Cant get Bunch agreement

Interactive modelling (MASSHA, SASpy) and global minimization programs (SASREF, BUNCH, CORAL, GLOBSYMM)
Post Reply
Message
Author
tzacharchenko
Guest
Posts: 1
Joined: 2013.02.28 12:54

Cant get Bunch agreement

#1 Post by tzacharchenko » 2013.02.28 13:08

Hello, im working on a large multi-domain protein of which there are NMR structures, however beyond agreement with ab initio models in GASBOR. Bunch fails to generate a structure in agreement with the scattering data. Ive also tried fixing domains i know are rigid relative to mobile domains and vice versa and still there is no agreement. Annoyingly i can manually change the PDB in PyMol and produce co-ordinates that agree with the scattering curve, however when i use these in Bunch it produces a model that agrees to a lesser extent than the the supplied co-ordinates.

The EOM method however produces agreement with the scattering curves in individual models and the ensemble , why can EOM find agreement with the data and not bunch , also is there anything i can do with BUNCH to make it work?

grondinjm
Active member
Posts: 36
Joined: 2012.06.27 16:27

Re: Cant get Bunch agreement

#2 Post by grondinjm » 2014.05.28 23:24

I could be mistaken, but if you know that some of your domains are mobile, you just have a flexible system and the fit in EOM is reflecting that. I've been told that when you are getting poor fit in BUNCH, you should try EOM.

Alex
Active member
Posts: 692
Joined: 2007.08.09 21:10
Location: Planet Earth

Re: Cant get Bunch agreement

#3 Post by Alex » 2014.05.29 10:49

Hello, im working on a large multi-domain protein of which there are NMR structures, however beyond agreement with ab initio models in GASBOR. Bunch fails to generate a structure in agreement with the scattering data. Ive also tried fixing domains i know are rigid relative to mobile domains and vice versa and still there is no agreement. Annoyingly i can manually change the PDB in PyMol and produce co-ordinates that agree with the scattering curve, however when i use these in Bunch it produces a model that agrees to a lesser extent than the the supplied co-ordinates.
BUNCH, CORAL, SASREF, DAMMI[NF] and GASBOR assume that there is a single conformation in solution. I bet you see some heterogeneity in your
NMR ensembles, like floppy tails, etc.. A good way to start is to check Kratky plot. Personally, I prefer dimensionless Kratky plot compared to well
folded, completely unfolded and "moderately" flexible proteins. This way your can judge where on "flexibility scale" your protein is.

Getting agreement with ab initio methods is much easier, as compared to when you restrict your system and model using combined ab initio/rigid body modeling.
The EOM method however produces agreement with the scattering curves in individual models and the ensemble , why can EOM find agreement with the data and not bunch , also is there anything i can do with BUNCH to make it work?
Typically, the models selected in ensemble can fit data rather poorly when taken individually. However, I did see cases that you describe too. When models are generated with EOM suite, there are only several criteria taken into account - depending on what you select (Native- or Random-like angle distribution) the angle distribution in generated models has to comply and c-alpha clash check. The former gives you more options in terms of conformational sampling (especially if you selected Random-like) as compared to BUNCH. The latter may lead to problematic models which have problems even beyond side chain clashes, hence leading to data overfitting.

HTH,
Alex

Post Reply