Re: [modeller_usage] Modeling a sequence based on a low identity to a template
To: ESTER FUSTE DOMINGUEZ <>
Subject: Re: [modeller_usage] Modeling a sequence based on a low identity to a template
From: Modeller Caretaker <>
Date: Sun, 26 Aug 2007 18:54:14 -0700
Cc:
ESTER FUSTE DOMINGUEZ wrote:
QUESTION 1:
I follow the example "difficult modeling" from Modeller tutorial. However, in
the last step, when I run "mod9v2 model.py" I obtain this misake described below:
...
_modeller.error: read_te_291E> Sequence difference between alignment and pdb :>>
I tried to solve it, but I could'nt. Could you help me, please?
When my sequence was submitted to the mGenThreader server to fold assignment,
the server didn't returned me any significat hit, all confidence levels were
GUESS (p-value >= 0.1)
Then, which hit should I take? The hit which have the smallest p-value or the
hit with the highest percentage identity?
Or maybe If I don't have any significat hit, all models built by MODELLER will
be wrong.Is that true?
With low quality templates, your models are likely to be incorrect.
QUESTION 3:
In "difficult example" in MODELLER's tutorial, in this alignment, I don't know
which is the criterion to choose 30 :A: 209 :A:::: in structureX. How do you
count it? I don't undestand the way you count it.
Actually, that's a typo - it should start from 40:A. If you look in the
original PDB file, the residue numbering starts at 30 (not 1) in chain
A. If you look at the REMARKS in the PDB file, they say:
REMARK 999 SEQUENCE
REMARK 999 RESIDUES 1-29 WERE NOT OBSERVED IN THE ELECTRON DENSITY.
REMARK 999 MASS SPECTROMETRY SHOWED THAT THE PROTEIN HAD BEEN
REMARK 999 PROTEOLYZED DURING CRYSTALLIZATION.
For modeling purposes we discard 10 more residues, because the alignment
between these residues and the first 40 or so residues of our target
sequence is so poor (thus, the model is only for part of the protein).