How can we use tpm values provided in GEO datasets for CNV or tumor cell prediction

diazlab / CONICS

CONICS: COpy-Number analysis In single-Cell RNA-Sequencing

73 stars 28 forks source link

How can we use tpm values provided in GEO datasets for CNV or tumor cell prediction #18

Open biobug16 opened 4 years ago

biobug16 commented 4 years ago

Hi @diazlab and @soerenmueller I want to predict tumor cells in Suva’s 2014 IDH wt Glioma dataset, for which I only have log2tpm values. How can we convert log2tpm to log2cpm or is there a way to use log2tpm itself with CONICSmat?

Thanks

soerenmueller commented 4 years ago

Hi @biobug16

here is some example code for the Patel et al dataset

a=read.table("Glioblastoma_expressed_genes.txt",sep="\t",header=T,row.names=1,check.names = F) a=((2^a)-1) a=log2(a/10+1)

biobug16 commented 4 years ago

Hi @soerenmueller Thank you for your reply. Can you please explain your answer a bit. I guess after reading, you first calculated the antilog for the log2tpm by using 2^a and then deducted 1 from it, followed you dividing it by 10 and then again you did log2 of it.

May be I am missing something, but are you sure it will be equivalent to log2CPM?

Correct me if I am getting it wrong.

Thanks again!