e2boxer

Contents

e2boxer

This page details the design and usage of e2boxer. Users will find the process of acquainting themselves with the software much easier if they take the time to read the contents of this page.

Getting Started

e2boxer can handle multiple images simultaneously. To get started load up several images using something similar to the command below

e2boxer.py images221?.mrc --gui --boxsize=96

Notice that the boxsize parameter is necessary. If you don't know it just take a guess and then change it later in the interface. Once you're in the interface tick the "Dynapix" option, select a few boxes, and see what happens.

Unfortunately a lot more information needs to be supplied here but that won't come for a few weeks yet... please email us for any queries if you are already using boxer and we will do our best to resolve your issues.

An early 3D concept of the e2boxer interfaces that existed for a short while. In the end we chose to stick with 2D

Tips And Tricks

Shrinking For Speed

A current development goal is to make the 'shrinking for speed' concept an easy to use feature of e2boxer. For now just do it this way.

Sometimes, and especially on low end machines, it is much faster to shrink your data by a factor of 2 (you could shrink by more than this but it may not be necessary) before boxing them. This practice was also common using boxer in eman1. To do this quickly from the command line you would do something like -

]$ mkdir Shrink2
]$ for i in `ls *.mrc`
>do
>e2proc2d.py $i Shrink2/$i --shrink=2
>done
]$

Then move into the new directory and run e2boxer as per normal -

]$ cd Shrink2
]$ e2boxer.py *.mrc --gui --boxsize=128
...

Once you're done and you've saved all of the .box (coordinate) files, use batchboxer from eman1 do box the original (big, high resolution images) using something like:

]$ for i in `ls *.mrc`
> do
> batchboxer input=../$i dbbox=... ouput=... scale=2
> done

And it's all done. The assumption is that you know how to use for loops from the command line... contact us if you need elaboration.

General Information

Correlation is the main tool for performing autoboxing in eman2. e2boxer only ever uses one reference to generate the correlation image - this reference is the translationally aligned, rotational average of the black boxed particles in e2boxer. In addition, autoboxing parameters are determined in an automated fashion. This information is saved in real time to a local database, and once you have established an accurate 'state' of the autoboxing method you can access the information from the command line to do autoboxing on arbitrarily large sets of images. However, the autoboxing routine is extremely fast and can be executed in real time. This means that if you are the kind of 'Boxer' who pays attention to detail and checks every frame, then you don't actually have to run autoboxing from the command line because autoboxing in the interface is fast enough in the first place.

When we first started writing boxer we thought that we should support the concept of many references just as boxer did in eman1. But after we got things working it seemed that the single template (rotational average) was good enough. It works better than you would expect because we use extra 'peak profile' information rather than just the threshold. We may yet add support for multiple references if users find that the current approach is not accurate enough, but thus far we are hearing no complaints.

For more general information see e2boxer FAQ.