From: Ping Wang
Dear all,
----------
From: Remy Loris
Dear Ping,
First thing to ask: Do you know with 100% certainty if your crystals contain a complex? If your crystals are large enough (and you have more than one) you can check this on SDS-PAGE or with mass-spec as long as you are sure to remove all surrounding mother liquid that may contaminate your result. Both proteins should be observed very clearly (80%-20% would be a contamination rather than a co-crystal!). Experience tells us that even with high affinity (nanomolar or better) and preformed complexes, still often only one of the two partners will crystallize.
Second: does your second protein contain more than one domain? If so, there may be domain movements that obscure the desnity or even lead to a completely wrong MR solution. Try rigid body refinement with individual domains first. If this does not work, try MR with the individual domains.
The MR solution of your second protein may also be wrong for any other reason. Check if you can trust the solution based on the statistics provided in the log file. If the LLG is only marginally better after adding the second protein, it will probably be wrong. Check also if there is no real solution hanging around but rejected by the program because (for example) it has just one clash too many.
Remy Loris
Vrije Universiteit Brussel and VIB
----------
From: Remy Loris
----------
From: Ping Wang
Dear Remy,
----------
From: Eleanor Dodson
One possibility is to use buccaneer to build the unknown protein, building into the map phased on the known one. This can wor very well..
Eleanor
Dear all,
Recently I have a dataset with a protein complex including two proteins. Each structure of the single protein is available. When I use phaser to solve the phase, one protein got good density while the other was not so ideal. In order to get a better phase, I try to cut the flexible region of the bad protein. But the result shows no improvement. Does anyone have some suggestion for me? Thanks!
Ping
----------
From: Remy Loris
Dear Ping,
First thing to ask: Do you know with 100% certainty if your crystals contain a complex? If your crystals are large enough (and you have more than one) you can check this on SDS-PAGE or with mass-spec as long as you are sure to remove all surrounding mother liquid that may contaminate your result. Both proteins should be observed very clearly (80%-20% would be a contamination rather than a co-crystal!). Experience tells us that even with high affinity (nanomolar or better) and preformed complexes, still often only one of the two partners will crystallize.
Second: does your second protein contain more than one domain? If so, there may be domain movements that obscure the desnity or even lead to a completely wrong MR solution. Try rigid body refinement with individual domains first. If this does not work, try MR with the individual domains.
The MR solution of your second protein may also be wrong for any other reason. Check if you can trust the solution based on the statistics provided in the log file. If the LLG is only marginally better after adding the second protein, it will probably be wrong. Check also if there is no real solution hanging around but rejected by the program because (for example) it has just one clash too many.
Remy Loris
Vrije Universiteit Brussel and VIB
----------
From: Remy Loris
----------
From: Ping Wang
Dear Remy,
Thank you for your suggestion. I have confirmed that two protein exit in the crystal. Both of protein are single domain protein. Maybe I will check whether the real solution is rejected by program. Thanks.
Ping
----------
From: Eleanor Dodson
One possibility is to use buccaneer to build the unknown protein, building into the map phased on the known one. This can wor very well..
Eleanor
No comments:
Post a Comment