Integrative clustering of high-dimensional data with joint and individual clusters

Abstract

When measuring a range of genomic, epigenomic, and transcriptomic variables for the same tissue sample, an integrative approach to analysis can strengthen inference and lead to new insights. This is also the case when clustering patient samples, and several integrative cluster procedures have been proposed. Common for these methodologies is the restriction to a joint cluster structure, equal in all data layers. We instead present a clustering extension of the Joint and Individual Variance Explained algorithm (JIVE), Joint and Individual Clustering (JIC), enabling the construction of both joint and data type-specific clusters simultaneously. The procedure builds on the connection between k-means clustering and principal component analysis, and hence, the number of clusters can be determined by the number of relevant principal components. The proposed procedure is compared with iCluster, a method restricted to only joint clusters, and simulations show that JIC is clearly advantageous when both individual and joint clusters are present. The procedure is illustrated using gene expression and miRNA levels measured in breast cancer tissue from The Cancer Genome Atlas. The analysis suggests a division into three joint clusters common for both data types and two expression-specific clusters.

Integrative factorization of bidimensionally linked matrices

“…The joint and individual variations explained (JIVE) method (Lock et al ., ) decomposes joint and individual low‐rank signals across matrices via the decomposition . In the context of vertical integration, the joint and individual scores and have been applied to risk prediction (Kaplan and Lock, ) and clustering (Hellton and Thoresen, ) for high‐dimensional data. Several related techniques, such as AJIVE (Feng et al ., ) and SLIDE (Gaynanova and Li, ), have been proposed (Zhou et al ., ), as well as extensions that allow the adjustment of covariates (Li and Jung, ) or accommodate heterogeneity in the distributional assumptions for different sources (Li and Gaynanova, ; Zhu et al ., ).…”

Brain Age Prediction: Cortical and Subcortical Shape Covariation in the Developing Human Brain

“…Specifically, JIVE decomposes the total variance in data into three terms: joint variation across all seven shape measures, structured variation unique to individual shape measures, and residual noise that should be discarded from analyses. JIVE has been used in cancer studies to identify genetic variants across different data platforms (e.g., genotyping, mRNA and miRNA expression) (Hellton and Thoresen, 2016;O'Connell and Lock, 2016) and to identify the common variance between task-based fMRI connectivity and a wide range of behavioral measures (Yu et al, 2017). We hypothesized that integrative analysis of different cortical and subcortical shape measures can provide a more comprehensive picture of brain cortical and subcortical development from childhood to early adulthood, and thus may generate insights into the relationship between brain maturation and cognitive development.…”

Assisted gene expression‐based clustering with AWNCut

“…For example, Shen et al have proposed a penalized latent variable regression method to identify common latent variables for detecting disease subtypes with multiple omics datasets . A clustering procedure based on the Joint and Individual Variance Explained algorithm has been developed to find joint and data type‐specific clusters simultaneously . Another relevant study is the recent ANCut, which aims at more effectively clustering GEs with the assistance of regulators.…”

“…22 A clustering procedure based on the Joint and Individual Variance Explained algorithm has been developed to find joint and data type-specific clusters simultaneously. 23 Another relevant study is the recent ANCut, 24 which aims at more effectively clustering GEs with the assistance of regulators. For more detailed discussions, we refer to the literature.…”