Supplementary MaterialsAdditional document 1: Code and data for teaching analysis in TCGA data

Supplementary MaterialsAdditional document 1: Code and data for teaching analysis in TCGA data. denotes tumor microsatellite instability (MSI) position. (TIFF 29531 kb) 40425_2018_472_MOESM4_ESM.tiff (29M) GUID:?88C52652-CBAA-474C-BCC0-E379C480D84D Extra document 5: Expression of tumor mismatch repair genes versus tumor mutation burden across histological subtypes of TCGA UCEC dataset. Each column displays data from an individual histological subtype in TCGA UCEC Xipamide dataset, and each row displays data from an individual gene. Color denotes tumor microsatellite instability (MSI) position. (TIFF 21093 kb) 40425_2018_472_MOESM5_ESM.tiff (21M) GUID:?315C8C30-B3C6-4957-B6C8-4BF1DE990281 Extra file 6: Gene arranged enrichment results. For many KEGG, Reactome, and Biocarta gene models, the percentage of genes that are up- and down-regulated with a FDR? ?0.05. (CSV 50 kb) 40425_2018_472_MOESM6_ESM.csv (51K) GUID:?C9916B42-8E98-4DAD-A9D6-56D9C3DA7D09 Additional file 7: Supplementary material regarding algorithm development and validation. (DOCX 30 kb) 40425_2018_472_MOESM7_ESM.docx (35K) GUID:?1486CB28-AA49-48BA-9754-0830F79C316E Additional file 8: Mismatch repair (MMR) Loss and Hypermutation Predictor scores plotted against each other across histological subtypes in TCGA COAD dataset. Curved lines show the decision boundaries corresponding, from top-left to bottom-right, to microsatellite instability (MSI) Predictor score colon adenocarcinoma, stomach adenocarcinoma, uterine corpus endometrial carcinoma Additional files 3, 4 and 5 display the results of Fig. ?Fig.11 stratified by histological subtypes. The observations of Fig. 1 hold across each cancers histological subtypes. Hypermutated tumors share common transcriptional patterns in colon, stomach, and endometrial cancers Approximately one third of the hypermutation or ultramutation events as measured by Xipamide next-generation sequencing in TCGA (a broader set than MSI-H tumors) cannot be detected by loss of MMR gene expression. In such cases, transcriptomic events downstream of MMRd might Xipamide enable detection of hypermutation independent of the expression levels of the classic MMR genes. In cancers where hypermutation has a common origin in MMRd, and possibly in CIMP, we hypothesized that hypermutated tumors would display common transcriptional patterns across tumor types. To evaluate whether broader expression patterns could predict tumor MSI and hypermutation status, we ran univariate linear models testing the association of hypermutation status with the expression levels of each gene in each of the 3 TCGA whole transcriptome RNA-Seq datasets considered. Genes with highly significant associations with tumor hypermutation status were abundant: a Benjamini-Hochberg false discovery rate (FDR)? ?0.05 was achieved by 7800 genes in Xipamide colon adenocarcinomas, 9337 genes in stomach adenocarcinomas, and 3848 genes in endometrial carcinomas. A number of these genes behaved similarly across all 3 cancer types: 420 genes had a FDR? ?0.05 and a positive association with tumor hypermutation status in all 3 datasets, and 672 genes had a FDR? ?0.05 and a negative association with tumor hypermutation status in all 3 cancer types (Fig.?2). Gene sets relating to DNA replication machinery and metabolism were highly enriched Mouse monoclonal to KSHV ORF26 for positive associations with hypermutation (Additional file 6). The results demonstrated that numerous genes display strong differential expression with tumor hypermutation status across all cancer types and suggest that a data-driven predictor of tumor hypermutation status could prove informative. Open in a separate window Fig. 2 Gene expression signature of hypermutation status in TCGA dataset. Volcano plots show genes associations with hypermutation for colon adenocarcinoma (COAD), stomach adenocarcinoma (STAD), and uterine corpus endometrial carcinoma (UCEC). Genes with a false discovery rate (FDR)? ?0.05 in COAD are colored orange and blue in all 3 panels based on the direction of the association with hypermutation in COAD. The genes utilized by the Hypermutation Predictor algorithm are highlighted in reddish colored (positive weights) and crimson (adverse weights) Book gene manifestation algorithms for predicting MMRd, Hypermutation, and MSI position In line with the above observations within the TCGA dataset, distinct gene manifestation algorithms had been qualified for predicting tumor Xipamide MMR Hypermutation and Reduction position, and mixed right into a sole MSI Predictor algorithm then. The MMR Reduction algorithm, informed from the outcomes of Fig. ?Fig.1,1, procedures lack of tumor manifestation for the 4 MMR genes (MLH1, MSH2, MSH6, and PMS2). The Hypermutation Predictor algorithm, educated by the outcomes of Fig. ?Fig.2,2, uses 10 genes indicated in hypermutated tumors to forecast a tumors hypermutation position differentially. Finally, to increase predictive value through the use of all available info, the MSI Predictor algorithm combines the MMR Reduction and Hypermutation Predictor ratings into a solitary score made to forecast tumor MSI position. The calculations and derivations of the algorithms are summarized below and described at length in Additional file 7. The MMR reduction algorithm for phoning tumor MSI position based on.