Closed lxhuang7 closed 3 years ago
"if.filter = FALSE" will force DeMixT algorithm to use all available genes to fit the deconvolution model, which is not recommended. We suggest to use "if.filter = TRUE" option, which will improve tumor proportion estimation. Currently, we have a built-in filter in the function DeMixT_S2 to filter out unidentifiable genes in terms of the deconvolved expression (tumor-only expression). Consequently, it is likely that some genes will be dropped since the deconvolved expression value will be unreliable.
Thank you very much for the explanation.
@lxhuang7 Hello! Sorry for bothering you. I am trying to identity a tool for tumor profile deconvolution and have ISOpure and DeMixT. I am curious what are the reasons for you to switch from ISOpure to DeMixT? Thanks!
Hi, Thanks for this wonderful tool. I'm trying to switch from ISOpure to DeMixT. One question though, can I extract tumor-only expression matrix for all genes in my admixed RNA-seq data by just setting if.filter = FALSE? Do I need to change other parameters accordingly? It is not very clear to me in the vignette. Thanks again!