Hi all,
I'm trying to use the multifit application on a system with multiple copies of some subunits, but I'm having a couple of problems:
1. I keep getting solutions in which the subunits are overlapping, with identical subunits in exactly (or close to) the same place, leaving empty space elsewhere in the map. If I reduce the max penetration allowed/increase the penetration coefficient, I get no solutions. Are there any other parameters I should try changing? (It's not at all clear what many of the parameters in the alignment.param file are for, so I don't know which others might be relevant). Or anything else I could do?
2. I would like to define x-links involving the subunits with multiple copies, but the multifit application currently doesn't support ambiguity for cross-links. I was planning to modify ProteomicsEMAlignmentAtomics's add_all_restraints() method to add x-links as ConnectivityRestraints instead, but I'm not sure how/where to introduce these changes. Editing the .h/.cpp files in my folder has no effect, and I get errors when changing multifit's __init__.py due to various functions/classes not being defined (they're imported to the .h/.cpp file instead of defined in __init__). I know little about python/c++ beyond syntax, so I may be going about this the wrong way. Do I need to be doing something involving swig? Or something else?
Any help will be greatly appreciated!
Thanks,
- Fiona
On 6/5/14, 12:00 AM, Fiona Bridget Naughton wrote: > 1. I keep getting solutions in which the subunits are overlapping, > with identical subunits in exactly (or close to) the same place, leaving > empty space elsewhere in the map.
My guess would be you need to ask for more fits in this case, but Keren is the ultimate authority here.
> 2. I would like to define x-links involving the subunits with > multiple copies, but the multifit application currently doesn't support > ambiguity for cross-links. I was planning to modify > ProteomicsEMAlignmentAtomics's add_all_restraints() method to add > x-links as ConnectivityRestraints instead, but I'm not sure how/where to > introduce these changes.
You would need to check out the IMP source code, modify the C++ code, and rerun cmake and make. This will regenerate the Python bindings from the C++ code (by running SWIG). Certainly some C++ experience would be required here.
Ben