Microarray support??

[siwelb]

I haven't seen any mention of this in the documentation, but I also wanted to check on the forums regarding this. Does OMERO support the creation/analysis of microarrays?

Thanks,
Brian

[wmoore]

Hi Brian,

There's no micro-array specific functionality in OMERO.
Can you tell us what sort of data you want to store and the kind of analysis you want to do (what software do you currently use)? E.g. Image data, tabular data, file data?

OMERO is quite adaptable to custom data and analysis, but may require a bit of work to hook up,

Cheers,

Will.

[siwelb]

Will -

Thanks for the reply. We are currently using an open-source program called TMAJ (http://tmaj.pathology.jhmi.edu/) for our microarray storage/analysis. It allows for uploading of a variety of image file (jpeg, tiff, png, etc.) that are stored in a grid-like layout (one image per array core). The analysis part of TMAJ is quite similar to OMERO in that it allows users to draw shapes in the image and make annotations.

It would be great if there was a way to customize OMERO to support this functionality - so far I have been pretty impressed with the features.

I had an additional question as well - does OMERO.web support direct linking to images? That is, a URL is provided for each image that when entered in a browser would take a user directly to that image for viewing/analysis?

Thanks,
~ Brian

[wmoore]

Hi,

How much of the TMAJ functionality would you be wanting from OMERO? We already support the High Content Screening "Screen - Plate - Well" hierarchy and have the infrastructure to do "Dynamic Fields". This is a work in progress but you can see an early demo of this - Download movie here:
http://trac.openmicroscopy.org.uk/ome/a ... ns.mov.zip

There is a lot of other Domain-specific functionality in TMAJ - Modelling of Patients, Blocks, Samples etc.
Can you manage without this, or handle it with some other type of annotations?

Seems like the tricky part may be the Import. Do you have the images in some sort of "Plate" format, or are they a collection of single images?
How big (pixels) are your individual images?
Currently we don't support drawing of shapes (ROIs) etc on "Big Images" (above 3000 x 3000 pixels) because of the technical limitations on our clients - Pan & Zoom etc, but we are working on it.
What sort of measurements do you require from these Shapes, or are they really just human readable annotations?

For linking to images, you can link direct to the image viewer - E.g. webclient/img_detail/22 but linking to an image in the Project Dataset hierarchy is still in development. See viewtopic.php?f=6&t=732&p=2652

Will.

[siwelb]

Will -

We wouldn't necessarily need the domain-specific functionality you mentioned (patient/sample data). Although we might be able to add some of our data using the metadata features using the OMERO.editor tool. The thing we'd really want (if it were possible without too much of a headache) is to be able to import the files and then view the images in a grid-like format.

The images we usually import into TMAJ are a collection of individual JPEG/PNG files - the software takes care of arranging the images into a grid. A normal image file would be approximately 2200 x 1900 pixels. All annotations would be human readable.

~ Brian

[wmoore]

OK, I think I understand what you need.

Right now, it is possible to "arrange the images in a grid" in Insight, although this is just a display layout - the data are not saved in a grid.
However, if you name your images E.g. image_x_01_y_01.tiff, image_x_02_y_01.tiff etc. then you can set the number of columns in Insight so that the X and Y coordinates will be laid out across rows and columns. I'll attach screen shot of this, using images named with Z and T indexes. You enter the number of columns in a field at the top right of the thumbnail panel, "22" in the screen-shot (NB - this can be obscured sometimes when this panel is not very large).

To fit your needs precisely, you would need to use the High Content Screening model, and specify a "Plate" of "Wells", each with one or more images. But there is currently no way to get your images into this data structure.

We'll keep this in mind and see if we can work out the best way to do this.
I created a ticket http://trac.openmicroscopy.org.uk/ome/ticket/6281

Can you tell me if the rows and columns have any significance. E.g. columns are different tissue types, rows are different treatments. Or are they purely set out "at random"?

Will.

[siwelb]

The rows and columns can be from different specimens although they aren't always. The images at each row/column positions are set based on how the scientist set up the tissue micro-array.

TMAJ has an application that designs array blocks (specify how many rows/columns and then can assign a specific diagnosis for each array core in the array block). After the array block has been designed, the user opens an image import application that lets them select the specific array block, array slide and scan to associate with the group of images. In the past we have always named our image files using the format you described in your post above (something like y_01_x_01_spot.jpg) when importing into TMAJ although we have had to manually name the files since they didn't come from an array slide scanner.

The method you described (setting number of columns in Insight) sounds reasonable. Would this behavior persist for every user or would this need to be done whenever a user went to view the images?

~ Brian

[wmoore]

Hi Brian,

Setting the number of columns in Insight is purely for display - It's not saved at all, not even for the 1 user.

Do the various microarray scanners adopt any convention on naming their images? E.g. image_X_12_Y_10.tiff or image-y-10-x-12.jpg. Are they at least consistent for a given scanner / format (or can the user choose a different naming pattern each time?

I see TMAJ supports Ariol (Applied Imaging), Bliss (Bacus Labs), Acis (Clarient) & Scanscope (Aperio).

Is there typically a "master" file that contains all the info on the images (the image counts and positions for a given scan) or are the images by themselves? (I guess if you've manually acquired the images, without a scanner then you only have the images themselves).

Is the naming pattern you've chosen widely used?

The answer to these questions should allow us to "tell" our import library (BioFormats) how to recognize a set of tissue micro-array images, so that they are imported correctly into our Screen/Plate/Well hierarchy.

Cheers,

Will.

[siwelb]

Will -

In the past the files we have loaded didn't come from one of the supported scanners so I'm very familiar at all with the way that the various scanners name the exported images. According to TMAJ documentation, you can still rename the files to the y_01_x_01_spot.jpg format. It also mentions that ARIOL machines have a separate XML file that contain the mapping metadata for each image (how each image corresponds to X/Y coordinates).

I did a search for file name conventions for the various scanners, but only came up with one link for BLISS machines: http://genome-www.stanford.edu/TMA/nomenclature.shtml

I'm guessing that all scanners have different ways they name the individual image files. Because of this, it might be better to do something similar to TMAJ and just come up with a format that can be used regardless of the type of scanner.

~ Brian

[wmoore]

OK, thanks for the info.
I think we have all the requirements for now - Just a case of scheduling a time to look at this: Can't promise anything right now. Keep an eye on the ticket I mentioned above, http://trac.openmicroscopy.org.uk/ome/ticket/6281

Cheers,

Will.