TSD Homology modelling protocol

Back to results.

Template selection

HHpred and COMA were employed in order to complement the 3 sequence identity classes.

SWISS-MODEL

The modelling was performed with the webserver of SWISS-MODEL in automated mode and alignment mode.

Visualisation with Pymol

SWISS-MODEL provides a model file containing the error coding in the B factor column. This error coding was adopted for the superimposed reference-model illustration. The command therefore is:

spectrum b, red_blue, <model-name>

Modeller

Get all necessary files <source lang="bash"> cd ../input/

1. HEXA sequence

wget http://www.uniprot.org/uniprot/P06865.fasta sed -ri 's/^>.+/>P06865/' P06865.fasta

1. Reference structure

wget http://www.pdb.org/pdb/files/2GJX.pdb

1. Templates

wget http://www.pdb.org/pdb/files/2GK1.pdb wget http://www.pdb.org/pdb/files/3GH5.pdb wget http://www.pdb.org/pdb/files/1O7A.pdb </source>

Create the pairwise alignments, using Modeller internal methods <source lang=python>

1. !/usr/bin/env python

from modeller import *

OUT='../prediction/'

def makeAli(targetId, targetFile, templateId, templateFile, templateChain) :

       env = environ()
       aln = alignment(env)
       mdl = model(env, file=templateFile, model_segment=('FIRST:'+templateChain, 'LAST:'+templateChain))
       aln.append_model(mdl, align_codes=templateId, atom_files=templateFile)
       aln.append(file=targetFile, align_codes=targetId, alignment_format='FASTA')
       aln.align2d()
       aln.check()
       aln.write(file=OUT+targetId+templateId+'-2d.ali', alignment_format='PIR')
       aln.malign()
       aln.check()
       aln.write(file=OUT+targetId+templateId+'.ali', alignment_format='PIR')
       return

makeAli('P06865','../input/P06865.fasta','2GK1', '../input/2GK1.pdb','A' ) makeAli('P06865','../input/P06865.fasta','1O7A', '../input/1O7A.pdb','D' ) makeAli('P06865','../input/P06865.fasta','3GH5', '../input/3GH5.pdb','A' ) </source>

Call modeller on a single alignment file <source lang="python">

1. !/usr/bin/env python

from modeller import * from modeller.automodel import * import os import sys

ali = sys.argv[1] strc = sys.argv[2] wd = sys.argv[3]

os.chdir(wd)

env = environ()

a = automodel(env,

            alnfile  = ali,
            knowns   = strc,
            sequence = 'P06865',
            assess_methods=(assess.DOPE, assess.GA341))

a.starting_model= 1 a.ending_model = 1 a.make() </source>

Caller script for all modeller predictions

<source lang="bash">

1. !/bin/bash

1. Ugly hotfix so we can work in our own folders later on (who hardcodes filepaths AND doesn't allow output folders?!)

sed -i 's|X:../input|X:../../input|' ../prediction/*.ali

PDBS=( 2GK1 1O7A 3GH5 )

for pdb in "${PDBS[@]}" #Gotta love bash syntax do

 echo "$pdb"
 mkdir -p ../prediction/$pdb/
 mkdir -p ../prediction/${pdb}_2D/

 ./createModel.py ../P06865${pdb}.ali ${pdb} ../prediction/$pdb/ &> ../prediction/$pdb/log
 ./createModel.py ../P06865${pdb}-2d.ali ${pdb} ../prediction/${pdb}_2D/ &> ../prediction/${pdb}_2D/log

done

mkdir -p ../prediction/3GH5_ED/ ./createModel.py ../P068653GH5_edited.pir 3GH5 ../prediction/3GH5_ED/ &> ../prediction/3GH5_ED/log </source>

Evaluation

6Å RMSD using Pymol

• Load template and model
• For each select active site
• Expand selections by 6Å
• Extract selections into objects
• Align objects