[BioC] some mismatches between affymetrix CGOS MAS5 detection calls and BioC mas5calls

lgautier at altern.org lgautier at altern.org
Mon Dec 11 16:51:27 CET 2006


"mas5call" is an implementation, that is a physical realization
of something that is described in a documentation.
The effective presence of bugs in an implementation cannot be ruled
out, but there are a number of other reasons for differences: numerical
precision issues and rounding errors, but also a reader's interpretation
of the described procedure.
The best way to have MAS5.0 results as done by CGOS... is probably
to use CGOS itself.

However (and beside further statements on the nature of things), BioC is
giving you the tools to investigate at the probe level why these 300
probesets are either M or P, and decide for yourself (and eventually
send us a bug report if you find a mistake on our side).
If the these genes pass from one side to the other, they could be close to
the borderline in the first place.

Hoping this helps,


> Hi,all
> Recently I download some CEL files from GEO and get it's detection
> calls using BioC mas5calls . When I compare this file with standard file
> produced by Affymetrix company's CGOS platform,I found there are about 300
> probesets called "M" by mas5call function but called "P" in my standard
> file. These mismatches may affect our analysis result if we use detection
> calls as a filter condition or for special algorithm using calls
> infomation.  For many of us ,one may download CEL files from GEO or
> ArrayExpress but can not get raw files contain detection calls from those
> website.So we use BioC's mas5call function.But there are differences
> between
> two method! What should I do?
> Regards
> haiyan
> 	[[alternative HTML version deleted]]
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives:
> http://news.gmane.org/gmane.science.biology.informatics.conductor
> !DSPAM:457d68ac276685618593157!

More information about the Bioconductor mailing list