnitude more than a copy of MS Office. Many users might prefer a reliable
ready-to-run precompiled executables than potentially dealing with compilation
or run-time issues (many useful posts on this BB are about dealing with
those!) since not everyone has the knowledge time or desire to work them
out.
Hope this does not provoke a 'religious war' from the OS zealots - as I
indicated I think the optimal OS for crystallography depends more on other
factors in your work environment rather than benchmark issues. I would not
take this survey to mean that there is no interest in ARP/wARP on Windows -
especially as it is one of the more unique programs, perhaps without a direct
substitute.
As a sidebar, Windows interoperability and reduction of cost/complexity in the
crystallographic computing environment is the main reason for preserving
arp-waters in CCP4.
Thanks
John Badger, Director of Structural Biology, Zenobia Therapeutics
