To expand on that:
If it chose not to dock sequence, then that basically means that the
electron density doesn't provide compelling evidence for a particular
sequence being right.
A score of 0.5 means that the next best sequence dock had exactly the
same score as the best one. In practice, that means that you would do
just as well docking random sequences into the density. Your score of
0.508934 basically says that.
In practice I'd only expect an 18 residue trace to dock if the density
is quite good and the C-alpha positions correspondingly good. That will
vary a lot depending on resolution and the quality of the density. For a
clear region in a 2.5A map, you've got a good chance, for a mobile loop
then no chance. If it is mistraced of course, all bets are off.
Did you RSR everything first?
Also, the algorithm in Coot uses much of the same code, but run in a
rather different manner from buccaneer, in order to achieve interactive
speed. It might be worth trying buccaneer on it, but getting a longer
fragment is more likely to make a difference.
If you want a test, get a known structure, convert to poly-ala, and chop
it up a bit, and try sequencing the fragments of different length.
Kevin
Paul Emsley wrote:
Phil Evans wrote:
How does the Extensions->Dock sequence command work?
I've got a polyAla model, so I open the dialog, import a sequence
file, click the "Sequence closest fragment" or "Sequence all fragments!"
Then I get in the terminal window something like
Sequence: ??QKDIGVKPEFSFN??
Confidence: 0.508934
From: LQQKDIGVKPEFSFNIP
Chain id: 0 Offset: 114
but nothing seems to have happened.
has it failed? or worked?
OK, well that interface could do with improving. As could the
documentation - thanks for prodding me.
It has failed. It needs 0.9 or greater confidence before it will add
the sidechains (this basically Kevin's buccaneer code).
Paul.