I have a very specific problem while modeling PZ (Protein Z). The protein is a molecule of 360 residues with a Gla domain (1-46), two EGFs (47-126) and an SP like domain (135-360). When looked for similarity, some PDBs matched 144-360 residues of PZ with high coverage. However, there are no PDBs that corresponds to the GLA domain (N terminal) and subsequently the whole protein. Now, my study involves the N terminal domain, so I need to model the complete protein.
There is only 1 PDB available (1DAN in RSCB); that has 4 chains in the molecule. But, whenever SWISS modeller is being used, it is only taking a single chain. Is it possible to do in Swiss modeler?
What I have seen, that the PZ is having a high sequence alignment (covering the whole PZ) with 1DAN only when the 2 chains of 1DAN is taken and reorganized (Light chain is added to the heavy chain).
Now, my question is how to model this protein? If MODELLER (developed by Andrej Sali) can be used, then which technique should be followed?
I am attaching the sequences:
(1) PZ (2) 1DAN (3) Sequence alignment of PZ and 1DAN.
Thanks.
1. Sequence of PZ:
>PZ
AGSYLLEELFEGNLEKECYEEICVYEEAREVFENEVVTDEF
WRRYKGGSPCISQPCLHNGSCQDSIWGYTCTCSPGYEGSNCELAKNECHPERTDGCQHFC
LPGQESYTCSCAQGYRLGEDHKQCVPHDQCACGVLTSEKRAPDLQDLPWQVKLTNSEGKD
FCGGVIIRENFVLTTAKCSLLHRNITVKTYFNRTSQDPLMIKITHVHVHMRYDADAGEND
LSLLELEWPIQCPGAGLPVCTPEKDFAEHLLIPRTRGLLSGWARNGTDLGNSLTTRPVTL
VEGEECGQVLNVTVTTRTYCERSSVAAMHWMDGSVVTREHRGSWFLTGVLGSQPVGGQAH
MVLVTKVSRYSLWFKQIMN
2. Sequence of 1DAN:
>1DAN:H|PDBID|CHAIN|SEQUENCE
IVGGKVCPKGECPWQVLLLVNGAQLCGGTLINTIWVVSAAHCFDKIKNWRNLIAVLGEHDLSEHDGDEQSRRVAQVIIPS
TYVPGTTNHDIALLRLHQPVVLTDHVVPLCLPERTFSERTLAFVRFSLVSGWGQLLDRGATALELMVLNVPRLMTQDCLQ
QSRKVGDSPNITEYMFCAGYSDGSKDSCKGDSGGPHATHYRGTWYLTGIVSWGQGCATVGHFGVYTRVSQYIEWLQKLMR
SEPRPGVLLRAPFP
>1DAN:L|PDBID|CHAIN|SEQUENCE
ANAFLEELRPGSLERECKEEQCSFEEAREIFKDAERTKLFWISYSDGDQCASSPCQNGGSCKDQLQSYICFCLPAFEGRN
CETHKDDQLICVNENGGCEQYCSDHTGTKRSCRCHEGYSLLADGVSCTPTVEYPCGKIPILEKRNASKPQGR
>1DAN:T|PDBID|CHAIN|SEQUENCE
NTVAAYNLTWKSTNFKTILEWEPKPVNQVYTVQISTKSGDWKSKCFYTTDTECDLTDEIVKDVKQTYLARVFSYPAGNVE
>1DAN:U|PDBID|CHAIN|SEQUENCE
GEPLYENSPEFTPYLETNLGQPTIQSFEQVGTKVNVTVEDERTLVRRNNTFLSLRDVFGKDLIYTLYYWKSSSSGKKTAK
TNTNEFLIDVDKGENYCFSVQAVIPSRTVNRKSTDSPVECM
3. The sequence alignment created:
PZ AGSYLLEELFEGNLEKECYEEICVYEEAREVFENEVVTDEFWRRYKGGSPCISQPCLHNG 60 1DAN -ANAFLEELRPGSLERECKEEQCSFEEAREIFKDAERTKLFWISYSDGDQCASSPCQNGG 59 .. :**** *.**:** ** * :*****:*:: *. ** *..*. * *.** :.* PZ SCQDSIWGYTCTCSPGYEGSNCELAKNECH--PERTDGCQHFCLPGQ-ESYTCSCAQGYR 117 1DAN SCKDQLQSYICFCLPAFEGRNCETHKDDQLICVNENGGCEQYCSDHTGTKRSCRCHEGYS 119 **:*.: .* * * *.:** *** *:: :...**:::* . :* * :** PZ LGEDHKQCVPHDQCACGVLTSEKRA--------------PDLQDLPWQVKLTNSEGKDFC 163 1DAN LLADGVSCTPTVEYPCGKIPILEKRNASKPQGRIVGGKVCPKGECPWQVLLL-VNGAQLC 178 * * .*.