*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
I've put out a new version (0.2.0) for linux and OSX. Only the binary
has changed, so if you already installed the previous release all you
need to do is copy the binary to $CBIN.
The changes are to the joinging and pruning stages, and enhance the
connectivity of the resulting model.
In the joinging stage, I've implemented a threading algorithm for
tracing the longest path through branched chains. (Tassos has discussed
this on occasion). This uses a variant of a critical path analysis
algorithm, modified to handle loops. This method is very fast,
guarantees optimal results in the case of non-looped chains, and gives
good results in most cases of looped chains.
In the pruning stage, I've modified the algorithm to favour long chains
over shorter ones. This also cleans up a few odd orphan residues which
were left by the previous method.
I've also done some analysis of the results against JCSG test data.
Program performance degrades gracefully with both resolution and phasing
quality, without a hard cutoff. As a result, if the phases are good
enough it will work at 3.0 Angstroms. Here are a couple of models built
from real 3A JCSG data:
http://www.ysbl.york.ac.uk/~cowtan/buccaneer/1vp7-buc.pdb
This is an easy case because it is almost all helix. About 300 of 420
residues were built correctly, including some loops, plus about 15
incorrect residues. The calculation took 24 mins.
http://www.ysbl.york.ac.uk/~cowtan/buccaneer/1vkn-buc.pdb
In this case about 1000 from 1400 residues are approximately correct,
plus another 100 incorrect residues. The calculation took 20 mins.
For more details, see:
http://www.ysbl.york.ac.uk/~cowtan/buccaneer/buccaneer.html
