ABSTRACT
The iterative signature algorithm (ISA) was proposed by Bergmann et al. (2003) as a new approach for the analysis of a large-scale expression data. It can be applied to any data matrix with continuous response. Let E be a M × N data matrix of M conditions and N features. Note that E is the transpose of the expression matrix Y˜ used in the other chapters of this book. E can be represented as a collection of row or column vectors defined by
EG =
gT1 gT2 ... gTN
and EC = ( c1, c2, · · · , cM ) ,
where EG is a M ×N matrix of column vectors and EC is a N ×M of row vectors. The matrices EG and EC can be normalized by
g′ = g−g¯sd(g) and c ′ = c−c¯sd(c) .
