[BioC] 217216_x_at is NOT dihydrolipoamide S-succinytransferase

Robert Gentleman rgentlem at fhcrc.org
Wed Mar 30 03:32:55 CEST 2005

Hi Kevin,

  Thank-you for the report but I think you are under some  
misapprehensions about both what we do, and what we can do. First, we  
"merely" (in quotes because it is a big job) map from what the  
manufacturer tells us to a set of best matches from published and  
reliable sources, such as the NCBI. The routines used are well  
documented (see the AnnBuilder package and relevant publications named  
therein) and applied as documented. While you view this as an error, we  
do not. An error, is when we have either misidentified the match or  
misaligned it with the public data source. Not, if the knowledge about  
the biology has changed, or if the public source itself has made a  
mistake. We are repackagers of data not arbiters; we do not have the  
resources nor the capability to do that. We have tried to be clear  
about what we do and why. If you find it useful that is great. If you  
do not, that is your choice - we do provide AnnBuilder as a tool so  
that you could build your own set of packages if that suits you.

   In this case, the data we are using from Affymetrix maps as follows:
217216_x_at     AC006530

  So the question then becomes what is known about AC006530, and we tend  
to rely on Entrez Gene for that information:

(or however you want to get there).
That is the information we primarily use and our data packages reflect  

Note that Affymetrix has: 215210_s_at     S72422, so a different  

We do not have the resources (nor does anyone that I know of) to hand  
curate annotations as you are suggesting. If you would like our matches  
to be different, I can suggest making your improvements known to the  
NCBI or who ever their primary sources are and as they propagate we  
will be able to pick them up.

You may want to change your local copy to reflect your local conditions  
and views. If you do so, please make sure to change the version number  
to something that is different from our set, perhaps using -'s or  
similar, so that there can be no possible chance of confusion.

Best wishes,

On Mar 29, 2005, at 3:58 PM, Kevin Dawson wrote:

> Dear Kasper,
> Apparently my previous message did not appear on the Bioconductor list  
> serve
> due to its attachment. I think, Robert should have received it though.
> The annotation error was caused by the fact that DLAT and MLH3 are  
> close
> neighbors on chromosome 14 and at one point, somebody used the same  
> locus
> info for both. I cannot attach an image that appears on the list serve;
> therefore, probably, the easiest way to be demonstrate what I am  
> talking
> about is if you do the followings:
> (1) Go to affymetrix.com and go to NetAffy
> (2) Look up 217216_x_at. It will tell you, it refers to MLH3.
> (3) Click on <details>
> (4) At the alignments, click on the chr14 entry: This will take you to  
> the
> UCSC genome browser with the Affymetrix probe set information aligned  
> with
> the gene information
> (5) Extend the chromosomal location to chr14:74,422,000-74,600,000 to  
> see
> both genes at a time.
> On the left side of the map, you'll see DLST, and on the right side  
> you'll
> see MLH3. The 217216_x_at probeset matches MLH3 and NOT DLST.
> A second problem caused by the misannotation is that in the annotation
> packages, the 215210_s_at is annotating the DLST pseudogene. It is  
> true that
> the probeset also matches the pseudogene on chromosome 1; however it
> primarily matches DLST itself on chromosome 14. That is why I am  
> suggesting
> to clarify the issues and change the annotation of 215210_s_at to DLST  
> and
> the annotation of 217216_x_at to MLH3 as described in my previous post
> Thank you,
> Kevin
| Robert Gentleman              phone: (206) 667-7700                    
| Head, Program in Computational Biology   fax:  (206) 667-1319   |
| Division of Public Health Sciences       office: M2-B865               
| Fred Hutchinson Cancer Research Center                                 
| email: rgentlem at fhcrc.org                                              

More information about the Bioconductor mailing list