Fig. 1

Adjustment RNA-Seq data based on RNA-Seq mixture modeling. a Simulation results from in-silico mixture of different purified immune cells with purified monocytes from dataset GSE60424 [51]. Deviation of TPM values from ground truth (unmixed sample) was quantified as the mean absolute error (MAE). Purple: uncorrected samples; green: corrected samples. Each dot represents one simulation with a random mixture percentage between 0 % and 50 %. Violin plots show the distribution of MAE values. See “Results” for description of dataset used. The algorithm was applied for estimation of contamination and data adjustment as described in Additional file 1. b Estimated TAM contamination of tumor samples used in the present study, based on RNA-Seq mixture modeling. c Estimated tumor cell contamination of TAM samples. Striped bars in (b) and (c) denote samples excluded from further analysis. d, e Effect of adjustment by RNA-Seq mixture modeling on marker gene expression (PAX8, CD163) in tumor cell samples. ori, original TPM values; adj, adjusted TPM