Does my p(r) function look OK?

Frequently Asked Questions about ATSAS small-angle scattering analysis program package
Post Reply
Message
Author
User avatar
AL
Administrator
Posts: 690
Joined: 2007.08.03 18:55
Location: EMBL Hamburg, Germany
Contact:

Does my p(r) function look OK?

#1 Post by AL » 2010.04.04 16:56

Q: Could you please have a look at my p(r) function - is it OK? What should a proper p(r) function look like?
A: Let's have a look at five objects of different shapes but same Dmax and same molecular weight.

Image Image Image
Adopted from Svergun & Koch SAS studies of biological macromolecules in solution. Rep. Prog. Phys. 66 (2003) 1735-1782, Fig. 5 (c)IOP Publishing Ltd.

As you can see the p(r) functions can be quite different. But they are all smooth, positive and smoothly converge to zero at Dmax.
If you use GNOM you can improve your p(r) function by altering these parameters:
  • Dmax;
    alpha;
    nBeg (smin);
    nEnd (smax).

Aidee
Active member
Posts: 19
Joined: 2010.08.10 13:27
Location: UNIVERSITY OF MANCHESTER

Scattering pattern

#2 Post by Aidee » 2011.01.10 19:49

I have used Gnom to work out the p(r) function and it is suggestive of an ellipsoid/sphere but I dont know what software to use for working out the scattering profile on the left [lg I(s), relative]. Do I just plot it from the raw data? I should expect that both patterns will suggest the same shape right? Thanks.

User avatar
AL
Administrator
Posts: 690
Joined: 2007.08.03 18:55
Location: EMBL Hamburg, Germany
Contact:

#3 Post by AL » 2011.01.10 20:30

For given experimental data GNOM finds such a p(r) function (right) from which a Fourier transform (left) makes the best fit to the experimental data (not shown). The *.out file produced by GNOM contains the p(r) function, the regularized curve and the input experimental data.

In order to look at the shape of the particle directly one can use an ab initio modeling program, e.g. DAMMIF.

Aidee
Active member
Posts: 19
Joined: 2010.08.10 13:27
Location: UNIVERSITY OF MANCHESTER

#4 Post by Aidee » 2011.01.10 20:49

From the gnom.out file, how do I know which column is for the Fourier transform plot?

User avatar
AL
Administrator
Posts: 690
Joined: 2007.08.03 18:55
Location: EMBL Hamburg, Germany
Contact:

GNOM output

#5 Post by AL » 2011.01.10 21:13

After "Total estimate" there are five columns including:
  • S - modulus of the scattering vector;
    J EXP - experimental intensity;
    ERROR - experimental error;
    J REG - regularized curve fitted to the experimental data;
    I REG - same as J REG1, but extrapolated to S=0, this is the Fourier transform of the p(r).
After these five columns you will find the distance distribution function including:
  • R - distance;
    P(R) - p(r);
    ERROR - computed error of p(r).
1J REG and I REG are same only if smearing was not used. Otherwise J REG is the smeared fit and I REG is de-smeared.

Aidee
Active member
Posts: 19
Joined: 2010.08.10 13:27
Location: UNIVERSITY OF MANCHESTER

#6 Post by Aidee » 2011.01.11 01:05

Thanks again for the reply. I have plotted logI vs s but the resulting pattern does not correspond to any of the shapes. Is this normal or did I miss anything?
Attachments
Gnom.JPG
Gnom.JPG (32.43 KiB) Viewed 27313 times

User avatar
AL
Administrator
Posts: 690
Joined: 2007.08.03 18:55
Location: EMBL Hamburg, Germany
Contact:

#7 Post by AL » 2011.01.11 13:01

Aidee wrote:Is this normal or did I miss anything?
Yes, you did. Above are just five examples of simple shapes. In reality the particle of interest can be of any shape - far more complex than a sphere or a rod.

Aidee
Active member
Posts: 19
Joined: 2010.08.10 13:27
Location: UNIVERSITY OF MANCHESTER

#8 Post by Aidee » 2011.01.11 14:20

Thank you so much. I guessed it will be complex too. The p(r) function is sort of bell-shaped and the Dammin runs come out with different shapes as well, mostly spheroids.

spark
Active member
Posts: 156
Joined: 2008.06.08 17:38

Re: GNOM output

#9 Post by spark » 2012.06.25 20:36

AL wrote:After "Total estimate" there are five columns including:
  • S - modulus of the scattering vector;
    J EXP - experimental intensity;
    ERROR - experimental error;
    J REG - regularized curve fitted to the experimental data;
    I REG - same as J REG, but extrapolated to S=0, this is the Fourier transform of the p(r).
After these five columns you will find the distance distribution function including:
  • R - distance;
    P(R) - p(r);
    ERROR - computed error of p(r).
Just what I need. Thanks Al :-)
Is s in this case really s or q? And what is the unit? Again the input unit?

User avatar
AL
Administrator
Posts: 690
Joined: 2007.08.03 18:55
Location: EMBL Hamburg, Germany
Contact:

Re: GNOM output

#10 Post by AL » 2012.06.26 12:13

You are welcome!
spark wrote:Is s in this case really s or q?
Eh... what is the difference?
s or q or Q denotes the modulus of the scattering vector. Usually s=q=4πsin(θ)/λ.
spark wrote:And what is the unit? Again the input unit?
Correct.
Technically you can choose the units in GNOM.

spark
Active member
Posts: 156
Joined: 2008.06.08 17:38

Re: Does my p(r) function look OK?

#11 Post by spark » 2012.06.26 12:25

Hi Al, thanks for the answer.

s and q are not the same for me. They differ by a factor of 2pi (definition from Roe's book).
This is why I get always confused. The scatering factor for me is always q :-) (not s as the definition of s is slightly different).
Thank you. I guessed yesterday night that it is probably nm^{-1} in my case. Thank you for the confirmation.

Best regards
Spark.

User avatar
AL
Administrator
Posts: 690
Joined: 2007.08.03 18:55
Location: EMBL Hamburg, Germany
Contact:

s vs. q

#12 Post by AL » 2012.06.26 13:02

spark wrote:s and q are not the same for me. They differ by a factor of 2pi (definition from Roe's book).
Could you please post the definition from Roe's book?
I guess q=2sin(θ)/λ is more common for polymer people. In solution scattering there are no strict rules, usually it is defined in every paper or presentation what letter is used for the modulus of the scattering vector and what exactly it means.

spark
Active member
Posts: 156
Joined: 2008.06.08 17:38

Re: s vs. q

#13 Post by spark » 2012.06.26 13:50

AL wrote:
spark wrote:s and q are not the same for me. They differ by a factor of 2pi (definition from Roe's book).
Could you please post the definition from Roe's book?
I guess q=2sin(θ)/λ is more common for polymer people. In solution scattering there are no strict rules, usually it is defined in every paper or presentation what letter is used for the modulus of the scattering vector and what exactly it means.
Yes, you are right. He defines s as s=2sin(θ)/λ and q=2*pi*s.
Sorry, if this was a stupid question, but I am always not sure :-)

Post Reply