Re: [modeller_usage] Modeling multiple chains with two ligands
To: "Dimitry.A.Suplatov" <>
Subject: Re: [modeller_usage] Modeling multiple chains with two ligands
From: Modeller Caretaker <>
Date: Sun, 08 Apr 2007 11:57:16 -0700
Cc:
Dimitry.A.Suplatov wrote:
I am having a great problem with modeller 9v1.
My aim is to model 2 chain protein from a 2 chain template with a
ligands. I want two ligands: Ca metal ion and SOX (Penicillin G
sulphoxide). My template pdb file briefly looks like this:
ATOM 1 N SER A 3 27.496 31.222 28.262 1.00
48.33 N
ATOM 2 CA SER A 3 27.785 31.042 26.803 1.00
43.97 C
... etc to the end of chain a
TER 1679 ALA A 209
ATOM 1684 N SER B 1 14.845 1.596 -1.736 1.00
6.93 N
ATOM 1685 CA SER B 1 15.548 2.917 -1.659 1.00
6.75 C
ATOM 1686 C SER B 1 16.489 3.059 -2.844 1.00
8.04 C
.....
ATOM 6113 NE ARG B 557 26.460 16.023 30.916 1.00
23.74 N
ATOM 6114 CZ ARG B 557 27.175 15.276 30.094 1.00
19.20 C
ATOM 6115 NH1 ARG B 557 28.016 15.921 29.310 1.00
21.83 N
ATOM 6116 NH2 ARG B 557 27.069 13.952 30.055 1.00
20.49 N
ATOM 6117 OXT ARG B 557 23.973 18.664 34.656 1.00
29.98 O
TER 6118 ARG B 557
.... Cutting useless HETATM
HETATM 6159 CA CA B1568 23.187 -16.230 6.242 1.00
10.86 CA << Ca ion
HETATM 6160 O8 SOX B1569 13.279 0.047 0.037 1.00
37.00 O << SOX ...
HETATM 6161 C7 SOX B1569 13.688 -1.062 0.060 1.00
38.52 C
..
HETATM 6166 O12 SOX B1569 13.790 -4.072 -2.049 1.00
54.27 O
...etc to the end of SOX (I guess resid for SOX = 569, chain B1, ?)
Your SOX residue is number 1569 in chain B. The PDB file format has only
one character for the chain ID, so chain "B1" is impossible.
My alignment looks like this:
-----------------------------------------------------------------------------------------------------
>P1;1gm9
structureX:1gm9.pdb:3:A:569:B1:::: #Starting from A resid =3(because
first two are missing) and finishing with SOX
Consequently, this should also be 3:A:1569:B.
--SSSEIKIVRDEYG-PHIYANDTWHLFYGYGYVVAQDRLFQMEMARRSTQGTVAEVLGK
DFVKFDKDIRRNYWPDAIRAQIAALSPEDMSILQGYADGMNAWIDKVNTNPETLLPKQFN
TFGFTPKRWEPFDVAMIFVGTMANRFSDSTSEIDNLALLTALKDKYGVSQGMAVFNQLKW
LVNPSAPTTIAVQESNYPLKFNQQNSQTA/ #Chain break = ending chain A
SNMWVIGKSKAQDAKAIMVNGPQFGWYAPAY
TYGIGLHGAGYDVTGNTPFAYPGLVFGHNGVISWGSTAGFGDDVDIFAERLSAEKPGYYL
HNGKWVKMLSREETITVKNGQAETFTVWRTVHGNILQTDQTTQTAYAKSRAWDGKEVASL
LAWTHQMKAKNWQEWTQQAAKQALTINWYYADVNGNIGYVHTGAYPDRQSGHDPRLPVPG
TGKWDWKGLLPFEMNPKVYNPQSGYIANWNNSPQKDYPASDLFAFLWGGADRVTEIDRLL
EQKPRLTADQAWDVIRQTSRQDLNLRLFLPTLQAATSGLTQSDPRRQLVETLTRWDGINL
LNDDGKTWQQPGSAILNVWLTSMLKRTVVAAVPMPFDKWYSASGYETTQDGPTGSLNISV
GAKILYEAVQGDKSPIPQAVDLFAGKPQQEVVLAALEDTWETLSKRYGNNVSNWKTPAMA
LTFRANNFFGVPQAAAEETRHQAEYQNRGTENDMIVFSPTTSDRPVLAWDVVAPGQSGFI
APDGTVDKHYEDQLKMYENFGRKSLWLTKQDVEAHKESQEVLHVQR/..*
#Chain break = ending chain B and ".." for two HETATM
That's fine except that there is no chain break between chain B and your
two HETATMs in your PDB. The HETATMs are also in chain B, so you should
leave out that last chain break.
The last chain break is thus also not necessary here. Plus, Modeller
won't build any ligands in your model because you aligned your ligands
with gaps in the model. Replace that final '--' with '..' to get the
ligands.
Modeller starts with this error: "import site' failed; use -v for
traceback"
KeyError: 'No such atom: O12:766:C'
----------------------------------------------------------------------------------------------------------
That means, that it can not see atom O12:766:C.
It can't find the SOX residue because your alignment does not say to
build any ligands. To be sure of the residue numbers, look at the
initial model (.ini) file that Modeller generated.
If I remove user restraints and just use automodel, it output good
model, but without ligands at all!
And the latter is what I absolutely do no understand.
Fix your alignment and everything should work correctly.
My log has this note about SOX:
-----------------------------------------------------------------------------------------------------------
read_pd_459W> Residue type SOX not recognized. 'automodel' model building
will treat this residue as a rigid body.
To use real parameters, add the residue type to
${LIB}/restyp.lib,
its topology to ${LIB}/top_*.lib, and suitable forcefield
parameters to ${LIB}/par.lib.
Indeed.
And nothing about CA (may be the program takes CA as Calpha?).
No, of course not - Calpha is an atom type, not a residue. Modeller
already knows what the "CA" residue type is - it's defined in restyp.lib.
2. Why modeller outputs model with no ligands if I use default automodel
with no restraints (env.io.hetatm =True)?.
Turning on env.io.hetatm instructs Modeller to read HETATM residues from
your template PDBs. It does not force those HETATMs to appear in your
model - the alignment controls that.
P.S. I am using windows compilation. May it be some platform specific
modeller bug?