Auxiliary metadata file standards/procedures

The Single Cell Expression Atlas would like to show data from spatial omics experiments. In order to pick best standards for new data, I’m trying to wrangle a published dataset into the IDR style I see in git.

In a related topic I saw you mention pattern files and companion OME files. I wondered if you had some standards about these files for multiplex FISH or Visium? We use MAGE-TAB for single-cell RNA-seq and I see very similar files in IDR. I wondered if the attributes here are controlled or required?

I’m also interested to understand how you generate and validate these files? Do you do this from OME-TIFF automatically?

Our resource is mainly interested in the expression matrix (cellxgene), plus some extra metadata about targets and cells rather than the raw images (although we would like to link to them if possible). Do you have any examples with these matrices yet so I can see how to format highly multiplexed datasets?

Thanks for your help,


Hi @hewgreen

The screen files were introduced in IDR to workaround random collections of high-content screening files and should be avoided. When working with TIFFs, there’s nothing in a screen file that can’t be done with OME-XML.

There haven’t been any fields added to the OME Model specifically for capturing *-FISH like data. In terms of storing the positional metadata, both individual images with stage position information as well as a single well with multiple positions will accurately capture the information. In the GUI, the latter has the slight advantage of showing the relative position of the images to one another but at the cost of slightly misrepresenting the data as a plate rather than a tissue.

As you’d expect, some fields are controlled, some required, some both. A user-friendly representation of these is missing at the moment. We’ll get something together and ask you for feedback. :wink:

This is currently an IDR-specific pipeline, so the metadata isn’t validated as part of OME-TIFF. There’s a separate library, omero-metadata, used to parse the files and attach the metadata to the already imported images in OMERO.

I don’t believe we have any submissions where we’ve received cellxgene output, but I’ll double check. Update: @fwong confirms that we don’t currently have any.


Thanks for the detailed answer Josh. I have a couple of followups.

There haven’t been any fields added to the OME Model specifically for capturing * -FISH like data

I was going to ask if you spotted any gaps in the metadata yet? But as you don’t have any of this data yet I suppose you haven’t. Do you have any plans to ingest this data type?
Maybe we could collaborate on getting the first one published?

IDR-specific pipeline

So is the manually curated MAGE-TAB just used to add metadata to the OME-TIFF?


Comparing OME-XML to e.g. the spacetx format, what’s primarily missing is the relationship between the overarching (tissue) sample and the individual close-up images.


Replace “OME-TIFF” with “OMERO” and then yes.


1 Like