Re: [modeller_usage] not generating pdb files

[Modeller Caretaker <modeller-care@ucsf.edu>]

gerald.quon AT utoronto.ca wrote:
> I'm running modeller 8v0, and am trying to model 2 different proteins on the
> same template.  For some reason, for one of the alignments (#1), the only output
> being produced is the .D* and .V* files - the models aren't being generated (and
> the D* and V* files are being produced quite quickly).
> For alignment #2, I'm getting the models out so that isn't a problem.
>
> Any help would be appreciated...the same commands are being used for both
> alignments, which i've shown below..

There is probably something very wrong with the geometry of your 
starting model. Each optimization is being aborted because the molpdf 
exceeds automodel.max_molpdf (see 
http://salilab.org/modeller/manual/node43.html). You could just increase 
max_molpdf, but if the initial model is very bad, that might not help.

Modeller should list these errors at the end of the run (assuming it 
doesn't fall over during final_malign3d), and you can see from the .D* 
files that the energy is crazy, but it would also be nice to get an 
error message in the log at the moment the error happens. This will be 
rolled into the next release - or you can use the simple patch below:

--- modlib/modeller/automodel/automodel.py
+++ modlib/modeller/automodel/automodel.py
@@ -290,9 +290,10 @@
                                    trace_output=self.trace_output)
             self.write_int(step + 1, num)
             if molpdf > self.max_molpdf:
-                raise ModellerError, \
-                      "Obj. func. (%.3f) exceeded max_molpdf (%.3f) " \
+                err = "Obj. func. (%.3f) exceeded max_molpdf (%.3f) " \
                                    % (molpdf, self.max_molpdf)
+                print "single_mod_E> " + err
+                raise ModellerError, err
         self.schedule.step = len(self.schedule)
         self.refine()


	Ben Webb, Modeller Caretaker
--
modeller-care@ucsf.edu             http://www.salilab.org/modeller/
Modeller mail list: http://salilab.org/mailman/listinfo/modeller_usage