question regarding GASBOR - model seems to be inside out

Ab initio modelling: DAMMIF, DAMMIN, GASBOR, MONSA
Post Reply
Message
Author
jimmyg
Guest
Posts: 2
Joined: 2018.04.04 13:11

question regarding GASBOR - model seems to be inside out

#1 Post by jimmyg » 2018.04.05 09:06

Dear all,

we have collected SEC-SAXS data on a 'cage-like' protein. This is a synthetic and engineered protein and we have biochemical data to confirm that it is very likely prolate in shape with a trimer capping the poles, with each component of the trimer connected to each other by a 'curved rod' with spaces in between the rods (e.g. it is not completely enclosed). We envisage the structure will have pseudo-P32 with an ellipsoid-like shape and flat trimer poles. The core should be hollow.

We have analysed different parts of the SAXS SEC profile and we get consistent Rg's and Dmax's etc which ever part of the curve we use. Back in February we used the online GASBOR software and playing around with different parameters we obtained a model that matched what we were expecting using 'prolate' and P32 symmetry. This gave an excellent visual fit of the calculated SAXS profile to experimental curve, although the Chi2 was quite high - I put this down to incorrect errors. However, we have now (from late March) come back to finish this off (reduce the Chi2) and we don’t get any consistent results, even when we run the same data and parameters as before. Our models now appear like the two trimer poles are packing back-to-back with half length ‘rods’ sticking out away from the central apparent trimer-dimers. It is as if the cage has been sliced along the two-fold axis and both flipped over so the trimer-dimers now forms a core. Chi2s are still high and there is a less well visual fit of the calculated SAXS profile to experimental curve at higher q-values.

Does anyone know what may be going on here? Could there have been an update in the software between February and late March that would account for these discrepancies? Is this a problem with how GASBOR builds models – favouring a compact core? Or are we doing something really stupid? We are at a loss so any help will be greatly received – and I can post any further information if requested.

Many thanks and sorry for the long post

James

User avatar
AL
Administrator
Posts: 632
Joined: 2007.08.03 18:55
Location: EMBL Hamburg, Germany
Contact:

Re: question regarding GASBOR - model seems to be inside out

#2 Post by AL » 2018.04.05 19:43

jimmyg wrote:Could there have been an update in the software between February and late March that would account for these discrepancies?
For sure this is not the case - ATSAS 2.8.3 was released in December'2017, the next release is not available yet. You may check the version/revision in the log file, most probably the first line will comtain "GASBOR 2.3i (r9988)".

Have you checked your data with ambimeter? (Petoukhov & Svergun 2015)

Do you have any atomic models that you could fit to your data using e.g. SASREF?

jimmyg
Guest
Posts: 2
Joined: 2018.04.04 13:11

Re: question regarding GASBOR - model seems to be inside out

#3 Post by jimmyg » 2018.04.10 23:44

Dear AL,

thanks for you suggestions. I have gone away and looked into this and yes it is the same version of GASBOR so this isn't the problem. I ran AMBIMETER and it does give a score indicating there will be some significant ambiguity. I have now run GASBOR locally and I find that ~1 out of 8 runs gives me the sensible envelope whilst the others are not reliable so at least I can now reproduce and average these. I am going to give SASREF a go as well now a a complimentary approach.

Cheers

J

Post Reply