Re: [modeller_usage] Sequence alignment with missing residues
To: "He, Amy" <>, "" <>
Subject: Re: [modeller_usage] Sequence alignment with missing residues
From: Modeller Caretaker <>
Date: Mon, 26 Jul 2021 18:26:03 -0700
On 7/21/21 2:53 PM, He, Amy wrote:
My apologies for the multiple posts. I’m using Modeller to align
sequences, find missing residues in pdb and then perform loop modeling
for the missing regions. And I have a small issue with sequence
alignment when there’re missing residues.
...
The first sequence is read from a pdb file, with some residues (“-”).
And the second sequence is the full sequence. However, because of the
peculiar pattern, the alignment is not correct in this region:
GCPPGG---NDTEL (from the pdb file)
PPGGTGGNDTEL (the full sequence)
I checked the structure file and then I found that the alignment should
be as follows:
GCPP---GGNDTEL (from the pdb file)
PPGGTGGNDTEL (the full sequence)
Is there any way to avoid this type of “mismatch” in the sequence
alignment? Thank you very much for your kind advice in advance.
If you know what the alignment is you want to get, you can just create
that by hand (alignment files are just text files, so you can use any
text editor). You could even script it if you trust the numbering or
SEQRES records in the PDB file (this would be an exercise for the
reader, but it would be easy enough to compare ATOM/HETATM with SEQRES
records to get the two sequences). You don't have to use Modeller's
alignment methods to generate an alignment file.