Homology based Structure prediction protocol BCKDHA
Calculation of Models
To get Modeller started, we first had to create a few starting files.
For each of our template sequences 2bfd and 2r8o we had to create a pairwise alignment with our target sequence 1u5b. This pairwise alignment should be in the modeller PIR format. To build this alignments we wrote a script which can be also run by modeller. This script bases on an example script located in /apps/modeller9.9/examples/comments.
Modeller takes an py-file which specifies the sequences and alignment file to be used. The model-default.py can be found in /apps/modeller9.9/examples/automodel. Here we changed 'alignment.ali' to 1w85_1u5b.pir, knowns = '1W85' (our first target) and sequence = '1U5B' (out template). The file was saved as model1w85.py (for more information see )
Make sure to provide the PDB coordinate files in the folder /atom-files.
Modeller is executed by the following command:
Modeller can also be called with more than one template. A description how this is done can be found here:
The alignment mode can be used for multiple sequence alignments where at least one PDB structure is known.
1. Create an (multiple) alignment of at least the target and the template sequence (Format: FASTA, MSF, CLUSTALW, PFAM and SELEX) We created a multiple alignment using ClustalW<ref>http://www.ebi.ac.uk/Tools/msa/clustalw2/</ref>.
2. Submit the alignment to Swissmodel's alignment mode and make sure it is interpreted correctly
As we want to compare the predicted models to the existing PDB entry we need the PDB file for 1U5B. Here only chain A is important, as the sequence for chain A was used to create the models. Therefore only chain A should be present in a PDB file. Herefore we used a program named getFirstChain.pl provided by <ref>http://www.rosettacommons.org/guide/PDB+Manipulation+Scripts</ref>.
First we calculated the C_alpha RMSD using a tool called DaliLite <ref>http://www.ebi.ac.uk/Tools/dalilite/index.html</ref> which is provided by EMBL-EBI. Then we also used the command line tool "sap".
The TM scores were calculated using the program 'TMScore'<ref>http://zhanglab.ccmb.med.umich.edu/TM-score/</ref>. This program also return a pdb file with superpositioned template and structure in the C-alpha trace, which was used to visualize the modelling results.
All Atom RMSD
To calculate the All atom RMSD score we first had to find out where the catalytic centers are located. This can looked up in UniProt. Then we loaded the target sequence and one model in pymol and aligned them.
align "target sequence", "model"
After that we displayed the sequences
Menubar: Display -> Sequence
and marked the position of the catalytic center on the target sequence by clicking on it. To get the radius of 6 Angstrom around the catalytic centre select "(sele)" on the right side and press
A -> modify -> expand -> by 6A, residues
To build an object containing the region for the model press
A -> modify -> exclude -> object -> target sequence A -> extract object
To build another object containing the region for the target sequence do the steps above again. To calculated the RMSD align the two objects again
align obj01, obj02
The RMS-value of pymol corresponds to the RMSD value.
The created models were visualized using Pymol <ref> http://pymol.org</ref>. The target structure (1U5B, Chain A) and each template were loaded and superpositioned. As the secondary structure elements are best visualized using the 'cartoon' mode, the structures were displayed in cartoon mode by the following commands:
hide everything, all
show cartoon, all
To view two superimposed structures the
align command was applied:
align iTasser_2bfd_mod, 1U5B_chain
A more detailed tutorial can be found here<ref>http://www.mrc-lmb.cam.ac.uk/rlw/text/MacPyMOL_tutorial.html</ref>.