Limits...
Patient-specific driver gene prediction and risk assessment through integrated network analysis of cancer omics profiles.

Bertrand D, Chng KR, Sherbaf FG, Kiesel A, Chia BK, Sia YY, Huang SK, Hoon DS, Liu ET, Hillmer A, Nagarajan N - Nucleic Acids Res. (2015)

Bottom Line: Extensive in silico and in vitro validation helped establish OncoIMPACT's robustness, improved precision over competing approaches and verifiable patient and cell line specific predictions (2/2 and 6/7 true positives and negatives, respectively).In particular, we computationally predicted and experimentally validated the gene TRIM24 as a putative novel amplified driver in a melanoma patient.We also provide the first demonstration that computationally derived driver mutation signatures can be overall superior to single gene and gene expression based signatures in enabling patient stratification and prognostication.

View Article: PubMed Central - PubMed

Affiliation: Computational and Systems Biology, Genome Institute of Singapore, Singapore 138672, Singapore.

Show MeSH

Related in: MedlinePlus

Tumor stratification using predicted driver gene profiles. (a) Heatmaps depicting consistency of clustering (fraction of bootstrap replicates in which patients clustered together) for predicted driver gene mutational profiles (binary 0–1 vectors) using NMF. (b) Survival profiles of glioblastoma and ovarian cancer patients stratified by consensus clustering in (a). (c) Box plots showing the distribution of P-values (log rank test) for survival profiles of random subsets of glioblastoma (sample size 275) and ovarian cancer (sample size 250) patients, clustered into the same number of groups using different gene signatures (OncoIMPACT predicted driver genes; DriverNET predicted driver genes; Randomly selected sets of genes of the same size as OncoIMPACT predicted drivers; Randomly selected single genes).
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4402507&req=5

Figure 5: Tumor stratification using predicted driver gene profiles. (a) Heatmaps depicting consistency of clustering (fraction of bootstrap replicates in which patients clustered together) for predicted driver gene mutational profiles (binary 0–1 vectors) using NMF. (b) Survival profiles of glioblastoma and ovarian cancer patients stratified by consensus clustering in (a). (c) Box plots showing the distribution of P-values (log rank test) for survival profiles of random subsets of glioblastoma (sample size 275) and ovarian cancer (sample size 250) patients, clustered into the same number of groups using different gene signatures (OncoIMPACT predicted driver genes; DriverNET predicted driver genes; Randomly selected sets of genes of the same size as OncoIMPACT predicted drivers; Randomly selected single genes).

Mentions: Patient-specific driver mutational profiles are potentially promising inputs for tumor stratification since by definition, they are likely causative events for carcinogenesis and metastasis. However, while the mutational status of selected single genes has been shown to be of value in various cancers (16,55–56), unsupervised stratification using whole-exome mutation profiles is significantly more challenging (29), and the use of a small, computationally derived driver gene list for this purpose has not been demonstrated before. As a first, pilot exploration of this concept, we tested the utility of OncoIMPACT's predictions for stratifying patients according to their survival outcomes. Specifically, we used unsupervised consensus clustering using NMF to cluster patient-specific driver mutational profiles. Despite the sparseness of mutational profiles and the use of only a subset of genes containing predicted driver mutations (307 and 183 genes for Glioblastoma and Ovarian Cancer, respectively), we obtained robust clustering of patients (Figure 5a). In addition, we found that most clusters are defined by a few key driver genes that are predominantly mutated in tumors belonging to that cluster and serve to distinguish them from tumors in other clusters (Figure 5a).


Patient-specific driver gene prediction and risk assessment through integrated network analysis of cancer omics profiles.

Bertrand D, Chng KR, Sherbaf FG, Kiesel A, Chia BK, Sia YY, Huang SK, Hoon DS, Liu ET, Hillmer A, Nagarajan N - Nucleic Acids Res. (2015)

Tumor stratification using predicted driver gene profiles. (a) Heatmaps depicting consistency of clustering (fraction of bootstrap replicates in which patients clustered together) for predicted driver gene mutational profiles (binary 0–1 vectors) using NMF. (b) Survival profiles of glioblastoma and ovarian cancer patients stratified by consensus clustering in (a). (c) Box plots showing the distribution of P-values (log rank test) for survival profiles of random subsets of glioblastoma (sample size 275) and ovarian cancer (sample size 250) patients, clustered into the same number of groups using different gene signatures (OncoIMPACT predicted driver genes; DriverNET predicted driver genes; Randomly selected sets of genes of the same size as OncoIMPACT predicted drivers; Randomly selected single genes).
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4402507&req=5

Figure 5: Tumor stratification using predicted driver gene profiles. (a) Heatmaps depicting consistency of clustering (fraction of bootstrap replicates in which patients clustered together) for predicted driver gene mutational profiles (binary 0–1 vectors) using NMF. (b) Survival profiles of glioblastoma and ovarian cancer patients stratified by consensus clustering in (a). (c) Box plots showing the distribution of P-values (log rank test) for survival profiles of random subsets of glioblastoma (sample size 275) and ovarian cancer (sample size 250) patients, clustered into the same number of groups using different gene signatures (OncoIMPACT predicted driver genes; DriverNET predicted driver genes; Randomly selected sets of genes of the same size as OncoIMPACT predicted drivers; Randomly selected single genes).
Mentions: Patient-specific driver mutational profiles are potentially promising inputs for tumor stratification since by definition, they are likely causative events for carcinogenesis and metastasis. However, while the mutational status of selected single genes has been shown to be of value in various cancers (16,55–56), unsupervised stratification using whole-exome mutation profiles is significantly more challenging (29), and the use of a small, computationally derived driver gene list for this purpose has not been demonstrated before. As a first, pilot exploration of this concept, we tested the utility of OncoIMPACT's predictions for stratifying patients according to their survival outcomes. Specifically, we used unsupervised consensus clustering using NMF to cluster patient-specific driver mutational profiles. Despite the sparseness of mutational profiles and the use of only a subset of genes containing predicted driver mutations (307 and 183 genes for Glioblastoma and Ovarian Cancer, respectively), we obtained robust clustering of patients (Figure 5a). In addition, we found that most clusters are defined by a few key driver genes that are predominantly mutated in tumors belonging to that cluster and serve to distinguish them from tumors in other clusters (Figure 5a).

Bottom Line: Extensive in silico and in vitro validation helped establish OncoIMPACT's robustness, improved precision over competing approaches and verifiable patient and cell line specific predictions (2/2 and 6/7 true positives and negatives, respectively).In particular, we computationally predicted and experimentally validated the gene TRIM24 as a putative novel amplified driver in a melanoma patient.We also provide the first demonstration that computationally derived driver mutation signatures can be overall superior to single gene and gene expression based signatures in enabling patient stratification and prognostication.

View Article: PubMed Central - PubMed

Affiliation: Computational and Systems Biology, Genome Institute of Singapore, Singapore 138672, Singapore.

Show MeSH
Related in: MedlinePlus