[ccp4bb] Molrep with two models
Dear all, I tried MolRep with a two domain protein. I have cut the two domain as one domain rotates which prevent a search with the complete model. After I finished the first run. I put this solution as fixed input model in the second molrep run with the second domain. The resulting solution positioned the tow expected molecules, but not nearby the two already found domains. Interestingly a symmetry mate would be at the correct position. I could manually write out the coordinates of the symmetry molecule to put the two domains together, but I thought Molrep would position it close to the already found domains because they belong together and in fact they are one polypeptide chain. Why does this not happen? Do I have to incorporate additional information or constrains for the Molrep run? Thanks and Best Regards Christian
Re: [ccp4bb] Molrep with two models
http://www.ysbl.york.ac.uk/~alexei/molrep.html#stick On Thu, 2011-02-17 at 19:32 +0100, Christian Roth wrote: Dear all, I tried MolRep with a two domain protein. I have cut the two domain as one domain rotates which prevent a search with the complete model. After I finished the first run. I put this solution as fixed input model in the second molrep run with the second domain. The resulting solution positioned the tow expected molecules, but not nearby the two already found domains. Interestingly a symmetry mate would be at the correct position. I could manually write out the coordinates of the symmetry molecule to put the two domains together, but I thought Molrep would position it close to the already found domains because they belong together and in fact they are one polypeptide chain. Why does this not happen? Do I have to incorporate additional information or constrains for the Molrep run? Thanks and Best Regards Christian -- I'd jump in myself, if I weren't so good at whistling. Julian, King of Lemurs
Re: [ccp4bb] Molrep with two models
The molecular replacement program does not know about your molecule being a single polypeptide chain. The problem is fit two bodies therefore the program fits two bodies. The centre of mass of whatever you wish to position is placed the standard asymmetric unit used by the program. If it happens that the center of mass of the second domain - linked to the (already positioned) first domain by the polypeptide linkages - does not fall within that asymmetric unit, then it won't be this one that will be provided in the resulting pdb file. An equivalent will be provided. Checking the symmetry equivalents (including limited translations) and writing out the correct equivalent to replace that you do not wish to have is the way to go. Just as you did. In case of a search using the entire molecule, this problem does not occur. And if you go from a molecular replacement solution provided by program X to an automatic model rebuilding program called Y, these two programs will not necessarily use the same conventions. Hence you may end up with the model after automatic model rebuilding that does not superimpose with the input file coming from molecular replacement. Same reason there. Does that answer your question? Fred. Christian Roth wrote: Dear all, I tried MolRep with a two domain protein. I have cut the two domain as one domain rotates which prevent a search with the complete model. After I finished the first run. I put this solution as fixed input model in the second molrep run with the second domain. The resulting solution positioned the tow expected molecules, but not nearby the two already found domains. Interestingly a symmetry mate would be at the correct position. I could manually write out the coordinates of the symmetry molecule to put the two domains together, but I thought Molrep would position it close to the already found domains because they belong together and in fact they are one polypeptide chain. Why does this not happen? Do I have to incorporate additional information or constrains for the Molrep run? Thanks and Best Regards Christian