On 10/17/22 11:21 AM, Amir via modeller_usage wrote: > I'm currently using multi-chain modeller script to build a hetero-dimer protein complex > ... > structureX:file.pdb:1:A:+227:B::: 2.30:-1.00 > ...TKVEIKR/QVQLVES... > >P1;target > sequence:target:: :: ::: 0.00: 0.00 > ...TKVEIKR-QVQLVES...
If you don't have a chain break (/) in your target sequence, it is considered a single chain (and so every residue will be labeled in the 'A' chain, no 'B' chain). This is not what you want in this instance since Modeller will add a peptide bond between R and Q here. As you correctly discovered, the solution here is to use a / character instead of the gap.
> I get another error : (modeller.ModellerError: define__595E> Number > of selected atoms in sets 2 & 3 is not the same: 109 118)
You have tried to impose a symmetry restraint between the two chains. That will only work if you have the same number of atoms, which is what the error is telling you here. In any case, since you are modeling a heterodimer you don't want to add a symmetry restraint anyway, so just remove it (as is stated in the manual). You can just use the regular AutoModel class here - no need to create a custom MyModel subclass to impose symmetry.
> And I have another question. Is it possible to build a single > polypeptide chain model with multi-chain approach? Is there a way to > combine multi-chain script with multi-template script to keep the > orientation of the different protein domains?
If you mean "can I use multiple templates when building a multi-chain model" then yes, you don't have to do anything special. You just need to list all of the templates in the `knowns` argument, and provide an alignment file containing "/" chain break characters in the target sequence, as per usual.
Ben Webb, Modeller Caretaker