A mixture model-based strategy for selecting sets of genes in multiclass response microarray experiments

Broët, Philippe; Lewin, Arie Y.; Richardson, Sylvia; Dalmasso, Cyril; Magdelénat, H.

doi:10.1093/bioinformatics/bth285

Cited by 57 publications

(41 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our cross validation results show that with the proper estimation of missing values, the gene selection and classification accuracy can be significantly improved [12,14]. So, for the proof of concept, an alternative way is to test imputation methods by randomly removing values from the data and testing the impact on decision making techniques such as gene selections and classification.…”

Section: Classification Error Measurementioning

confidence: 94%

“…Alternatively, univariate algorithms are used, for example; t-test, signal to noise ratio [11], BSS/WSS [4], Significance Analysis of Microarray (SAM) [12] which are either made for binary class response or they consider each relevant gene individually which selects the genes which are highly correlated which it introduces redundancy [13]. The problem can be avoided if multivariate gene selection is applied to simultaneously consider multiple genes and class information, hence reducing redundancy of covariate genes and keeping the class discrimination intact.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Collateral Missing Value Estimation: Robust Missing Value Estimation for Consequent Microarray Data Processing

Sehgal

Gondal

Dooley

2005

AI 2005: Advances in Artificial Intelligence

View full text Add to dashboard Cite

Abstract. Microarrays have unique ability to probe thousands of genes at a time that makes it a useful tool for variety of applications, ranging from diagnosis to drug discovery. However, data generated by microarrays often contains multiple missing gene expressions that affect the subsequent analysis, as most of the times these missing values are ignored. In this paper we have analyzed how accurate estimation of missing values can lead to better subsequent gene selection and class prediction. Collateral Missing Values Estimation (CMVE), which demonstrates superior imputation performance compared to Bayesian Principal Component Analysis (BPCA) Impute, K-Nearest Neighbour (KNN) algorithm, when estimating missing values in the BRCA1, BRCA2 and Sporadic genetic mutation samples present in ovarian cancer by exploiting both local/global and positive/negative correlation values. CMVE also consistently outperforms, in terms of classification accuracies, BPCA, KNN and ZeroImpute techniques. The imputation is followed by gene selection using fusion of Between Group to within Group Sum of Squares and Weighted Partial Least Squares where Ridge Partial Least Square algorithm is used as a class predictor.

show abstract

Section: Classification Error Measurementioning

confidence: 94%

Section: Introductionmentioning

confidence: 99%

Collateral Missing Value Estimation: Robust Missing Value Estimation for Consequent Microarray Data Processing

Sehgal

Gondal

Dooley

2005

AI 2005: Advances in Artificial Intelligence

View full text Add to dashboard Cite

show abstract

“…We compare our findings with those of Storey and Tibshirani (2003), and of Broët et al (2004), who also analysed this data set using different approaches.…”

Section: Introductionmentioning

confidence: 94%

“…The functional classes (where known) of the remaining 23 genes are shown in Table 2, and interestingly include several genes involved in cell death as well as cell cycle control. Broët et al (2004) recently also applied a mixture model appproach to identify differentially expressed genes in this data set. However, they implemented a Bayesian approach, in contrast to the frequentist approach as applied here.…”

Section: Application To Hedenfalk Breast Cancer Datamentioning

confidence: 99%

“…In particular, genes which score as most significant using standard methods for multiple hypothesis testing may not necessarily be of most biological relevance (see Broët et al 2004). Genes with more subtle changes in their expression levels, indicating that they are more tightly regulated, may be of more importance in the biology of tumour formation.…”

Section: Application To Hedenfalk Breast Cancer Datamentioning

confidence: 99%

See 1 more Smart Citation

Application of Mixture Models to Detect Differentially Expressed Genes

Jones

Bean

McLachlan

et al. 2005

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. An important and common problem in microarray experiments is the detection of genes that are differentially expressed in a given number of classes. As this problem concerns the selection of significant genes from a large pool of candidate genes, it needs to be carried out within the framework of multiple hypothesis testing. In this paper, we focus on the use of mixture models to handle the multiplicity issue. With this approach, a measure of the local FDR (false discovery rate) is provided for each gene. An attractive feature of the mixture model approach is that it provides a framework for the estimation of the prior probability that a gene is not differentially expressed, and this probability can subsequently be used in forming a decision rule. The rule can also be formed to take the false negative rate into account. We apply this approach to a well-known publicly available data set on breast cancer, and discuss our findings with reference to other approaches.

show abstract