* : ** : :: : **** * :* ::* PZ GGVIIRENFVLTTAKCSLLHRNITVK------TYFNRTSQDPLMIKITHVHVHMRYDADA 217 1DAN GGTLINTIWVVSAAHCFDKIKNWRNLIAVLGEHDLSEHDGDEQSRRVAQVIIPSTYVPGT 238 **.:*. :*:::*:* :* :.. . * ::::* : * .: PZ GENDLSLLELEWPIQCPGAGLPVCTPEKDFAEHLLIPRTRGLLSGWARNGTDLGNSLTTR 277 1DAN TNHDIALLRLHQPVVLTDHVVPLCLPERTFSERTLAFVRFSLVSGWGQLLDRGATALELM 298 ::*::**.*. *: . :*:* **: *:*: * .*:***.: ..:* PZ --PVTLVEGEECGQVLN-----VTVTTRTYCERSSVA---AMHWMDGSVVTREHRGSWFL 327 1DAN VLNVPRLMTQDCLQQSRKVGDSPNITEYMFCAGYSDGSKDSCKGDSGGPHATHYRGTWYL 358 * : ::* * . .:* :* * . : : .*. : .:**:*:* PZ TGVLGSQ-PVGGQAHMVLVTKVSRYSLWFKQIMN-------------- 360 1DAN TGIVSWGQGCATVGHFGVYTRVSQYIEWLQKLMRSEPRPGVLLRAPFP 406 **::. . .*: : *:**:* *::::*.
I would first model each domain separately using the best method based on the level of sequence homology, (homology modeling for good similarity, threading for real but low similarity, and for the domain for which you could not find a template, ab-initio modeling, e.g. with rosetta), then arrange the relative orientation of the individual domain model the best you can - docking or leaving space for unfolded linker segment. You can then combine these domains in a reasonable relative orientation into one pdb file and use this pdb file as a template for homology modeling the full length of the protein.
Hi Soumyadev,
The problem which you have put forth is a very frequent one, I would suggest you to go for Multiple template modelling approach if you choose to use MODELLER, to make your work easy try installing easymodeller.
Alternatively, you can use ITASSER as well, which is one of the best protein modelling servers available.
Hope this helps,
Cheers,
Pankaj
I would first model each domain separately using the best method based on the level of sequence homology, (homology modeling for good similarity, threading for real but low similarity, and for the domain for which you could not find a template, ab-initio modeling, e.g. with rosetta), then arrange the relative orientation of the individual domain model the best you can - docking or leaving space for unfolded linker segment. You can then combine these domains in a reasonable relative orientation into one pdb file and use this pdb file as a template for homology modeling the full length of the protein.
Annemarie Honegger Thank you for the idea. I will try this. Let's see how this works out.
Hi Soumya,
I usually determine the secondary structure of the query protein based on amino acid sequence (u can use PSIPRED or other web servers), then try to find out templates based on secondary structure. Use the template/s for modeling (using MODELLER), where you can specify the chain that you wish to use for modeling, followed by energy minimization of the model generated.
Sudip Kumar Dutta Thanks Sudip, I would like to tell you that I have managed to generate a good structure of the protein. Thanks for your input. Much appreciated.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics