Hi Francis,
You may want to look into ACMI, which is designed for building into low
resolution maps.
In my experience with lower resolution data (4.2 - 4.5), "performing
well" is a different story than at higher resolutions. At low
resolution, I'd regard a composite omit map calculated from a final
autobuilt model as a more useful tool than the output model itself. The
sequence and connectivity are less likely to be accurate, simply because
the density doesn't provide sufficient information to deal with them
correctly. But the map can provide some insight into which loop
conformations are accurate, or which regions of density can be
productively modeled. Basically, let the program iterate, go down
(possibly) blind alleys, and backtrack if necessary - because it's
faster, even if it's less accurate.
I've had reasonable luck with buccanneer for these purposes (I'm in the
process of testing arp/warp on my current project - but the runtimes for
large models make this is a slow process).
Pete
This turned out to be more opinion than I'd initially expected; usual
disclaimers apply (your data could behave differently, don't trust my
conclusions without verifying them yourself, etc).
Francis E Reyes wrote:
Hi ccp4bb'ers,
Of the automatic model builders out there (autobuild, arp/warp, buccaneer, insert your own here), are there any opinions/personal experience on which of these perform well in low resolution cases ( worse than say 3.5-3.8 ) ?
Thanks!
F
---------------------------------------------
Francis E. Reyes M.Sc.
215 UCB
University of Colorado at Boulder
