Multi-omic and multi-view clustering algorithms: review and cancer benchmark - PubMed (original) (raw)

Review

Multi-omic and multi-view clustering algorithms: review and cancer benchmark

Nimrod Rappoport et al. Nucleic Acids Res. 2018.

Erratum in

Abstract

Recent high throughput experimental methods have been used to collect large biomedical omics datasets. Clustering of single omic datasets has proven invaluable for biological and medical research. The decreasing cost and development of additional high throughput methods now enable measurement of multi-omic data. Clustering multi-omic data has the potential to reveal further systems-level insights, but raises computational and biological challenges. Here, we review algorithms for multi-omics clustering, and discuss key issues in applying these algorithms. Our review covers methods developed specifically for omic data as well as generic multi-view methods developed in the machine learning community for joint clustering of multiple data types. In addition, using cancer data from TCGA, we perform an extensive benchmark spanning ten different cancer types, providing the first systematic comparison of leading multi-omics and multi-view clustering algorithms. The results highlight key issues regarding the use of single- versus multi-omics, the choice of clustering strategy, the power of generic multi-view methods and the use of approximated p-values for gauging solution quality. Due to the growing use of multi-omics data, we expect these issues to be important for future progress in the field.

PubMed Disclaimer

Figures

Figure 1.

Figure 1.

Overview of multi-omics clustering approaches.

Figure 2.

Figure 2.

Performance of the algorithms on ten multi-omics cancer datasets. For each plot, the x-axis measures the differential survival between clusters (–log10 of logrank’s test _P_-value), and the y-axis is the number of clinical parameters enriched in the clusters. Red vertical lines indicate the threshold for significantly different survival (_P_-value ≤ 0.05)

Figure 3.

Figure 3.

Mean performance of the algorithms on ten multi-omics cancer datasets. The x-axis measures the differential survival between clusters (mean –log10 of logrank’s test _P_-value), and the y-axis is the mean number of clinical parameters enriched in the clusters.

Figure 4.

Figure 4.

Summarized performance of the algorithms across ten cancer datasets. For each plot, the x-axis measures the total differential prognosis between clusters (sum across all datasets of –log10 of logrank’s test _P_-value), and the y-axis is the total number of clinical parameters enriched in the clusters across all cancer types. (A–C) Results for single-omic datasets. (D) Results when each method uses the single omic that achieves the highest significance in survival. (E) Same with respect to enrichment of clinical labels.

Similar articles

Cited by

References

    1. Goodwin S., McPherson J.D., McCombie W.R.. Coming of age: ten years of next-generation sequencing technologies. Nat. Rev. Genet. 2016; 17:333–351. - PMC - PubMed
    1. Ozsolak F., Milos P.M.. RNA sequencing: advances, challenges and opportunities. Nat. Rev. Genet. 2011; 12:87–98. - PMC - PubMed
    1. Allison D.B., Cui X., Page G.P., Sabripour M.. Microarray data analysis: From disarray to consolidation and consensus. Nat. Rev. Genet. 2006; 7:55–65. - PubMed
    1. Yong W.-S., Hsu F.-M., Chen P.-Y.. Profiling genome-wide DNA methylation. Epigenet. Chromatin. 2016; 9:26 - PMC - PubMed
    1. Jain A.K., Murty M.N., Flynn P.J.. Data clustering: a review. ACM Comput. Surv. 1999; 31:264–323.

Publication types

MeSH terms

LinkOut - more resources