After taking a look at the note regarding the hyperparemeters, I have a few questions. The tissue samples we're working with are liver tissue samples from adult and adolecent individuals and we're comparing three conditions, NASH1, NASH2, and NAFLD. After examining the tissue samples to determine the nuclei per spot, we've quantified that the nuclei per spot in the representative areas from each condition are very similar with values of 10 to 11 per spot. However the variance is large both between zonation and within tissue regardless of tissue disease state, for example, portal tracts on portal tract in portal inflammation range from 20 to 25 nuclei per spot whereas steatoic spots in NAFLD contain as few as 3 to 4 nuclei per spot. How can we surmount this problem of such a large variance of nuclei per spot between our samples? What would it look like to use a variable hyper parameter for n_cell_per_location? Please provide a code example. For our data should we run the model on the data on a per-condition basis?
Single cell reference data: number of cells, number of cell types, number of genes
The single cell reference data comes from the liver cell atlas and has 167598 cells X 13450 genes. The visium reference data also is from the liver cell atlas database and has 4333 cells X 11827 genes. Our experimental data set has 11602 cells and 8540 genes.
...
Single cell reference data: technology type (e.g. mix of 10X 3' and 5')
These samples are 10X 3' samples
...
Spatial data: number of locations numbers, technology type (e.g. Visium, ISS, Nanostring WTA)
The number of locations range from 640 to 1624 locations per sample with a total of 14 visium samples
...
Please use the template below to post a question to https://discourse.scverse.org/c/ecosytem/cell2location/.
Hello,
After taking a look at the note regarding the hyperparemeters, I have a few questions. The tissue samples we're working with are liver tissue samples from adult and adolecent individuals and we're comparing three conditions, NASH1, NASH2, and NAFLD. After examining the tissue samples to determine the nuclei per spot, we've quantified that the nuclei per spot in the representative areas from each condition are very similar with values of 10 to 11 per spot. However the variance is large both between zonation and within tissue regardless of tissue disease state, for example, portal tracts on portal tract in portal inflammation range from 20 to 25 nuclei per spot whereas steatoic spots in NAFLD contain as few as 3 to 4 nuclei per spot. How can we surmount this problem of such a large variance of nuclei per spot between our samples? What would it look like to use a variable hyper parameter for n_cell_per_location? Please provide a code example. For our data should we run the model on the data on a per-condition basis?
Single cell reference data: number of cells, number of cell types, number of genes
The single cell reference data comes from the liver cell atlas and has 167598 cells X 13450 genes. The visium reference data also is from the liver cell atlas database and has 4333 cells X 11827 genes. Our experimental data set has 11602 cells and 8540 genes. ...
Single cell reference data: technology type (e.g. mix of 10X 3' and 5')
These samples are 10X 3' samples ...
Spatial data: number of locations numbers, technology type (e.g. Visium, ISS, Nanostring WTA)
The number of locations range from 640 to 1624 locations per sample with a total of 14 visium samples ...