Thousands of samples are needed to generate a robust gene list for predicting outcome in cancer.

By L. Ein-Dor, Or Zuk, Eytan Domany. Proceedings of the National Academy of Sciences of the United States of America, 2006

720 citations recorded by the source corpus.

Subfields: Molecular Biology

View Paper: https://www.semanticscholar.org/paper/99015cffdecaa58600198d44029565851fcc1b7f

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Molecular Biology

Citations in the corpus

Machine learning applications in cancer prognosis and prediction (2014, background): “…of lack of agreement between the predictive gene lists discovered by different groups,theneed of thousands of samples inorder toachievethe desired outcomes, the lack of biological interpretation of predictive signatures and the dangers of information leak recorded in published studies [15 – 18].”
Prognostic gene signatures for non-small-cell lung cancer (2009, background): “Ein-Dor and coworkers (20) demonstrated that biological heterogeneity leads to thousands of samples being required to identify robust and reproducible subsets for most tumor types.”
The Influence of Feature Selection Methods on Accuracy, Stability and Interpretability of Molecular Signatures (2011, methodology): “While the limits of some basic methods for feature selection have been highlighted in the context of molecular signatures, such as gene selection by Pearson correlation with the output (Ein-Dor et al., 2006), there are surprisingly very few and only partial investigations that focus on the influence”
GOBO: Gene Expression-Based Outcome for Breast Cancer Online (2011, positive): “However, it has become evident that gene signatures often represent the same biological processes [22], are associated with outcome only in subgroups of breast cancer [18,19], and that validation of gene signatures needs to be performed in large independent data sets [45].”
The MicroArray Quality Control (MAQC) project shows inter- and intraplatform reproducibility of gene expression measurements (2006, background): “…to appear in peer-reviewed journals 9,10 , raising doubts about the repeatability, reproducibility and comparability of microarray technology 11–13 , several studies have also been recently published showing increased reproducibility of microarray data generated at different test sites and/or…”
Near-optimal Supervised Feature Selection among Frequent Subgraphs (2009, background): “Pearson’s correlation (PC) is commonly used in microarray data analysis [26, 10], where discriminative genes for phenotype prediction need to be selected from thousands of uninformative ones.”
The balance of reproducibility, sensitivity, and specificity of lists of differentially expressed genes in microarray studies (2008, methodology): “Correlation coefficient, which behaves similarly to the t-statistic, has also been widely used as a gene selection method in the identification of signature genes for classification purposes [15,47,48].”
Supervised feature selection via dependence estimation (2007, methodology): “Pearson’s correlation is commonly used in microarray analysis [13, 14 ], and is defined as rxy = ( Pm i=1 (xi ¯x)(yi ¯ y))/(sxsy), for each column x of X (scores are computed separately for each feature).”
Network-based support vector machine for classification of microarray samples (2009, background): “In recent years, researchers have realized that gene markers identified from microarrays drawn from difierent studies on the same disease across similar cohorts lack consistency [6,7].”
Predictive data mining in clinical medicine: Current issues and guidelines (2008, methodology): “[58] used a theoretical analysis to show that thousands of patients may be needed to obtain reliable gene lists.”
Identifying protein interaction subnetworks by a bagging Markov random field-based method (2012, background): “Finally, because of the heterogeneity in tissue samples and the inherent noise in microarray data (17), reproducibility of currently identified subnetworks is often low when tested on independent data sets (18,19).”
Algorithms for Discovery of Multiple Markov Boundaries (2013, result): “Thus it is not an exclusively small-sample phenomenon, as it was postulated by earlier research (Ein-Dor et al., 2005, 2006).”
Batch effect removal methods for microarray gene expression data integration: a survey (2013, background): “A clear application, where this limitation has been pointed out, is the prediction of disease outcome where thousands of samples are needed to generate robust gene/protein signatures [5, 6].”
Deep learning in medical imaging and radiation therapy. (2018, background): “It is, for example, reminiscent of issues observed in genomics where lack of reproducibility was observed when looking for predictive gene lists in small datasets (~100s of cases).(368,369) There, thousands of samples are needed to generate a robust gene list to predict the outcome in cancer.”
Statistical strategies for avoiding false discoveries in metabolomics and related experiments (2007, methodology): “, 2002) ‘significant’ genes, but only 3 of them were in fact the same (Ein-Dor et al., 2006)! Equally, permuting subsets of the samples also ‘discovered’ entirely different lists of genes.”