Open lila167 opened 4 years ago
Hi,
Not sure what you mean. What do you mean a "different reference profile"? It is not straightforward in xCell to use a different reference and requires development.
Regarding genes - xCell signatures are based on a list of 10K genes that were common across the 6 sources uses for the xCell development. Since there are other protocols with other sets of genes, at least 5K of them should be available in the input data, but more is better for more reliable results.
Best, Dvir
Thank you very much @dviraran for your quick answer. Indeed I have a pbmc reference profile (sorted bulks) with slightly different cell types than yours. I want to use your gene list with that reference profile. Do you think it is enough to filter those 10K genes and perform the regression? Or does your method score the genes before performing deconvolution?
Sorry, still confused about what you are trying to do. If you want to create a new reference, with spillover matrices, it will require going over the dev_scripts and the method in the paper.
If you are trying to run xCell on your data, it will internally choose those 10K genes and you don't need to do anything. You can find the list of genes in xCell.data$genes.
@dviraran I have a reference profile containing (gene x cell_type) which has more broader cell types than yours. It only has 6 major blood cell types instead of 64 in xCell. Now I want to use this reference profile for deconvolution using your method + genes. From your explanation, this is not possible I believe. Thanks
Yup. Hopefully one day we will create a customizable xCell..
Hi @dviraran,
I would like to use xCell for deconvoluting some blood samples with a different reference profile. I am wondering if you are using all the 5079 genes listed in "13059_2017_1349_MOESM1_ESM.xlsx" in your tool?
Thanks alot.