2-Deoxy-D-glucose br Survival oncoprint DNA methylation stat
3.2. Survival, oncoprint, DNA methylation status and mRNA co-occurrence analyses
≥5% of genetic alterations from the mRNAs selected from microarray (Fig. 2a). S100A3 from microarray and S100A7 from TCGA were having 3% of genetic alterations and in both cases amplification was observed (Fig. 3a and c) while CXCL10 and PKIA were altered in 2.1% patients (Fig. 3c). Besides oncoprint analysis, heatmap plots of the most sig-nificant DNA methylation are also shown for mRNAs selected from microarray (Fig. 3b) and TCGA (Fig. 3d). From the genes selected from TCGA, most of the genes in CC patients were found to be hy-permethylated (HBB, HBA2, SLC2A1, S100A7 and CXCL10) (Fig. 3d). DNA hypermethylation was found in seven mRNAs (CXCL10, ACTG2, MYH11, PMEP1, LAG3, CD247 and HLA-DPB1) selected from micro-array (Fig. 3b).
Out of the 703 pairs of interactions from 38 input mRNAs, S100A3-S100A7, HBA1-HBA2, CD247-S100A3 and CD247-S100A7 showed significant co-occurrence (p < 0.001) by mRNA co-occurrence ana-lysis. From microarray data, 10 genes (SLC2A1, CD24, ADM, RARRES3, NKG7, CD247, LAG3, LAMC2, PMEPA1 and KCNM1) were found as survival markers. Interestingly, mRNAs enriched in collagen assembly pathway in Reactome (LAMC2, MMP7, DST, COL7A1 and COL8A1) act together as a single combinatorial prognostic marker (p = 0.040) (Fig. 3e). From Cox regression analysis of mRNAs, SLC2A1 (Fig. 3f) and PMEPA1 (Fig. 3g) selected from TCGA and microarray, respectively, were also found as significant prognostic markers. Expression profile, corresponding survival status as well as p value of selected prognostic mRNA-based staging markers are provided in Supplementary Table S5. Combined gene 2-Deoxy-D-glucose survival plots of these mRNAs did not show any significant p value for colorectal cancer (p = 0.58) and head and neck cancer (p = 0.81) in TCGA data, indicating their specificity for CC (p = 0.02). Kaplan–Meier (KM) plots of combined gene expression for CC, colorectal cancer as well as head and neck cancer from TCGA are provided in Supplementary Fig. 1. When survival analysis of validated
Stage specific selected lncRNAs and mRNAs from TCGA validated in two GEO datasets (GSE52903 and GSE29817).
Stage mRNA lncRNA
Selected mRNAs from microarray.
interacting miRNA partners of these selected mRNAs and lncRNAs was performed, hsa-miR-150-5p (Fig. 3h) from the interacting partners of mRNAs and hsa-miR-188-3p (Fig. 3i), hsa-miR-101-3p and hsa-miR-378a-3p (Fig. 3j) from the interacting partners of lncRNAs were found as prognostic markers.
3.3. Identification of enriched miRNA families from mRNA-miRNA and lncRNA-miRNA interaction networks followed by stage specific mRNA-miRNA-lncRNA association network generation
Using miRNA family enrichment analysis for interacting partners of mRNAs selected from microarray with validated mRNA-miRNA inter-action network, we found that miR-30 (Yellow), miR-17 (Green), let-7 (Pink) and miR-130 (Cyan) families were enriched (Fig. 4a), while miR-let7 family was found to be significant with the upregulated mRNAs as input. miR-let7 family was enriched among the interacting miRNA partners of selected mRNAs obtained from TCGA (Fig. 4b) and enriched among the interacting miRNA partners of a selected lncRNA LINC00263 (Fig. 4c) too (p < 0.5). Venn diagram shown in Fig. 4d suggests the presence of seven miR -let-7 family candidate miRNAs (hsa-let-7a-5p, hsa-let-7b-5p, hsa-let-7d-5p, hsa-let-7e-5p, hsa-let-7f-5p, hsa-let-7 g-5p and hsa-let-7i-5p) which can interact with selected mRNAs in TCGA, microarray and lncRNAs. During mRNA-miRNA-lncRNA triad selection using common interacting miRNAs which could interact with selected mRNAs and lncRNAs, hsa-miR-766-3p and hsa-miR-769-3p were found to modulate SLC2A1, which is also known to bind with two lncRNAs, RP11-1299A16.3 and RP5-890E16.2 for stage I (Fig. 4e). In Stage III, hsa-miR-101-3p and hsa-miR-34a-5p were found to interact with PKIA, an mRNA, and RP11-38OL11.4, a lncRNA (Fig. 4f). r> 3.4. Pathway, GO and DSigDB analysis
Reactome 2016 analysis (with cut-off of Z Score < 1.95, p < 0.05 and combined score < 10) of mRNAs identified from microarray showed that Assembly of collagen fibrils and other multimeric structures_Homo sapiens_R-HSA-2022090, Collagen formation_Homo sapiens_R-HSA-1474290, Extracellular matrix organization_Homo sapiens_R-HSA-1474244, Type I hemidesmosome assembly_Homosapiens_R-HSA-446107 and Class B/2 (Secretin family receptors)_Homo sapiens_R-HSA-373080 were significant out of the best 10 enriched pathways. When GO mole-cular function (MF) 2017b was analysed for the mRNAs obtained from microarray (cut-off of Z score < 2, p < 0.05 and combined score < 10), only protein tyrosine kinase binding (GO: 19900782) was found to be significant. When all mRNAs selected from microarray and TCGA were considered for Reactome 2016, Assembly of collagen fibrils and other multimeric structures_Homo sapiens_R-HSA-2022090, Collagen formation_Homo sapiens_R-HSA-1474290 and Extracellular matrix organization_Homo sapiens_R-HSA-1474244 were also found to be sig-nificant with combined score > 20 (Fig. 5a). In GO biological process (BP), all the best 10 enriched biological processes were found to be sig-nificant, of which oxygen transport (GO:0015671) and epidermis devel-opment (GO:0008544) are worth mentioning (Fig. 5b).