Robust PCA and classification in biosciences

Hubert, Mia; Engelen, Sanne

doi:10.1093/bioinformatics/bth158

Cited by 164 publications

(81 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…As an initial assessment of the patterns of U251 gene expression in the three model systems, PCA was applied. This unsupervised method allows for the comparison of complex data sets in a 3D graph (10). As shown in Fig.…”

Section: Resultsmentioning

confidence: 99%

Influence of in vivo growth on human glioma cell line gene expression: Convergent profiles under orthotopic conditions

Camphausen

Purow

Sproull

et al. 2005

Proc. Natl. Acad. Sci. U.S.A.

108

View full text Add to dashboard Cite

Defining the molecules that regulate tumor cell survival is an essential prerequisite for the development of targeted approaches to cancer treatment. Whereas many studies aimed at identifying such targets use human tumor cells grown in vitro or as s.c. xenografts, it is unclear whether such experimental models replicate the phenotype of the in situ tumor cell. To begin addressing this issue, we have used microarray analysis to define the gene expression profile of two human glioma cell lines (U251 and U87) when grown in vitro and in vivo as s.c. or as intracerebral (i.c.) xenografts. For each cell line, the gene expression profile generated from tissue culture was significantly different from that generated from the s.c. tumor, which was significantly different from those grown i.c. The disparity between the i.c gene expression profiles and those generated from s.c. xenografts suggests that whereas an in vivo growth environment modulates gene expression, orthotopic growth conditions induce a different set of modifications. In this study the U251 and U87 gene expression profiles generated under the three growth conditions were also compared. As expected, the profiles of the two glioma cell lines were significantly different when grown as monolayer cultures. However, the glioma cell lines had similar gene expression profiles when grown i.c. These results suggest that tumor cell gene expression, and thus phenotype, as defined in vitro is affected not only by in vivo growth but also by orthotopic growth, which may have implications regarding the identification of relevant targets for cancer therapy.microarray ͉ intracerebral ͉ brain tumor

show abstract

Section: Resultsmentioning

confidence: 99%

Influence of in vivo growth on human glioma cell line gene expression: Convergent profiles under orthotopic conditions

Camphausen

Purow

Sproull

et al. 2005

Proc. Natl. Acad. Sci. U.S.A.

108

View full text Add to dashboard Cite

show abstract

“…Essentially, SIMCA can be summarized as an approach that combines robust PCA within each group based on robust covariance estimation to achieve good predictive performances in classification. More details on SIMCA can be found in [21], [20] [2] and [38]. SIMCA has been widely applied to areas as diverse as image analysis, microarray gene expression classification, and many other fields where data exists with n much less than p. An implementation of Robust SIMCA is provided through the R package rrcovHD, and will be used in our comparison of predictive performances of high dimensional robust classifiers.…”

Section: Robust Estimation Methods For Linear Discriminant Analysismentioning

confidence: 99%

Predictive performances of implicitly and explicitly robust classifiers on high dimensional data

́E¹

2017

Communications Faculty of Science University of Ankara Series A1Mathematics and Statistics

View full text Add to dashboard Cite

Abstract. The goal of this paper is to demonstrate via extensive simulation that implicit robustness can substantially outperform explicit robust in the pattern recognition of contaminated high dimension low sample size data. Our work specifically demonstrates via extensive computational simulations and applications to real life data, that random subspace ensemble learning machines, although not explicitly structurally designed as a robustness-inducing supervised learning paradigms, outperforms the structurally robustness-seeking classifiers on high dimension low sample size datasets. Random forest (RF), which is arguably the most commonly used random subspace ensemble learning method, is compared to various robust extensions/adaptations of the discriminant analysis classifier, and our work reveals that RF, although not inherently designed to be robust to outliers, substantially outperforms the existing techniques specifically designed to achieve robustness. Specifically, by exploring different scenarios of the sample size n and the input space dimensionality p along with the corresponding capacity κ = n/p with κ < 1, we demonstrate through extensive simulations that regardless of the contamination rate ϵ, RF predictively outperforms the explicitly robustness-inducing classification techniques when the intrinsic dimensionality of the data is large.

show abstract

“…Good empirical results were reported, e.g. in Hubert and Engelen (2004). In this article local robustness and efficiency of the SD so estimator are studied.…”

Section: Accepted Manuscriptmentioning

confidence: 97%

The influence function of the Stahel–Donoho covariance estimator of smallest outlyingness

Debruyne¹,

Hubert²

2009

Statistics & Probability Letters

Self Cite

View full text Add to dashboard Cite

To cite this version:M. Debruyne, M. Hubert. The influence function of the Stahel-Donoho covariance estimator of smallest outlyingness. Statistics and Probability Letters, Elsevier, 2009, 79 (3) This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. A C C E P T E D M A N U S C R I P T ACCEPTED MANUSCRIPTThe influence function of the Stahel-Donoho covariance estimator of smallest outlyingness Leuven, Belgium AbstractIn traditional multivariate location and scatter estimation based on the StahelDonoho outlyingness, a weight function is applied, usually calibrated with respect to the multivariate Gaussian distribution. Other robust methods compute the covariance matrix of a fixed size subset of the data (e.g. the MCD estimator). In this paper we study a combination of both ideas. Location and scatter are estimated using a fixed size subset of the data containing the points with smallest Stahel-Donoho outlyingness. Local robustness and asymptotic relative efficiency are investigated.

show abstract

Robust PCA and classification in biosciences

Abstract: All the programs are part of the Matlab Toolbox for Robust Calibration, available at http://www.wis.kuleuven.ac.be/stat/robust.html.

Cited by 164 publications

References 18 publications

Influence of in vivo growth on human glioma cell line gene expression: Convergent profiles under orthotopic conditions

Influence of in vivo growth on human glioma cell line gene expression: Convergent profiles under orthotopic conditions

Predictive performances of implicitly and explicitly robust classifiers on high dimensional data

The influence function of the Stahel–Donoho covariance estimator of smallest outlyingness

Contact Info

Product

Resources

About