On Mon, Apr 26, 2010 at 8:19 PM, Daniel Fernandez <dfernan@gmail.com> wrote:

Hi,

Please help me with this error. I have been more than a week trying to solve this issue and I still can't solve it.

Let me explain my approach to use modeller.
First I search for templates against the pdb database

I select as templates the one with low e-value and reasonable similarity percentage
I use clustalW (or TCoffee) to align the target and the selected templates.
I use modeller to model the target based on the TCoffee alignment file and the PDB files.

I do the whole pipeline but modeller works for some sequences but for most of them it gives me the following error and at this point I am clueless on how to solve it. Here I attach my input files to modeller in case someone wants to take a look at them and help me solve this issue.

INPUT: finalseq.pir (as in modeller format, i tried all different formats here, I attach my last approach that was to only save PDB files with the data from a specific chain...)
             template PDB files (the PDB files with the actual chain)
             target.fasta

OUTPUT: error:
get_ran_648E> Alignment sequence not found in PDB file:        3 2H5Y_A.pdb
              (You didn't specify the starting and ending residue numbers and
              chain IDs in the alignment, so Modeller tried to guess these from
              the PDB file.)
              Suggestion: put in the residue numbers and chain IDs (see the
              manual) and run again for more detailed diagnostics.
              You could also try running with allow_alternates=True to accept
              alternate one-letter code matches (e.g. B to N, Z to Q).
Traceback (most recent call last):
File "testclean.py", line 18, in <module>
    a.make()          # do the homollogy modelling
File "/n/sw/modeller-9v7/modlib/modeller/automodel/automodel.py", line 98, in make
    self.homcsr(exit_stage)
File "/n/sw/modeller-9v7/modlib/modeller/automodel/automodel.py", line 411, in homcsr
    aln = self.read_alignment()
File "/n/sw/modeller-9v7/modlib/modeller/automodel/automodel.py", line 401, in read_alignment
    aln.append(file=self.alnfile, align_codes=self.knowns+[self.sequence])
File "/n/sw/modeller-9v7/modlib/modeller/alignment.py", line 79, in append
    allow_alternates)
_modeller.SequenceMismatchError: get_ran_648E> Alignment sequence not found in PDB file:        3 2H5Y_A.pdb (You didn't specify the starting and ending residue numbers and chain IDs in the alignment, so Modeller tried to guess these from the PDB file.) Suggestion: put in the residue numbers and chain IDs (see the manual) and run again for more detailed diagnostics. You could also try running with allow_alternates=True to accept alternate one-letter code matches (e.g. B to N, Z to Q).

I am completely clueless on where to look the starting and ending residue numbers and chain IDs in the alignment, clustalW does not give me that information at all so not sure where to look that info and if possible with the approach I am using...

Thanks,

Daniel F.