Xenium data compatibility?

[hartandrew]

This tool is very easy to download and to use with the Demo data. I am working with Xenium data and originally exported the cell_id, x, and y coordinated to a csv. The file is able to load and I can select the backbone points. However, when I attemp to "Unroll image" the process crashes. Im wondering if this is because there are CODEX-specific required fields? Or is ther another mistake I could be making. Thank you

[JiangyanYu]

Thanks for your interests on our tool! No, this tool is not specific for CODEX data, since it only takes the x-y position for the unrolling process. So it shall work for Xenium data

Are you able to load the data and visualize your image, and select the backbone points?

Best, Jiangyan

[hartandrew]

Thanks for getting back to me. Yes. I was able to load the data, visualize the image, and select backbone points. It only had a problem when I tried to use the backbone points to unroll the image. I am working on a remote server and the error is that it immediately "disconnects" from server. However, it unrolling the swiss roll works with the demodata when I select backbone points.

[JiangyanYu]

Glad to hear that the first QC and backbone selection worked fine, indicating the script shall be working fine. One possibility might be the large number of cells in your Xenium data. How many cells in there? For example, in the example file there is 10k cells only.

If there are millions of cells, the server might can not process them. Is it possible to try the tool by using less number of cells? If not, is it possible to get a higher performance server?

One minor note is you can save the selected points and upload it later into the app. Just to avoid selecting them every time :).

[JiangyanYu]

Hi @hartandrew , I was wondering whether sampling less cells solved your problem. Looking forward to your feedback!

Best, Jiangyan

[hartandrew]

Hi Jiangyan,

I wasn't able to solve the problem by sampling fewer cells. The issue could have been that I was deploying intestline on a remote server. Instead, I was able to use Intestline to pick backbone coordinates on either side of the swiss roll from center to outer edge. I exported the backbone coordinates and interpreted them as a polygon in R. I assigned each cell a relative value of "distance" by mapping them to the nearest intersect with the centerline of the polygon from root to tip. For my purposes this worked quite nicely.