FCCC LOGO Faculty Publications
Wang G , Ochs MF
Estimating gene function with least squares nonnegative matrix factorization
Methods Mol Biol. 2007 ;408 :35-47
PMID: 18314576   
Back to previous list
Abstract
Nonnegative matrix factorization is a machine learning algorithm that has extracted information from data in a number of fields, including imaging and spectral analysis, text mining, and microarray data analysis. One limitation with the method for linking genes through microarray data in order to estimate gene function is the high variance observed in transcription levels between different genes. Least squares nonnegative matrix factorization uses estimates of the uncertainties on the mRNA levels for each gene in each condition, to guide the algorithm to a local minimum in normalized chi2, rather than a Euclidean distance or divergence between the reconstructed data and the data itself. Herein, application of this method to microarray data is demonstrated in order to predict gene function.
Notes
Wang, Guoli Ochs, Michael F United States Methods in molecular biology (Clifton, N.J.) Methods Mol Biol. 2007;408:35-47.