SegGridExplore output

[Andrew Kehr]

Hi Carsten and others,

I'm brand new to EM and helical processing, but our lab is looking at a protein that I'm helping with and I was given some preliminary helical parameters to explore and see what spring thought of them.  After doing SegClassReconstruct I'm looking at the output in SegGridExplore and I'm getting a really broad distribution of best mean helical ccc over the range I searched for rise and rotation as well as an upward trend in rise as rotation increases.  This is clearly at odds with what the TMV dataset looks (nice and tight distribution with obvious periodic amplitude increases) like in the tutorial and I'm curious if this indicates something done poorly in earlier states of processing that I may be misunderstanding.

Any advice appreciated.

Andy

[Carsten Sachse]

Dear Andrew,

The peakedness of the cross correlation landscape depends completely on the features of the particular helical structure. In this sense, it is not easy to know what you should get. I suggest to redo the analysis in corresponding pitch and 'number of units per turn' space. Often the pitch can be estimated from the classes. When you sample in pitch/unit_number space you are moving towards a more helically defined space as shown in the tutorial. Later on you can explore the peaks by running reconstructions on these multiple correlation peaks that are expected.

Best wishes,

Carsten