ENTROPY-BASED CLUSTER VALIDATION AND ESTIMATION OF THE NUMBER OF CLUSTERS IN GENE EXPRESSION DATA

2012 ◽  
Vol 10 (05) ◽  
pp. 1250011
Author(s):  
NATALIA NOVOSELOVA ◽  
IGOR TOM
Keyword(s):  
Gene Expression ◽  
Gene Expression Data ◽  
Clustering Algorithm ◽  
Selection Procedure ◽  
Biological Knowledge ◽  
Consensus Clustering ◽  
Expression Data ◽  
Cluster Validation ◽  
Number Of Clusters ◽  
Validity Measure

Many external and internal validity measures have been proposed in order to estimate the number of clusters in gene expression data but as a rule they do not consider the analysis of the stability of the groupings produced by a clustering algorithm. Based on the approach assessing the predictive power or stability of a partitioning, we propose the new measure of cluster validation and the selection procedure to determine the suitable number of clusters. The validity measure is based on the estimation of the "clearness" of the consensus matrix, which is the result of a resampling clustering scheme or consensus clustering. According to the proposed selection procedure the stable clustering result is determined with the reference to the validity measure for the null hypothesis encoding for the absence of clusters. The final number of clusters is selected by analyzing the distance between the validity plots for initial and permutated data sets. We applied the selection procedure to estimate the clustering results on several datasets. As a result the proposed procedure produced an accurate and robust estimate of the number of clusters, which are in agreement with the biological knowledge and gold standards of cluster quality.

scholarly journals Machaon CVE: cluster validation for gene expression data

2003 ◽  
Vol 19 (18) ◽  
pp. 2494-2495 ◽  
Author(s):  
N. Bolshakova ◽  
F. Azuaje
Keyword(s):  
Gene Expression ◽  
Gene Expression Data ◽  
Expression Data ◽  
Cluster Validation

scholarly journals Combinatorial analysis of Saccharomyces cerevisiae regulatory elements

2019 ◽  
Author(s):  
N. Dhillon ◽  
R. Shelansky ◽  
B. Townshend ◽  
M. Jain ◽  
H. Boeger ◽  
...  
Keyword(s):  
Gene Expression ◽  
Saccharomyces Cerevisiae ◽  
Specific Interaction ◽  
Principal Component ◽  
Regulatory Elements ◽  
Untranslated Regions ◽  
Core Promoters ◽  
Regulatory Circuits ◽  
Expression Cluster ◽  
Multiple Levels

AbstractGene expression in Saccharomyces cerevisiae is regulated at multiple levels. Genomic and epigenomic mapping of transcription factors and chromatin components has led to the definition and delineation of various regulatory elements. Enhancers, promoters, 5’ untranslated regions (5’UTR) and transcription terminators/3’ untranslated regions (3’UTR) have all been defined. However, the specific contributions of each of these features as part of a regulatory unit and the functional communications between these regulatory elements remains under explored.We built a combinatorial library of 26 different enhancers, core promoters, 5’UTRs and transcription terminators/3’UTRs. This library was analyzed with respect to gene expression in order to better understand the interactions between different regulatory elements. In the process we developed new methods to estimate the contribution of individual regulatory parts from just a few simple measurements. Our data show that different pairs of regulatory parts follow specific interaction rules affecting overall activity either positively or negatively. We find that while enhancers are the initiators of gene activity, core promoters modulate the levels of enhancer mediated expression. Cluster analysis based on expression show that TATA-box containing core promoters appear to increase enhancer-driven transcription to a greater extent than TATA-less promoters. Principal component analysis highlight outliers and suggest differences in mechanisms of regulation. These results provide a system to characterize regulatory elements and use these elements in the design of synthetic regulatory circuits.

Independent Validation of Prognostic Relevance of a Previously Reported Gene-Expression Signature in Acute Myeloid Leukemia (AML) with Normal Cytogenetics (NC): A Cancer and Leukemia Group B (CALGB) Study.

Blood ◽  
2005 ◽  
Vol 106 (11) ◽  
pp. 755-755 ◽  
Author(s):  
Guido Marcucci ◽  
M. D. Radmacher ◽  
A. S. Ruppert ◽  
K. Mrózek ◽  
J. E. Kolitz ◽  
...  
Keyword(s):  
Gene Expression ◽  
Gene Expression Signature ◽  
Independent Set ◽  
Risk Category ◽  
Cluster Membership ◽  
Expression Signature ◽  
Expression Cluster ◽  
Outcome Group ◽  
Normal Cytogenetics ◽  
Group B

Abstract The heterogeneous clinical outcome of NC AML patients (pts), who constitute about 45% of AML pts and are assigned to an intermediate-risk category, likely reflects a biologic diversity. Although mutations in FLT3, MLL, CEBPA and NPM genes and overexpression of BAALC at diagnosis can stratify prognostically NC AML, novel approaches may improve the predictive value of these markers. cDNA microarrays were used recently to gain insight into the biologic and clinical heterogeneity of NC AML. Bullinger et al. (NEJM2004;350:1605) reported a set of 133 genes represented by 149 cDNAs (Suppl. Table 6, www.ncbi.nlm.nih.gov/geo/, accession #GSE425) that separated NC AML into 2 cluster-based groups with significant differences in overall survival (OS) (Bullinger et al. Fig. 4D), albeit the number of pts analyzed was small. Hence, we sought to validate this signature in a larger, independent set of adults (n=64), <60 years, with primary NC AML, treated on CALGB 9621. Pretreatment samples were analyzed using Affymetrix U133 plus 2.0 GeneChips (Affymetrix, Santa Clara, CA). Normalization to a baseline array with median overall brightness and probe-set level expression computation were performed with dChip version 1.3. The log (base 2) of expression values were exported to BRB-ArrayTools v3.2.3 for analysis. Of the 133 genes from Bullinger et al., 81, represented by 157 probe sets, were on the Affymetrix chip and were expressed in ≥ 25% of our specimens. Hierarchical clustering, using the 157 probe sets, resulted in cluster I, corresponding to the poor-, and cluster II, corresponding to the good-outcome group in Bullinger et al. With a 4.7-year median follow-up, cluster I pts had lower estimated 5-year DFS (24% vs. 60%; P=.001; Fig. 1A) and OS (28% vs. 64%; P=.001; Fig. 1B) rates than cluster II pts. The 2 groups had similar complete remission rates (81% vs. 89%). With regard to pretreatment features, cluster I pts were younger (P=.06), had lower % skin infiltrates (P=.03), higher % marrow blasts (P=.06), different distribution of FAB subtypes (P=.04) with a high % M4, and higher incidence (67% vs. 12%) of FLT3 internal tandem duplication (ITD) (P<.001). A multivariable model including gene-expression cluster membership, FLT3 ITD status by Genescan analysis, and their interaction was fit to determine if the gene-expression signature added any information once accounting for FLT3 ITD status. A significant interaction of the gene-expression signature and FLT3 status was observed for DFS (P=.009) and OS (P=.01), where the gene-expression signature predicted DFS (P=.02) and OS (P=.02) only in pts with wild-type (wt) FLT3 but not in those with FLT3 ITD, who had adverse outcome irrespective of the gene-expression signature. Cluster I pts with wtFLT3 had >3x the risk of relapse and death than the corresponding cluster II pts. In conclusion, we independently validated for the first time the prognostic value of a gene-expression signature in NC AML. This expression profiling appears particularly useful in predicting outcome of pts lacking the prognostically adverse FLT3 ITD. Fig. 1A, Fig. 1B Fig. 1A, Fig. 1B.

Sign in / Sign up

Export Citation Format

Share Document