Gasbor output graph for errors

Ab initio modelling: DAMMIF, DAMMIN, GASBOR, MONSA
Post Reply
Message
Author
uschulze-gahmen
Member
Posts: 7
Joined: 2009.09.04 22:55

Gasbor output graph for errors

#1 Post by uschulze-gahmen » 2015.09.08 20:52

I ran Gasbor on my data to s=0.35 1/A with a 80 kDa protein using the online website. The resulting shape seems reasonable. I am having difficulties understanding the graphical output for the errors. The red line with the fit error is gradually declining towards higher resolution. But there are 5-6 really high and narrow spikes in this curve. Are these real or is this a bug in the program?
Then there are also points for a "rel Error". What is this error describing? I tried to attach the file with the graphical output.

Final Chi^2 against raw data ........................... : 1.934
Final Chi^2 against corrected data ..................... : 0.8616

Which of the chi^2 is more important?

Ursula
Attachments
prt1_ga1.png
prt1_ga1.png (6.8 KiB) Viewed 3911 times

lwc
Member
Posts: 12
Joined: 2012.07.25 07:52
Location: University of Queensland, Australia

Re: Gasbor output graph for errors

#2 Post by lwc » 2015.09.11 03:54

The line labelled "fit" is almost certainly plotting your experimental errors, and not the actual fit. I haven't used the online version of gasbor myself in several years so I can't comment on why it has plotted that. The spikes are not unusual and are probably due to masks or detector edges in the beamline setup.

Regardless, if you look at the .fir file itself, it should have four or five labelled columns. If you plot sExp vs iExp and sExp vs iFit in your favourite program, you should be able to appraise the fit visually.

For raw vs corrected, gasbor has probably used a background constant as a fitting parameter. iExp will be the original data and iExp* will be the corrected data. Which one you use depends on how suspicious the original subtraction was. Personally I avoid these corrections wherever possible.

- Lachlan

Post Reply