Another possible incompatibility with SPIDER
Forums
I've come across an apparent inconsistency in the implementation of the CTF between SPIDER and CTFFIND/ CTFTILT with respect to how amplitude contrast is dealt with. One of the pages in the documentation on 3D reconstruction ( http://www.wadsworth.org/spider_doc/spider/docs/techs/recon/mr.html ) describes doing CTF determination with CTFFIND and subsequently correcting the CTF with SPIDER so this type of inconsistency may be a problem. According to the paper on CTFTILT (Mindell and Grigorieff, 2003, JSB 142, 334-347, a coefficient -SQRT(1-A_CTFTILT**2) goes in front of the sine term in the equation for the CTF and a coefficient -A_CTFTILT goes in front of the cosine term where A_CTFTILT is the amplitude contrast as used in CTFTILT. It appears that this is what is implemented in the code for both CTFTILT and CTFFIND (For typical values quoted in the literature for amplitude contrast - either for cryo or negative stain electron microscopy - these will both be negative.) On the other hand, in the equation for the CTF on the page at http://www.wadsworth.org/spider_doc/spider/docs/man/tf.html , the coefficient in front of the sine term is (1-A_SPIDER) and the coefficient in front of the cosine term is -A_SPIDER so the coefficient in front of the sine term will be positive for typical values of the amplitude contrast in SPIDER, A_SPIDER, and the coefficient in front of the cosine term will be negative. So presumably it would be problematic to correct the CTF in SPIDER using defocusses that were fit by CTFFIND with these differences in the equation for the CTF?
(My algebra may well be wrong but it looked like you would need to use a negative value for the amplitude contrast in SPIDER with a similar - but not identical - magnitude to the positive value for the amplitude contrast used in CTFFIND or CTFTILT and also use the contrast reversal option in the SPIDER TF C operation, in order to get the same result, which is responsible for the placement of the zeroes in the CTF, I think. There are other differences in the implementation of the CTF in TF C with respect to envelopes, etc but I think these just affect magnitudes rather where the sign reversals occur in the CTF),
William
There is certainly an
There is certainly an discrepancy between the definition on the Spider web page and the implementation in CTFFIND3 and CTFTILT. However, the implementation in CTFFIND3/CTFTILT follows that of the MRC 2D crystallography programs. The CTFFIND3/CTFTILT implementation also agrees with a recent review by Penczek (Image restoration in cryo-electron microscopy, Methods Enzymol. 2010; 482:35-72).
I think these discrepancies should not matter much and can probably be ignored.
I've just had a look at this
In reply to There is certainly an by niko
I've just had a look at this again and apparently the argument to the sine and cosine functions - i.e. the chi function in your paper - is multiplied by -1 in SPIDER which explains the discrepancy regarding the signs in front of the sine and cosine terms (i.e. it appears the equations really are equivalent, apart from how amplitude contrast is dealt with). It does seem though that if somebody used, for example, 0.07 (for cryo data) for the amplitude contrast in the different versions of the equation then they would get a slightly different result,
William