The expression profiles and clinical information of The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) were downloaded from the UCSC Xena (https://xenabrowser.net/datapages/) database. UCSC Xena is a cancer genomics data analysis platform that supports the visualization and analysis of a variety of omics data of cancer samples (18). The platform has built-in public data sets, including data from large cancer research projects such as TCGA and ICGC. The detailed sample sizes of tumour and normal tissues in the TCGA and GTEx databases (including tumour type, sample size of tumour tissues from TCGA, sample size of normal tissues from TCGA, and sample size of normal tissues from GTEx) are provided in Supplementary Table 1 .

The immune cell infiltration data of the TCGA database were downloaded from the Immune Cell Abundance Identifier (ImmuCellAI) (http://bioinfo.life.hust.edu.cn/ImmuCellAI#!/) and TIMER2 databases (http://timer.cistrome.org/). The immunotherapy datasets GSE135222 and GSE91061 were downloaded from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov).

The gene alterations of the TRP family, including mutation, copy number, methylation, were conducted using the Gene Set Cancer Analysis (GSCA) database (http://bioinfo.life.hust.edu.cn/GSCA/#/). GSCA is an integrated database for genomic and immunogenomic gene set cancer analysis.

The single-sample gene set enrichment analysis (ssGSEA) function of R v4.1.1 package ‘GSVA’ was used to calculate the TRP score of each patient in the TCGA cohort.

Univariate Cox regression (uniCox) analyses were performed to explore the effect of TRP score on the survival of patients, including overall survival (OS), disease-specific survival (DSS), disease-free interval (DFI), and progression-free interval (PFI) indicators, in pan-cancer using the R packages ‘survminer’ and ‘survival.’

To explore the biological functions of the TRP family and its role in various cancers, gene set variation analysis (GSVA) enrichment analysis was performed using the R package ‘GSVA’ to evaluate the correlation between TRP score and 50 HALLMARK pathways based on the MSigDB database (http://software.broadinstitute.org/gsea/msigdb/index.jsp).

The R package ‘ESTIMATE’ was used to calculate the stromal score, immune score, and tumour purity score of each patient in the TCGA cohort. The association between the TRP score and these scores was analysed. The TME-related pathways were obtained, and pathway scores were calculated according to a previous publication (19). We further analysed the association between TRP score and immune cell infiltration, immunomodulatory genes, MHC genes, and chemokine/chemokine receptors at the pan-cancer level. The visualization of all heatmaps in this step was conducted using the ‘ggplot2’ R package.

All data used in this article are presented as the mean ± standard deviation (SD). Differences between various groups were analysed and compared using the Student’s t-test. Statistical analysis was performed using R v4.1.1 (https://www.r-project.org/). The R packages ‘ggplot2’ and ‘ggpubr’ were used for the statistics of all histograms. Pearson correlation coefficient was used in all correlation analyses. P < 0.05 was considered statistically significant and indicated as follow: *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

We first explored the differential expression of 28 TRP family genes based on the TCGA and GTEx databases in 33 tumour types. The sample size for the cancer types is listed in Supplementary Table 1 . As shown in Figure 1A , 28 TRP family genes were commonly differentially expressed in 33 tumour types. We further assessed the correlation between TRP family genes based on TCGA pan-cancer data ( Figure 1B ). Additionally, we performed a uniCox analysis of each gene in the 33 tumours ( Figure 2A ). The results indicated that most TRP family genes were risk factors in tumours. We calculated the risk scores (i.e., the number of tumours in which genes are risk factors minus the number of tumours in which genes are protective factors) and found that TRPM4 was the most significant risk factor ( Figure 2B ). High expression of TRPM4 predicted worse overall survival in lower-grade glioma (LGG), uveal melanoma (UVM), kidney renal clear cell carcinoma (KIRC), pancreatic adenocarcinoma (PAAD), mesothelioma (MESO), and thyroid carcinoma (THCA).

We further explored the gene alterations of the TRP family and their association with the mRNA levels. We comprehensively described the genetic alteration numbers of TRP family genes, including mutation, fusion, amplification, homozygous deletion, and multiple alterations, in the TCGA cohort ( Figure 3 ). We described the variant classification, variant type, single-nucleotide variant (SNV) class, variants per sample, variant classification summary, and top10 mutated genes ( Figure 4A ). For the SNV percentage, the total deleterious mutation percentage (i.e., the number of samples with at least one deleterious mutation site/the number of samples with SNV mutation data) showed the highest deleterious mutation frequency of TRPM6 in skin cutaneous melanoma (SKCM) ( Figure 4B ). In addition, TRPM6 also had the highest mutation frequency (15%) among TRP family genes in pan-cancer ( Figure 4C ).

