search

rajewsky-lab / openst

Open-ST: profile and analyze tissue transcriptomes in 3D with high resolution in your lab
https://rajewsky-lab.github.io/openst/
Other
73 stars 10 forks source link

Spacemake run crashes #57

Open varsha090597 opened 3 weeks ago

varsha090597 commented 3 weeks ago

Hi,

When I tried running the spacemake run command after initializing and configuring I get the following error which wasn't very intuitive. It would be very helpful if you could point me to what exactly is going wrong. Thanks!

Screenshot from 2024-08-27 12-03-38

danilexn commented 3 weeks ago

Hi!

Thanks for opening this issue! It is a bug related to the generation of plots for QC sheets and automated analysis (with R). We're exploring why this is the case (might have to do with some tiles having an unexpected number of spatial units).

If you run spacemake with the --keep-going flag, this will generate all files that are necessary for proceeding with the openst workflow (except these QC sheets, which can be anyway generated externally e.g. with scanpy)

varsha090597 commented 2 weeks ago

Thanks for your response! I tried running it with the --keep-going flag and it ran past the qc sheet creation step. But now it gives me a different error:

Screenshot 2024-08-30 at 11 48 33 AM Screenshot 2024-08-30 at 11 48 57 AM

It also says that pandoc version of 1.12.3 or higher is required but I already have version 2.4 installed in my conda environment so I do not understand why I get that error either.

nukappa commented 2 weeks ago

Hi @varsha090597, the first error is related to the stitching of individual tiles into a larger area and is likely related to your coordinate system file. Can you verify if the your coordinate system file you have contains the sample_ 2175 etc? If you use the one shipped with spacemake/openst then you have to modify it.

The second error occurs during the automated analysis, which is also not a critical step -- you can do it with scanpy for instance. It did find only 1 cluster in the data though, which might be not what you were expecting.

varsha090597 commented 2 weeks ago

Thank you! You mentioned checking if sample_2175 exists in my coordinate system file. I used the openst_coordinate_system.csv file that is shipped with spacemake and this is how it looks when I opened it: Screenshot from 2024-09-04 11-31-04 Could you point me to where exactly I need to modify this file to include sample_2175? There are multiple spatial barcode coordinate files that get generated from step 1 of the openst run with sample_2175 being one of them. Should they also be included in the coordinate system file?

nukappa commented 2 weeks ago

Yes, you basically need to modify the coordinate system file so that the rows match the names of the files you have. For instance, if the sample_2175 comes from, say, lane 1, then you'd need to modify that row in the coordinate system from fc_1_L1_tile_2175 to sample_2175. You should have gotten about ~3,750 tiles from 4 lanes from step 1 of the openst run.