Novel normalization algorithms and QA measures for array CGH.
Normalization of arrayCGH data differs from that of expression data in
many respects including that: (1) hybridization signals are typically
smaller and with higher background due to the excess of non-target
labeling products. (2) small changes in copy number may mark
biologically significant events, such as single-copy gains or losses
that may accompany unbalanced translocations or chromosomal
rearrangements. (3) log2 ratios are expected to be highly correlated
between probes targeting adjacent regions in the genome. There is no
commercial normalization software, which addresses the unique aspects
of aCGH data. We have developed and refined novel approaches to
aCGH-specific normalization and QA measures based on more than 1000
hybridizations on Agilent cDNA and oligo expression arrays while in
development at Dana-Farber, and several hundred hybridizations on
oligo aCGH arrays recently at MSKCC.
Complete Methods Description
The following compressed tar archive has all of the necessary R-code
to run the gcNormalization algorithm used here at MSKCC.