Class prediction and discovery using gene microarray and proteomics mass spectroscopy data: curses, caveats, cautions

By R. Somorjai, Brion K. Dolenko, Richard Baumgartner. Bioinform., 2003

353 citations recorded by the source corpus.

Subfields: Analytical Chemistry, Biophysics, Molecular Biology

View Paper: https://www.semanticscholar.org/paper/25cad58bbfd78b3ec18f4125465551e7248becbf

How researchers see this paper

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

In Analytical Chemistry

David S. Wishart

In Biophysics

In Molecular Biology

Citations in the corpus

Algorithms for Discovery of Multiple Markov Boundaries (2013, background): “…as the “multiplicity” of molecular signatures (i.e., different gene/biomarker sets perform equally well in terms of predictive accuracy of phenotypes) suggests existence of multiple Markov boundaries in these distributions (Dougherty and Brun, 2006; Somorjai et al., 2003; Aliferis et al., 2010a).”
A Review of Ensemble Methods in Bioinformatics (2010, background): “This is commonly known as the “curse-of-dimensionality” [26], and selecting the most relevant features [27, 28] and making the most use of the limited data samples [29] are the key issues in microarray or MS-based proteomics classification problem.”
Big Data and machine learning in radiation oncology: State of the art and future prospects. (2016, background): “The patient-per-feature ratio is also very challenging when you want to integrate thousands of information (especially genomics) into a model, because it is considered that this ratio should be at least 5 to 10 [52].”
Applications of Machine Learning in Cancer Prediction and Prognosis (2006, background): “As a rule, an SFR of less than 5 does not necessarily guarantee a robust classi ﬁ er (Somorjai et al. 2003).”
Data mining in the Life Sciences with Random Forest: a walk in the park or lost in the jungle? (2012, neutral): “only dozens of samples [31–33].”
Distinguishing cell types or populations based on the computational analysis of their infrared spectra (2010, methodology): “To use LDA, the number of spectra needs to be at least five times greater than the number of variable”
Feature Importance in Nonlinear Embeddings (FINE): Applications in Digital Pathology (2016, methodology): “Thesechallengeshave to dowith the “curseof dimensionality” [8] and the “curse of data sparsity” [9] since the number of features is significantly larger than the number of training exemplars (typically patient studies).”
Gene selection and classification of microarray data using random forest (2006, neutral): “For the Leukemia data set our method consistently returns sets of two genes, similar to [27] using an exhaustive search method, and lower than the numbers given by [42] of 3 to 25.”