Improved annotation for the first (8247 gene) Affymetrix GeneChip

David Waner dwaner at
Fri Aug 2 13:39:33 EST 2002

<x-charset windows-1252>Dear Colleagues,

   If you are pursuing experiments with the first 8247 gene (8K) Affymetrix
Arabidopsis Genechip and have used publicly available annotations please read
on. We have now further improved our annotation of this chip and corrected
errors that have resulted from significant changes over time in gene locus
(AGI) identifiers.

   Since the release of our initial annotations Affymetrix has released the
target sequences of the first 8K Arabidopsis Genechip allowing us to double
check the automated annotations. We have found inconsistencies in a number of
genes when checking the original annotation and are now posting improved
annotations on our web sites. After in depth analyses we found that many of the
annotation inconsistencies could be traced back to changes in AGI identifiers
during TIGR's reannotation of the genome sequence. This results in TIGR and
MIPS protein identifiers being inconsistent with each other. For example, TIGR
At2g22200 is the same as MIPS At2g22210. (The numbers are often shifted by one
unit, i.e. one gene.) In addition, there are discrepancies between the some of
the genes identified in the annotations originally provided by Affymetrix, and
the TIGR genes that the target sequences actually match.  The annotations that
we assembled last Summer (Ghassemian et al. 2001 TIPS) were based on the MIPS
database, and therefore many of the genes now have different names from those
currently found at TIGR. Using the current TIGR database, and the recently
released target sequences by Affymetrix, we were able to detect these
inconsistencies and have improved the Affymetrix chip annotation.

   Those laboratories that have been analyzing Affymetrix chip annotations using
our database are encouraged to double check their data. We have posted the
updated Genechip annotations at the following web site:
Updated Genespring files are also available here. Note that we have used a
conservative strategy for identifying matches which left approximately 5% of
the probes unannotated. Details of our annotation method are posted on our web

   Cheri Chen and Yan-Xia Liu in Jen Sheen's lab have also assembled an
annotation list using different strategies and criteria. A search can also be
performed through their web site:
Their strategy is described at:

   Note that because of the different annotation criteria used there are a few
annotations that differ between the two web sites. If you have questions or
further suggestions please feel free to contact us at:
dwaner at or yliu at

Julian Schroeder, Jen Sheen, David Waner, Cheri Chen, Yan-Xia Liu, Michael
Gribskov, Jason Tchieu.



More information about the Arab-gen mailing list