Projecting spatial annotations to single cell

[tvegawaichman]

Hello!

I'm currently working with the MOSTA dataset (Mouse Stereo-seq), and I have two questions:

• Do you recommend using specific parameters to work with this kind of data?
• I'm interested in transferring the Spatial annotations from my spatial dataset to each single cell. I was thinking about calculating the cell x spot matrix, assigning the most likely spot, and determining the spatial domain from where that spot is assigned in my spatial data. However, there might be more sophisticated ways to achieve this! I would greatly appreciate your recommendations on how to do it.

Thank you very much!

[Hejin0701]

Hi @tvegawaichman ,

1. Using the default parameter should be fine for this kind of data.
2. instead of trying to find the spot with max probability, what I would recommend is to sum over the probability of the cells getting assigned to the spots of certain domain and compare between these probabilities. Then try to find the domain with max probability. If two domains shows comparable probability of getting assigned, it is possible that the cell is located at the boundary between the two domains.

[tvegawaichman]

I will try it! Thank you!