[ome-users] .mrc file import into omero

[Dougherty, Matthew T.]

Some background and few comments about emdatabank.org and the mrc format.

EMdatabank is collaboration between the cryo-EM community, the PDB and EBI, going back to 2002 when a declaration was made to encourage publishers to require open access to datasets that authors reference in their submissions.

The EM Community Outreach History is here:
http://emdatabank.org/?q=node/68

The baseline metadata specification, comparable to the OME specification, can be found here:
http://emdatabank.org/?q=dictionary

EBI hosts a web page with ccp4 and mrc map format definitions: 
http://www.ebi.ac.uk/pdbe/docs/iims/3D_EM.html



The MRC format is arguably the oldest biological digital image format, and predates TIFF  by several years.

The MRC format, a/k/a map format,  was derived from the MRC software suite used in high resolution EM image processing, dating to ~1969. 
 
The file format was added to the suite around 1982 as an ad hoc format similar to the philosophy of the MRC suite.  

The xray community picked it up a few years later in their formal standards activity, adding their metadata and calling their format the CCP4 map format.

At this point there was a loose relationship between the EM community and the X-ray, a common intersection being electron crystallography; but generally the formats drifted apart.

Around 2000 there was an effort to harmonize the formats, which has generally held.

Each header is demarcated as 256 32-bit words (1024 bytes).
Within the header  there are a number of undefined words, 38-52 are defined as "future use".  

Within the ad hoc EM community the philosophy has been from the beginning these are not reserved and may be used for any purpose the user or software developer chooses.

A significant complication is that neither format establishes versioning information, so sorting things out can be a problematic and mysterious.

Another major difficulty is the EM ad hoc metadata addition of "origin" (presumably the lowest corner of the bounding box) at words 50-52, which differs from the CCP4 ad hoc definition for another variable.  This is a critical problem, because a number of visualization programs use this for graphics in the EM community.

Within the IMOD program, it uses a number of "future use" words for various purposes.


We are currently in discussion with OME/LOCI to resolve this problem, as well as a number of other image format problems, systemically.

Besides the metadata issues, there is the looming problem regarding the quantity of pixels, which can make the image intractable.  

For example in my lab we are planning for a 10k^2 CCD, which will result potentially in 512GB 3D images.

If any one has suggestions, comments or wants to know more, please contact me directly.

 
Matthew Dougherty
713-433-3849 
National Center for Macromolecular Imaging
Baylor College of Medicine/Houston Texas USA
=========================================================================
=========================================================================

>
> As has been mentioned, we are planning to support EM images in OMERO  
> in a project spanning the next 3 years, in a collaboration with the  
> EM DB at EBI.
> This includes building a complete data model for EM data (as we have  
> done for light microscopy) and implementing this in BioFormats, the  
> OMERO server and clients.
>
> As the OMERO team has little experience of EM to date, any help that  
> you can provide in this area would be most welcome!

Quentin is the guy here wgho is involved and interested in image  
database solutions for electron microscopy,
and i wonder if he might like to help you define the EM meta data model.

He has a much better i dea than i do about that meta data is really  
important for EM.

cheers

Dan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openmicroscopy.org.uk/pipermail/ome-users/attachments/20090710/84fc5c28/attachment.html>