1 Introduction

1.1 Home

1.2 Lnc-Pathways

The results are displayed in order of P values.

1.3 Lnc-Cells

The results can be sorted according to different attributes by clicking on column names .

1.4 Lnc-Cancer

The results can be sorted according to different attributes by clicking on column names .

2. Method

2.1 ImmLnc: Identification of immune-related lncRNAs in cancer

To identify the potential lncRNA modulators of immune-related pathways, we proposed a computational method that integrated lncRNA and gene expression data. Briefly, all the coding genes were ranked based on their correlation in expression with a specific lncRNA. The ranked gene list was subjected to each immune-related pathways to explore whether the immune-genes are enriched in the top or bottom of the list. An lncRES score was calculated for each pair of lncRNA and pathway. This process was repeated for all combinations of lncRNA and immune-related pathways. Based on permutation test, all lncRNA-pathway pairs with significantly higher lncRES scores were identified in cancer.

(i) Identification of lncRNA correlated genes

For each lncRNA of interest, we firstly ranked all the coding genes based on their correlation in expression with this lncRNA. The expression of lncRNA i and gene j across tumor patients were defined as L(i)=(l_1,l_2,l_3,…,l_i,…,l_m) and G(j)=(g_1,g_2,g_3,…,g_j,…,g_m). The tumor purity scores across m patients are defined as P=(p_1,p_2,p_3,…,p_i,…,p_m). We firstly calculated the partial correlation coefficient (PCC) between expression of lncRNA i and gene j by considering the tumor purity as a co-variable:

Where R_LG, R_LP and R_GP are the correlation coefficient between the expression of lncRNA i and coding gene j, the expression of lncRNA i and tumor purity, the expression of gene j and tumor purity. In addition, we also obtained the p-value for the PCC and defined as P(ij). For each pair of lncRNA and gene, we calculated a rank score (RS) as follow:

All genes are ranked based on RS scores and then subjected to enrichment analysis.

(ii) LncRNA modulators of immune-related pathways

Motivated by the idea of gene set enrichment analysis (GSEA), we mapped the genes in each immune-related pathway to the ranked gene list. Next, we calculated the enrichment score (ES) based on the GSEA. In addition, a p-value was calculated for each pathway that includes N_I of genes.

Where ES_ik is the enrichment score between lncRNA i and immune pathway k, N is the number of genes in the ranked list, and N_I the number of genes in the specific immune pathway. P-values were adjusted by false discovery rate (FDR). Moreover, according to one of previous studies51, we combined the p-value and ES score to an lncRES score.

Thus, the lncRES scores range from -1 to 1. We identified the lncRNA-pathway pairs with the absolute lncRES scores greater than 0.995 and FDR less than 0.05 as significant ones.

2.2 Identification of lncRNAs with expression perturbation in cancer

We used two methods to identify differentially expressed lncRNAs in each cancer type. Here, we only considered the 17 cancer types with more than five normal samples. For lncRNAs with expression level 0 in less than 30% samples were subjected to t-test. LncRNAs with expression levels 0 in more than 30% samples are for on/off analysis. For each lncRNA, we determined in a binary fashion: on (expressed, FPKM>0), OFF (not expressed, FPKM=0). Fisher’s exact test was used to evaluate whether the distribution of samples are different. LncRNAs with false discovery rate (FDR) less than 0.01 were identified as differentially expressed lncRNAs. In order to evaluate whether the expression of immune-related lncRNAs are likely to be perturbed, we compared the proportion of differentially expressed lncRNAs with all lncRNAs using Fisher’s exact test. The odd ratios (ORs) and 95% confidence level were also calculated.

3. How to cite

Pan-cancer characterrization of immune-related IncRNAs identifies oncogenic biomarkers.