Appion

There is a new version of FindEM available. It is attached to this issue.

Dear Bridget,

You kindly provided me with a letter of support for various software development projects.
We have now completed one of the projects, and I am ready to distribute the code for a new fast local correlation function. In the package attached, you will find an updated version of FindEM for particle picking, called FindEM2.

This new FindEM2 uses a custom mask for the template. This could be an oval or more accurately defined mask,
and will be better than the circular mask previously used, for non circular shaped particles.
The code is multi-threaded, and so it is also much faster than the old one.

We are releasing it under the GNU GPLv3, maintaining our copyright.

I am happy to help with integrating the code into your Appion or other software.

To cite it:  we have not submitted our manuscript yet, but I will keep you updated.

Best regards,
Alan

From Neil:

I already modified FindEM1 to be multithreaded and other speed improvements, but it might be interesting to have multiple masks. I'll have to read the paper.

Neil

From Dmitry:

I'd be interested to see how much of a qualitative difference a customized mask makes ... I'd guess that the manuscript for FindEM2 should have that information? I find that for most applications, the regular template-correlator works very well, assuming that you then take some time to filter your "false positive" picks with alignment and classification. I can imagine that there would be cases where a customized mask would be beneficial for discriminating subtle differences. The speed has not been a bottleneck for me personally, as I find that rough (30°) angular increments are usually sufficient, but it would play a role as you add more templates and/or use finer angular increments for searching. Perhaps let's put it on the agenda for the next Appion workshop? 
Dmitry