Computational Drug Discovery and Design

The input PDB file must contain the residues to be modeled,

but their atomic coordinates may be arbitrary. The easiest way

to achieve this is to create a linear chain with the missing

residues in the extended conformation as a separate molecule,

then paste it in the original PDB file. Finally, launch the follow-

ing script (here called “Run_Modeller_Loop”) to generate N

models, numbered from MOD_FROM to MOD_TO (this

allows adding new models to the set of the existing ones):

#!/bin/sh

#

# Usage: Run_Modeller_Loop input_file MOD_FROM MOD_TO

#

fname=$1

from=$2

to=$3

cp $fname input.pdb

i=$from

while [ $i -le $to ]

do

echo "Model #"$i

python loop.py $i>$i.log

((i++))

done

At the end the best model can be extracted according to the

optimized energy score, e.g., with the command:

grep -H OBJECTIVE loop* | sort -n -k6

This produces sorted entries with the best models at the

top of the list. This approach uses only command lines, but

there is a graphic user interface (GUI) alternative: the Super-

Looper web site (http://bioinf-applied.charite.de/super

looper). The relatively easy Java-based interface allows the

selection of the gap region and the residue sequence to be

added. It then searches its database for a sequence match.

The score is attributed according to the similarity of the data-

base sequence to the one given in the input, and according to

the satisfaction of the spatial constraints of the loop within the

sequence (seeNote 2).

3. Clean the PDB file. Remove the CONECT records and all
   protons if they exist (as is the case with NMR structures).
   Add TER records after the last residue of each molecular entity
   (polypeptide chain, cofactor, ligand) you wish to include in the
   MD. Look for structure-embedded disordered water

150 Gre ́gory Menchon et al.