CIBERSORT and TIMER

The LM22 gene signature matrix and CIBERSORT algorithm can estimate the relative proportions of 22 human immune cell phenotypes, including T cells, B cells, NK cells, macrophages, DC cells, mast cells, and granulocytes, in complex bulk tumor tissue (14). CIBERSORT adaptively selects genes from the input matrix to deconvolve a given mixture using linear support vector regression (SVR) based on the LM22 signature matrix. The LM22 was validated using external datasets of each cell subset, and CIBERSORT results were well-matched (93%) with the phenotypes of these datasets (14). The input matrix of reference gene expression signatures was prepared using the standard annotation file. The CIBERSORT algorithm runs in R with 100 permutations using the LM22 signature, and p < 0.05 was set as the cutoff for statistically significance. TIMER algorithm can estimate the abundances of six immune infiltrates and evaluate the correlation of gene expression with immune infiltration level (15).