Alignment without histology image

[DruncBread]

Hi,

Thank you for your great work!

I have some questions after reading your paper and SpatialMETA tutorials though:

1. Is it possible to align ST and SM slices without histology image? Would you mind providing a tutorial?
2. Do you think SpatialMETA will work on other omics, like ATAC (peak matrix) and RNA (gene expression matrix) slices alignment?

[CHENyiru3]

Despite I am not the first author, but I think I know how to answer the second question: not recommended. The reason is that ATAC-seq may have different prior distribution, instead of current Gaussian distribution for metabolic. You can find out other tools that declare that they can do ATCT+RNA integration (maybe multiVI, MISO,... can do this)

[CHENyiru3]

For the first question, sorry I am not sure. If you don't have histology image, then you can't automatically align the spots between two slices. I assume that you can manually mapping the spots, and align them in space, and then skip the histology alignment step.

[DruncBread]

Hi, @CHENyiru3

Thank you for you advice!

I have another question, though:
I encountered a cuda out of memory error.

The author suggested that I should normalize the spatial coordinates first. I haven't tried it yet.

But this seems strange to me. Shouldn't spot coordinates be aligned to image? I think normalization would break this alignment.

It would be great if you could help me resolve this doubt!

Best regards

[RuonanTian]

You can try to run adata_sm.obsm['spatial'] = adata_sm.obsm['spatial_normalized'] before the alignment. If the cuda out of memory