Hi there. Thanks in advance for an easy-to-use and well documented tool, and any input you might be able to provide!
Was wondering if you had any thoughts on how applicable the method is to imaging-based spatial transcriptomics data? Specifically, there's a comment somewhere that things are "independent of normalization" as the ranking is done in a per-cell manner, however:
1) I am worried that rankings (where ties are resolved randomly?) could be off when the dynamic range is much smaller, e.g., counts are mostly 0 or 1 with few large values. My feeling is that things come down to "is a set of genes detected at large or not" rather than how strong expression levels are.
2) Also: Would you expect results to differ at all without any normalization? I.e., using raw counts? Asking because standard normalization procedures aren't applicable here/it is less straightforward to obtain values on an expression-like scale.
Hi there. Thanks in advance for an easy-to-use and well documented tool, and any input you might be able to provide!
Was wondering if you had any thoughts on how applicable the method is to imaging-based spatial transcriptomics data? Specifically, there's a comment somewhere that things are "independent of normalization" as the ranking is done in a per-cell manner, however:
1) I am worried that rankings (where ties are resolved randomly?) could be off when the dynamic range is much smaller, e.g., counts are mostly 0 or 1 with few large values. My feeling is that things come down to "is a set of genes detected at large or not" rather than how strong expression levels are.
2) Also: Would you expect results to differ at all without any normalization? I.e., using raw counts? Asking because standard normalization procedures aren't applicable here/it is less straightforward to obtain values on an expression-like scale.