Analyzing omics results from a single-cell genomics/single-cell spatial genomics experiment would reveal that there are these many cell clusters present/found. There will be cases, where the data and its analysis would "reveal" or declared that new cell subtypes have been observed.
At what point in the change of Gene expression of one category from other - would one call it as a different subtype OR State of cell??, and contrast it from a stochastic difference?
if n = total number of unique transcripts in tissue type or cell type - would change in the expression of one gene would be enough to call it a different subtype?
Or is there a totally different set of bioinformatic assumption (which I am missing) when defining cell-subtypes/states?
Usually we define a cell type based on marker genes from the literature. If there are multiple clusters belonging to one cell type you can go on and check for different subtypes and cell-states based on the literature. If that is not enought you should define them yourself based on genes and the functions behind it e.g. via GO-terms. If you really want to state that you found a new cell type/subclass you need to validate it.
But you need to be careful because as you know you can increase decrease the resolution of your dataset. If you have a too high resulution you might get multiple different clusters for one cell type which overestimates the heterogeneity and might have no biological meening.
I hope that can be helpful for you :)
- Badges
- Science method