Hi, all!
When using blob picker and template picker for particle picking, the picked particles are always off-centered. I’m wondering whether there are some parameters that can be tuned to make things more accurate? Currently the markers are always at the ridge of the particles and hard to be re-centered in the following classification processes.
If you use mis-centred classes for template picking, the resulting picks will be off-centre. With a subset of micrographs (50-100) use the blob picker and keep adjusting parameters until you find ones which center nicely, then apply to the whole dataset. Or, alternatively, use Topaz or cryolo. Or manually pick 200-500 particles across a dozen or so micrographs (adjust the particle and box size parameters so you know you’ve got them nicely centred), then extract, classify into 10-20 classes, pick the nicely centred ones and template pick from that. Templates don’t really have to look wonderful - they just have to look like something - as the low-pass filter will remove any sort of detail from them.
It’s fairly unusual to find a sample which feels all but impossible to get picked nicely (ignoring things like amyloids). For very oddly shaped complexes (like ATPase, where views differ in profile dramatically) once you’ve found some parameters which work well for one sort of view, it can be worth optimising for alternative views as well and double-picking. With some careful curation during Inspect Picks, you might find you don’t end up with any duplicate particles… although I would still advise running any dual- or multi-pick strategy through some stringent Duplicate Removal runs to save pain later . 
When using templates, tightening up the particle diameter a little (which affects the mask diameter) and increasing the picking distance from 0.5 to 0.7 or 0.8 can help as well.
4 Likes
Thank you for the details! I’ll try to tune the parameters to see which one works. 