Next, we described the methylation information of TRP family genes in pan-cancer. Results indicated that the methylation levels of TRP family genes were general negatively correlated with mRNA levels ( Figure 5A ). Subsequently, we assessed the methylation status of the TRP family genes and found that the methylation levels of TRP family genes differed between cancer types ( Figure 5B ). We further analysed the copy number variants (CNVs) of TRP family genes. The proportion of different types of CNV (including heterozygous amplification, heterozygous deletion, homozygous amplification, and homozygous deletion) of each gene in pan-cancer was shown ( Figure 6A ). The CNV of TRPC4AP was positively correlated with its mRNA level in 27 of 33 tumour types ( Figure 6B ). The CNV of homozygous or heterozygous amplification was positively correlated with mRNA level, and the CNV of homozygous or heterozygous deletion was negatively correlated with mRNA level ( Figures 6C, D ).

We conducted ssGSEA to estimate the TRP score across 33 tumour types in the TCGA cohort. The TRP score was the highest in SKCM and lowest in LGG ( Figure 7A ). In addition, the TRP score was higher in tumour tissues compared to the adjacent normal tissues in head and neck squamous cell carcinoma (HNSC), kidney chromophobe (KICH), KIRC, prostate adenocarcinoma (PRAD), and THCA ( Figures 7B–F ), while lower in colon adenocarcinoma (COAD), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), and lung squamous cell carcinoma (LUSC) ( Figures 7G–J ). As for the results of uniCox analysis: (1) For OS, TRP score was a risk factor in LGG, UVM, thymoma (THYM), and KIRC, whereas a protective factor in bladder urothelial carcinoma (BLCA) ( Figure 8A ); (2) For DSS, TRP score was a risk factor in LGG, UVM, and THYM, while a protective factor in BLCA and kidney renal papillary cell carcinoma (KIRP) ( Figure 8B ); (3) For DFI, TRP score was a risk factor in breast invasive carcinoma (BRCA), whereas a protective factor in LIHC, LUSC, and stomach adenocarcinoma (STAD) ( Figure 8C ); (4) For PFI, TRP score was a risk factor in LGG, UVM, glioblastoma multiforme (GBM), and THYM, while a protective factor in adrenocortical carcinoma (ACC), LIHC, and BLCA ( Figure 8D ).

To analyse the potential pathways affected by the TRP score, we performed a GSVA based on 50 HALLMARK pathways. The association between the TRP score and GSVA scores in pan-cancer is shown in Figure 9 . We observed that the TRP score was positively associated with many malignant pathways in pan-cancer, such as IL6–JAK–STAT3 signalling, interferon-gamma response, and inflammatory response. All these pathways were also closely associated with TIME.

We further explored the association between TRP, stromal, and immune scores in pan-cancer ( Figure 10A ). Results revealed that TRP score was positively associated with an immune score, stromal score, and ESTIMATE score. We further obtained and calculated TME-related pathways according to the published paper, including immune-related pathways, stromal-related pathways, and DNA repair-related pathways. The results indicated that the TRP score was closely associated with immune-related pathways, including immune checkpoint, CD8 T effector, and antigen processing machinery, in pan-cancer ( Figure 10B ).

The above results showed that the TRP score was highly correlated with the immune score. Therefore, we further explored the correlation of TRP score with immune cells in TME. Based on the ImmuCellAI database, we found that at the pan-cancer level, the TRP score is highly correlated with most immune cells, indicating an immune-activated TME ( Figure 11A ). The correlation of immune cell infiltration based on the TIMER2 database also illustrated that the TRP score had a significant positive correlation with immune-activated TME in pan-cancer ( Figure 11B ).

In addition, we observed that the TRP score was highly correlated with immune-activating genes ( Figure 12A ), chemokines ( Figure 12B ), chemokine receptors ( Figure 12C ), and MHC genes ( Figure 12D ) in most tumour types. The above results all indicated that patients with high TRP scores have elevated immune cell infiltration, which may have a positive impact on immunotherapy.

It was reported that patients with a high tumour mutation burden (TMB) or microsatellite instability (MSI) may be sensitive to immune checkpoint inhibitor (ICI) treatment (20–22). We found that the TRP score was associated with TMB in four cancer types and MSI in seven cancer types. TRP score was positively correlated with the TMB value in BRCA ( Figure 13A ) and MSI value in BRCA, HNSC, and uterine corpus endometrial carcinoma (UCEC) ( Figure 13B ). Based on these results, we suspected that patients with high TRP scores are sensitive to immunotherapy. To verify this hypothesis, we collected immunotherapy data and calculated the TRP score. Through Kaplan-Meier analysis, we found that patients undergoing ICI treatment with high TRP scores had better OS or progression-free survival (PFS). The percentage of responsive patients was higher in the high TRP score group ( Figures 14A–F ).