Re: [modeller_usage] Error in ini modeller-generated file

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

On 11/16/2010 07:34 AM, Felipe Villanelo wrote:
> I'm trying of add some residues to an multimeric PDB, with a very simple
> model script:
...
> But after a short run, the program crash with tue following message:
...
> _modeller.FileFormatError: read_pd_702E> File: modelo4-Al.ini, Line
> 2404 Could not parse data from file; file is probably corrupt.
>
> Curiosily, the 'modelo4-Al.ini' file is generated by modeller during the
> run, and I cannot bypass this problem
> The alignment file is ok.
> Maybe the problem is, that there is a ~40 Aa coil with no-template
> sequence to model (between other ssequences with template)

I would need to see your input files to be 100% sure, but my best guess 
is that if you look at line 2404 of that modelo4-Al.ini file, rather 
than atom coordinates you will see asterisks (****). Because PDB is a 
fixed-width format, any coordinates larger than 9999 (or smaller than 
-999) cannot be represented, and probably what is happening is that your 
very long insertion (which Modeller will initially generate as a simple 
extended chain, since it has no additional information) ends up with 
such large coordinates.

I suggest that you don't try to model that 40 aa region at all - 
Modeller will not be able to generate a good model anyway.

I will fix this "properly" in a future release, but if you really want 
to try to model that insertion, you can take that bad .ini file, put in 
proper coordinates using a text editor (even replacing each *** with 
zero coordinates would be an improvement, but sensible coordinates would 
be best) and then rerun your script, using the 'inifile' keyword to 
force use of your "fixed" initial model:
http://salilab.org/modeller/9v8/manual/node26.html

If your problem is something different (no asterisks in your PDB file) 
then please send me your input files so I can reproduce your problem.

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