CTF correction

[FridoF]

The CTF correction has several issues:

1. Signal near CTF zeros is set to zero, which should not be the case. As a result, the CTF corrected images have artifactual rings. The bug seem to be present already in v0.972.
2. The contrast is inverted, which differs from previous workflow. This may stem from different conventions of defocus (i.e., the word defocus already implying negative sign). in principle, we can agree on contrast inversion, but if so it should be a conscious choice.
3. Consistency of defocus determination in CTF plotter and CTF correction needs to be tested.
4. CTF correction needs to be verified on simple phantom tilt series with few simulated particles. Defocus can be constant at respective particle.

[FridoF]

updated doc of function gui/ctfCorrection.py problem of CTF 'rings' pushed to zero is either in function ModelCTF or in PhaseDeconv.

[gijsschot]

Dear Friedrich,

Are you in today? I do not see Mihajlo either.

In PhaseDeconv this happens.

Set sign of center pixel to (+)

otf[otf.shape[0]//2,otf.shape[1]//2] = 1

In ModelCTF

CHANGED sign for amplitude contrast to fix phase flip for low

frequencies -- SP 7.9.16

originally: ctf = -np.sqrt(1-A*2)np.sin(chi)-A*np.cos(chi)

Gijs

On Mon, Dec 9, 2019 at 11:57 AM FriedrichFoerster notifications@github.com wrote:

updated doc of function gui/ctfCorrection.py problem of CTF 'rings' pushed to zero is either in function ModelCTF or in PhaseDeconv.

— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub https://github.com/FridoF/PyTomPrivate/issues/14?email_source=notifications&email_token=AAG2BBLIK2Y3GJFBVVYE43TQXYQA7A5CNFSM4JXTERF2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEGIXOXQ#issuecomment-563181406, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAG2BBN3OFOGRSTADNOWCR3QXYQA7ANCNFSM4JXTERFQ .

[FridoF]

Yes, I am in and already documented the code and committed.

On Mon, 9 Dec 2019 at 13:40, gijsschot notifications@github.com wrote:

Dear Friedrich,

Are you in today? I do not see Mihajlo either.

In PhaseDeconv this happens.

Set sign of center pixel to (+)

otf[otf.shape[0]//2,otf.shape[1]//2] = 1

In ModelCTF

CHANGED sign for amplitude contrast to fix phase flip for low

frequencies -- SP 7.9.16

originally: ctf = -np.sqrt(1-A*2)np.sin(chi)-A*np.cos(chi)

Gijs

On Mon, Dec 9, 2019 at 11:57 AM FriedrichFoerster < notifications@github.com> wrote:

updated doc of function gui/ctfCorrection.py problem of CTF 'rings' pushed to zero is either in function ModelCTF or in PhaseDeconv.

— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub < https://github.com/FridoF/PyTomPrivate/issues/14?email_source=notifications&email_token=AAG2BBLIK2Y3GJFBVVYE43TQXYQA7A5CNFSM4JXTERF2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEGIXOXQ#issuecomment-563181406 , or unsubscribe < https://github.com/notifications/unsubscribe-auth/AAG2BBN3OFOGRSTADNOWCR3QXYQA7ANCNFSM4JXTERFQ

.

— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub https://github.com/FridoF/PyTomPrivate/issues/14?email_source=notifications&email_token=AMQ3BOKKL6LWZEBW2HONN23QXY4E3A5CNFSM4JXTERF2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEGJA53Q#issuecomment-563220206, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMQ3BOMGVBFR6GGN6VVMM5TQXY4E3ANCNFSM4JXTERFQ .

-- prof. dr. Friedrich Foerster

Bijvoet Center for Biomolecular Research Utrecht University Padualaan 8 3584 CH Utrecht The Netherlands

Tel: +31 30 253 4318

http://www.uu.nl/en/research/cryo-em

[FridoF]

update: a very possible problem for the kinks in the CTF corrected power spectra is that the symmetry of a real function is not maintained in the FFT of the modelled CTF. this could, for example, be due to a mis-centering of the CTF, i.e., the origin has the wrong coordinates (shifted by a single pixel).

[gijsschot]

We moved to CTF correction by IMOD. Its functionality has been validated using the mixed CTEM data thus this issue has been closed by implementing a new method