Spider to frealign

Submitted by adesgeorges on Fri, 2010-02-26 17:11

Forums

Frealign

Hello Niko,

What is the difference is between gather_spi.com and gather_spi_boot.com?
And I have used spider to calculate my defocus instead of ctftilt. Is there still an easy way to retrieve my spider defocus values ? Or do you strongly recommend to use ctftilt instead of spider for defocus estimation?

By advance, many thanks,

niko Fri, 2010-02-26 22:11

The _boot script does not require shift or defocus values and reconstructions calculated with FREALIGN will not extend to high resolution. This script is meant to generate a first parameter file for a rough structure calculation. When you have angles, shift and defocus values, you should use gather_spi.com to make use of all the information.

I do not know how you determined the defocus using Spider. CTFTILT determines the mean defocus value and specimen tilt to calculate defocus values for each particle within an image according to its x,y position. For nominally untilted specimens, the observed tilt angle is usually only a few degrees (for example 3 deg), due to bending of the sample grid or small inaccuracies in the goniometer. This will only be important for reconstructions with a resolution of 10 Å or higher. Also, if you have astigmatism in your images, it will be important to measure this and set the three defocus parameters for each particle correctly. For the conventions used in FREALIGN, please see the publication by Mindell & Grigorieff, 2003.

If you are just starting with FREALIGN I would use the CTF values determined by Spider until you reach a limit in the resolution of your reconstruction with FREALIGN. If you did not determine the astigmatism you should enter the same value for both defocus values (the angle is arbitrary). Once your refinement with FREALIGN hits a limit you can redetermine the CTF values with CTFTILT and update them in your parameter file to see if this improved the resolution.

adesgeorges Sat, 2010-05-01 13:48

Hello Niko,

Still in Gather_spi, what do you call image # , negative # ans substack image # ?
Is the image # equal to the particle #? Is the position in the stack what is called substack image #?

When you define starting and last particle # , is it particle 1 and the total number of particles? Or the ID number of the first particle and the ID number of the last one?

Further down in the script, you set even and odd variable. This toggles the script between a loop that sets variables for tilted images and one that sets variables for untilted images. The two loops are almost identical except for the line "set nn" where you print the negative # or the neg # +1.
What is this used for? And what is the relationship with odd and even variables?

By advance, thank you.

niko Mon, 2010-05-03 13:18

I have updated the archive with a README file and additional scripts. The script gather_spi.com you were using was designed for Spider parameters derived from the random conical tilt reconstruction procedure. It may not be that useful to you. Nevertheless, I have now also included a script, frealignprep.com, that will generate some of the input files to gather_spi.com. This might help you in figuring out what the different parameters mean.

The script you might be more interested in is the convertm.com script I have now included. It takes a list of Spider parameters and converts them to Frealign parameters. I have also included an example Spider parameter file that you can use as a template to generate your own input to convertm.com.

Let me know if this is useful.

adesgeorges Tue, 2010-05-04 17:53

One quick question:
the first column in the frealign alignment file, particle nb, does it have to be consecutive numbers or does it just reflect the position of the particle in the stack?
For example, if the first particle I want to use is particle nb 5 in the stack, do I have the nb 5 in the first column first row?

By advance, many thanks.

adesgeorges Tue, 2010-05-04 18:36

It is ignored in convertm, but I guess it isn't in the refinement program? Does frealign take particle nb 5 as the 5th particle in the stack? And can the alignment file actually start by particle nb 5?

niko Tue, 2010-05-04 18:43

You are right, Frealign does not ignore the first column. Frealign uses this number as the number of the particle in the stack. However, I would not recommend skipping line numbers in the parameter file since I have not tested running Frealign this way for a while. If you need to run Frealign with missing lines in the parameter file, please run a test beforehand to see if it behaves the way you think. Alternatively, you can specify a starting particle number when you run Frealign.

Aurelien Mon, 2010-08-30 18:59

Hello,

Is there an easy formula to convert spider euler angles to the frealign convention?
Or is it exactly the same? I can't figure it out looking at convertm.

Best,

niko Mon, 2010-08-30 20:54

The angles should be exactly the same as those used for the Spider 3D reconstruction commands. However, the PSI angle (in-plane angle) is sometimes already applied to the images in a Spider script before 3D reconstruction. One could keep PSI = 0 and use the already rotated images for further processing with Frealign but this means that some resolution is lost due to interpolation errors. It is better to use the original windowed and normalized particles instead. In this case, when converting to Frealign, the PSI angles will generally not be zero. Their signs might have to be reversed, depending on the Spider script. The shifts are should also be the same but could have reversed signs, depending on how the Spider script is written.