Hi,
I realized that using a 3D reference for filaments is not a good idea.
The autopicker projects the reference from all orientations, including
top views. This creates noises in the FOM map and makes filament tracking
even more difficult. Please use 2D class averages.
Best regards,
Takanori Nakane
> Dear all,
>
> I have a few questions regarding my Relion (3.0) Autopick job that I
> hope someone can help with. The job was carried out with Pick 2D helical
> segments, using a 3D reference generated in Spring, giving the following
> output star file header;
>
> data_
>
> loop_
> _rlnCoordinateX #1
> _rlnCoordinateY #2
> _rlnClassNumber #3
> _rlnAutopickFigureOfMerit #4
> _rlnHelicalTubeID #5
> _rlnAngleTiltPrior #6
> _rlnAnglePsiPrior #7
> _rlnHelicalTrackLength #8
> _rlnAnglePsiFlipRatio #9
>
> Rather than the HelicalTubeID numbers increasing sequentially, many of
> the HelicalTubeID numbers are skipped. Does this suggest that some
> filaments are 'found' but excluded on the basis of a setting given in
> the job setup (e.g. Minimum length) or something else? e.g.;
>
> ?3639.656845?? 396.164242????????? 115???? 2.952794 6??? 90.000000??
> 104.458071?? 176.775013???? 0.500000
> ?3635.292631?? 376.676277????????? 115???? 3.824260 6??? 90.000000???
> 99.964321?? 196.402246???? 0.500000
> ? 549.946390? 2950.144303?????????? 25???? 1.529888 23??? 90.000000??
> 151.066601???? 0.000000???? 0.500000
> ? 532.455421? 2940.475462????????? 115???? 2.023258 23??? 90.000000??
> 151.066601??? 19.641827???? 0.500000
> ? 514.964451? 2930.806621????????? 115???? 2.186421 23??? 90.000000??
> 151.066601??? 39.283654???? 0.500000
> ? 497.473482? 2921.137780?????????? 76???? 2.003261 23??? 90.000000??
> 151.066601??? 58.925481???? 0.500000
> ? 479.982512? 2911.468939????????? 115???? 2.178790 23??? 90.000000??
> 151.066601??? 78.567308???? 0.500000
> ? 462.491543? 2901.800098????????? 111???? 2.252027 23??? 90.000000??
> 151.066601??? 98.209134???? 0.500000
> ? 445.376324? 2891.484151????????? 111???? 2.165513 23??? 90.000000??
> 148.649027?? 117.849215???? 0.500000
> ? 428.308775? 2881.086110??????????? 9???? 1.640572 23??? 90.000000??
> 148.649027?? 137.491042???? 0.500000
> ?2754.733718? 1725.979348????????? 111???? 1.185446 50??? 90.000000???
> 53.629083?? 471.316624???? 0.500000
> ?2766.585329? 1709.887119?????????? 25???? 1.246539 50??? 90.000000???
> 53.629083?? 490.958451???? 0.500000
> ?2778.436939? 1693.794890????????? 111???? 1.454966 50??? 90.000000???
> 53.629083?? 510.600278???? 0.500000
> ?2790.288550? 1677.702662????????? 115???? 1.225620 50??? 90.000000???
> 53.629083?? 530.242105???? 0.500000
> ?2802.140161? 1661.610433?????????? 77???? 0.998342 50??? 90.000000???
> 53.629083?? 549.883931???? 0.500000
> ?1876.491081? 1319.906500????????? 102???? 0.358193 73??? 90.000000???
> 56.143737???? 0.000000???? 0.500000
> ?1887.625234? 1303.309780?????????? 86???? 0.188399 73??? 90.000000???
> 56.143737??? 19.641827???? 0.500000
> ?1898.759387? 1286.713059?????????? 93???? 0.362412 73??? 90.000000???
> 56.143737??? 39.283654???? 0.500000
> ?1909.893540? 1270.116339?????????? 43???? 0.186715 73??? 90.000000???
> 56.143737??? 58.925481???? 0.500000
>
> The HelicalTubeID jumps from 6 to 23 to 50 etc.
>
> Perhaps related to this, what is the relationship between the
> AutopickFigureOfMerit and the Picking threshold value given in the job
> setup (in this case 1.075), that determines which filaments are included
> in the star file? I find a wide range AutopickFigureOfMerit values (also
> negative values) assigned to the included filaments, and it's not clear
> to me what the threshold/deciding factor for inclusion is.
>
> In general the output includes many 'good filaments', but misses what
> appear to be very clear examples regardless of the threshold value used.
>
> Thank you very much!
>
> Charlotte
>
>
> ########################################################################
>
> To unsubscribe from the CCPEM list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1
>
########################################################################
To unsubscribe from the CCPEM list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1
|