[BioC] NetAffx annotation order conflicts with RMA

[Marie Chehani Alles]

Hi,

This is a warning message for Bioconductor users who use NetAffx for their
annotation rather than “affyannot”.

Affymetrix appears to have the order of its probes within the .csv files
available on NetAffx (eg. HG-U133_Plus_2_annot121004) so that they are not
the same as those appearing on in the RMA output, resulting in the errors in
the assignation of intensity data to the correct annotations.

For instance the order of probes seen after reading my HG-U133 Plus 2.0 .CEL
files into LIMMA using RMA was:

Probe ID	Intensity1	Intensity 2
177_at	4.490467	4.501346
1773_at	6.147345	6.863805


However, the order of probes in the HU-U133_Plus_2_annot121004 file
downloaded from NetAffx was:

Probe ID	Gene Title
1773_at	farnesyltransferase, CAAX box, beta
177_at	phospholipase D1, phophatidylcholine-specific


which when copied and pasted into an excel spreadsheet of my normalized
intensities resulted in the wrong annotations being assigned to the probes:

Probe ID	Intensity1	Intensity 2	Gene Title
177_at	4.490467	4.501346	farnesyltransferase, CAAX box, beta
1773_at	6.147345	6.863805	phospholipase D1, phophatidylcholine-specific


In total 1129 probes were "out of order" between the two lists, and perhaps
NetAffx users should consider using a "match" command on the Probe IDs of
both lists to check that there are no errors being carried over to further
analyses.

Regards,
Chehani

Marie Chehani Alles
Cancer Research Program
Garvan Institute of Medical Research
384 Victoria Street, Darlinghurst 2010
Web: http://www.garvan.org.au
Tel: (612) 9295 8354