Data Retrieval and Preprocessing

RNA sequencing data (FPKM value), mutation profiles, and clinical data for TCGA-KIRC were downloaded from the Genomic Data Commons (GDC, https://portal.gdc.cancer.gov/) using the R package TCGAbiolinks (17). The FPKM value was transformed into transcripts per kilobase million (TPM) value. After removing duplicated patients, we included 530 KIRC patients with full clinical information and 72 normal tissues for further analysis. The copy number variation (CNV) data, processed with the GISTIC algorithm, were downloaded from the UCSC Xena data portal (http://xena.ucsc.edu/). Somatic mutation data were analyzed using VarScan2 and used to calculate the tumor mutation burden (TMB). Microsatellite instability (MSI) data were collected from the supplementary files of Bonneville's study (18).

A KIRC cohort ({"type":"entrez-geo","attrs":{"text":"GSE22541","term_id":"22541"}}GSE22541) with detailed survival data and an RNA expression matrix was downloaded from GEO (https://www.ncbi.nlm.nih.gov/geo/). After removing 44 samples collected from pulmonary metastasis of KIRC, we included 24 samples collected from primary KIRC for further analysis. An immunotherapy cohort (PMID29301960) containing 33 KIRC patients was collected from the supplementary files of Miao's study (19). Based on the Creative Commons 3.0 License, an immunotherapy cohort (IMvigor210) containing 348 bladder cancer patients was obtained from http://research-pub.gene.com/IMvigor210CoreBiologies/ (20). Another immunotherapy cohort of melanoma ({"type":"entrez-geo","attrs":{"text":"GSE78220","term_id":"78220"}}GSE78220) was downloaded from GEO. After removing one duplicated patient and one patient without follow-up time, we included 26 patients of {"type":"entrez-geo","attrs":{"text":"GSE78220","term_id":"78220"}}GSE78220 for further analysis.

Detailed information on these cohorts is summarized in Supplementary Table 1.