Tumor-infiltrating immune cells

Question

[J.farideh76]

Hi, I have a list of genes and samples which got from TCGA in STAR count format and normalize them with edgeR and limma packages, now I want to assess Tumor-infiltrating immune cells, I try TIMER v2, but the input in this site should be TPM-normalized without log-transformation. Now how can I change my data to this format?

Answer 1

[Saeedeh Salehi]

Hi Farideh

It's very easy. You can use multiple solutions; however, I want to imply easiest one! You can use the function countoTpm in GeoTcgaData package and or count2Tpm function in IOBR package.

• Solution 1:

library(GeoTcgaData)

data(gene_cov)
countToTpm(countsMatrix, keyType = "SYMBOL", gene_cov)

• Solution2:

library(IOBR)
count2tpm(countMatrix, idType = "SYMBOL", org = "hsa", id = "id", gene_symbol = "symbol")

[Di Wu]

Hi Salehi, I also have a question about it. My data is microarrary expression, which program you suggest for infiltrating immune cells. And my data is Agilent two color microarray, I could use expr.MA to get the log-transformation expression, how can I transform the log-transformation microarray gene expression to without log-transformation. Thank you.

[Saeedeh Salehi]

Hi Di

Several methods, including TIMER 2, EPIC, MCP-counter, ImmuCellAI, DeconRNAseq, xCell, CIBERSORT, etc., have been developed for enumerating immune cells from microarray and RNAseq data. among them CIBERSORT and ImmuCellAI is appropriate for both microarray and RNAseq data. You can refer to related database and do analysis of interest.

[Di Wu]

Hi Dr. Saeedeh Salehi, many thanks you for your suggestions. The ImmuCellAI which could be available for mouse microarray, is did for infiltrating immune analyze. But they only give me the estimated results, and I dont know how to contrast with different groups. Should I just use T-test for each contrast which I interested? In addition, I think the accuracy of ImmuCellAI is not good enough. I give it my expression results which also did the DE analyse via limma. But it could give me similar results according to the functional enrichments results. Thanks again.