Independent Component Analysis of Simulated 2D Electrophoresis Gels
Nicolle M. Correa, Haleh Safavi, Wei Xiong, Anindya Roy, Tülay Adalı, Françoise Seillier‐Moiseiwitsch · Machine learning for signal processing ... · 2006
We present a novel method to detect differentially expressed proteins in simulated two-dimensional electrophoresis (2DE) gels using spatial independent component analysis (ICA). We use an analysis-of-variance (ANOVA) approach as a benchmark for comparison and show that it yields improved detection power compared to currently used t-test based methods for 2DE gel analysis. However, ICA proves to be much faster than ANOVA and unlike ANOVA, it does not depend on any threshold. We also use ICA on wavelet-transformed 2DE gels. ICA in the wavelet domain offers increase in speed, and also, better performance for misaligned gels as compared to the ICA in the spatial domain.