Clustering of miRseq mature expression: consensus hierarchical
Head and Neck Squamous Cell Carcinoma (Primary solid tumor)
16 April 2014  |  analyses__2014_04_16
Maintainer Information
Citation Information
Maintained by Hailei Zhang (Broad Institute)
Cite as Broad Institute TCGA Genome Data Analysis Center (2014): Clustering of miRseq mature expression: consensus hierarchical. Broad Institute of MIT and Harvard. doi:10.7908/C1BK19ZP
Overview
Introduction

This pipeline calculates clusters based on consensus hierarchical clustering with agglomerative average linkage[1],[2]. This pipeline has the following features:

  1. Classify samples into consensus clusters.

  2. Determine differentially expressed marker miRs for each subtype.

Summary

We filtered the data to 261 most variable miRs. Consensus average linkage hierarchical clustering of 475 samples and 261 miRs identified 3 subtypes with the stability of the clustering increasing for k = 2 to k = 8 and the average silhouette width calculation for selecting the robust clusters.

Results
Silhouette width of each sample in robust cluster

Figure 1.  Silhouette width was calculated and the average silhouette width for all samples within one cluster was shown below according to different clusters (left panel). The robust cluster was pointed out by blue symbol (left panel) and the silhouette width of each sample in robust cluster was shown on right panel.

miR expression patterns of molecular subtype

Figure 2.  Tumours were separated into 3 clusters on the basis of miR expression with 475 samples and 854 marker miRs. The color bar of the row indicates the marker miRs for the corresponding cluster.

Figure 3.  The miR expression heatmap with a standard hierarchical clustering for 475 samples and 261 most variable miRs.

Consensus and correlation matrix

Figure 4.  The consensus matrix after clustering shows 3 robust clusters with limited overlap between clusters.

Figure 5.  The correlation matrix also shows 3 robust clusters.

Samples assignment with silhouette width

Table 1.  Get Full Table List of samples with 3 subtypes and silhouette width.

SampleName cluster silhouetteValue
TCGA-BA-5559-01A-01R-1513-13 1 0.22
TCGA-BA-6873-01A-11R-1872-13 1 0.25
TCGA-BA-A4IF-01A-11R-A25Z-13 1 0.19
TCGA-BA-A6DB-01A-11R-A30I-13 1 0.35
TCGA-BA-A6DG-01A-21R-A30I-13 1 0.2
TCGA-BB-7872-01A-11R-2231-13 1 0.22
TCGA-BB-8596-01A-11R-2401-13 1 0.12
TCGA-C9-A480-01A-12R-A24J-13 1 0.23
TCGA-CN-4733-01A-02R-1872-13 1 0.33
TCGA-CN-4742-01A-02R-1513-13 1 0.1

Table 2.  Get Full Table List of samples belonging to each cluster in different k clusters.

TCGA_ID K.2 K.3 K.4 K.5 K.6 K.7 K.8
TCGA-BA-5559-01A-01R-1513-13 1 1 2 2 3 4 4
TCGA-BA-6873-01A-11R-1872-13 1 1 2 2 3 4 4
TCGA-BA-A4IF-01A-11R-A25Z-13 1 1 2 2 3 4 4
TCGA-BA-A6DB-01A-11R-A30I-13 1 1 2 2 3 4 4
TCGA-BA-A6DG-01A-21R-A30I-13 1 1 2 2 3 4 4
TCGA-BB-7872-01A-11R-2231-13 1 1 2 2 3 4 4
TCGA-BB-8596-01A-11R-2401-13 1 1 2 2 3 4 4
TCGA-C9-A480-01A-12R-A24J-13 1 1 2 2 3 4 4
TCGA-CN-4733-01A-02R-1872-13 1 1 2 2 3 4 4
TCGA-CN-4742-01A-02R-1513-13 1 1 2 2 3 4 4
Marker miRs of each subtype

Samples most representative of the clusters, hereby called core samples were identified based on positive silhouette width, indicating higher similarity to their own class than to any other class member. Core samples were used to select differentially expressed marker miRs for each subtype by comparing the subclass versus the other subclasses, using Student's t-test.

Table 3.  Get Full Table List of marker miRs with p<= 0.05 (The positive value of column difference means miR is upregulated in this subtype and vice versa).

Composite.Element.REF p difference q subclass
HSA-MIR-206|MIMAT0000462 9e-154 8 5.8e-151 1
HSA-MIR-133B|MIMAT0000770 3.5e-130 6.7 1.1e-127 1
HSA-MIR-133A|MIMAT0000427 1.1e-124 6.6 2.4e-122 1
HSA-MIR-1|MIMAT0000416 1.1e-119 6.5 1.8e-117 1
HSA-MIR-675-3P|MIMAT0006790 1.4e-24 2.1 1.5e-22 1
HSA-MIR-499A-5P|MIMAT0002870 1.4e-14 1.6 2.5e-13 1
HSA-MIR-1247-5P|MIMAT0005899 1.6e-16 1.5 3.9e-15 1
HSA-MIR-299-5P|MIMAT0002890 1.2e-28 1.4 1.6e-26 1
HSA-MIR-411-5P|MIMAT0003329 4.7e-21 1.3 2.5e-19 1
HSA-MIR-381-3P|MIMAT0000736 1e-22 1.3 9.6e-21 1
Methods & Data
Input

miRseq (MIMATs) of RPM value (reads per million reads aligned to miRBase mature) with log2 transformed was as the input data for the clustering

Consensus Hierarchical Clustering

Consensus Hierarchical clustering is a resampling-based clustering. It provides for a method to represent the consensus across multiple runs of a clustering algorithm and to assess the stability of the discovered clusters. To this end, perturbations of the original data are simulated by resampling techniques [1],[2].

Silhouette Width

Silhouette width is defined as the ratio of average distance of each sample to samples in the same cluster to the smallest distance to samples not in the same cluster. If silhouette width is close to 1, it means that sample is well clustered. If silhouette width is close to -1, it means that sample is misclassified [3][4].

Download Results

In addition to the links below, the full results of the analysis summarized in this report can also be downloaded programmatically using firehose_get, or interactively from either the Broad GDAC website or TCGA Data Coordination Center Portal.

References
[1] Bmonti, S., Tamayo, P., Mesirov, J. & Golub, T.R., Consensus Clustering: A Resampling-Based Method for Class Discovery and Visualization of Gene Expression Microarray Data, Machine Learning:91-118 (2003)
[3] Rousseeuw, P.J., Silhouettes: A graphical aid to the interpretation and validation of cluster analysis., J. Comput. Appl. Math. 20:53-65 (1987)
[5] RSEM