An Additive Model to Normalize Spotted Arrays Using Spike Controls
Motivation: The study of a metabolic pathway is focused on a limited number of genes in comparison to whole-genome studies. Therefore, a large number of (spike) controls may be printed within slide together with several ESTs replicates to reduce bias and variance of estimates.
Results: We develop a linear additive mixed effect model to remove dye and spatial biases using spike controls. We adapt the iterated weighted least squares algorithm to obtain a fast algorithm to search for the optimal model and to perform point estimates of model parameters. Actual data from a very noisy calibration slide have been successfully normalized following our model.
STEFANINI Federico;
MASCHERINI Massimiliano;
BRANDI Maria;
MAVILIA Carmelo;
2008-01-21
Societa' Italiana di Biometria
JRC36914
Additional supporting files
File name | Description | File type | |