search

daviddaiweizhang / istar

GNU General Public License v3.0
85 stars 12 forks source link

How to simulate Visium from Xenium? #8

Open Xiaohui-Jiang opened 7 months ago

Xiaohui-Jiang commented 7 months ago

Hi David, could you also share the code to simulate Visium from Xenium?

jw235 commented 1 month ago

I just changed the cnts file and put the Xenium cell position instead of the Visium spot positions.

DDDoGGie commented 2 weeks ago

I just changed the cnts file and put the Xenium cell position instead of the Visium spot positions.

Hi, I am facing the same problem and have made the same attempt as you! However, I got a worse result, even the shape of the slices from get_mask.py is different from the Figure1b in the article. Could you provide me with some guidance on the specific implementation?

DDDoGGie commented 2 weeks ago

Hi David, could you also share the code to simulate Visium from Xenium?

Hello, I encountered the same issue. How did you resolve it? Thanks in advance!

jw235 commented 1 day ago

I tried aligning the histology images with my Xenium data, but unfortunately, the alignment wasn't 100% accurate. Nevertheless, I proceeded and aligned them into a Zarr file. Using Import Spatial Data, I loaded the Zarr file and extracted the image data and XY positions, though the alignment was still slightly off. I then fed this as a count matrix (cnts.tsv->spot(different transcript names) /x/y) and loaded the image as he-raw.jpg in the same dimensions as the sample image from ISTAR. The run did work, but since the alignment isn't perfect, the results are not fully correct. I'll need to improve the alignment to get more accurate results. Let me know if you manage the alignment and the run.

DDDoGGie commented 19 hours ago

I tried aligning the histology images with my Xenium data, but unfortunately, the alignment wasn't 100% accurate. Nevertheless, I proceeded and aligned them into a Zarr file. Using Import Spatial Data, I loaded the Zarr file and extracted the image data and XY positions, though the alignment was still slightly off. I then fed this as a count matrix (cnts.tsv->spot(different transcript names) /x/y) and loaded the image as he-raw.jpg in the same dimensions as the sample image from ISTAR. The run did work, but since the alignment isn't perfect, the results are not fully correct. I'll need to improve the alignment to get more accurate results. Let me know if you manage the alignment and the run.

I have accomplished the alignment between the Xenium's gene count file (.h5) and the paired HE image (.ome.tif), I'm not sure if this is what you need??