Entering edit mode
a.karanikolou
▴
10
@akaranikolou-20591
Last seen 5.3 years ago
Hi all, I have recently run the HTA 2.0 annotation using the two packages affycoretools and hta20transcriptcluster.db as outlined in the post https://support.bioconductor.org/p/89308/. However, I have run into an error after running the rma() as follows:
> exprs_mainProbes<-getMainProbes(exprsgs, pd.hta.2.0)
Error in switch(level, core = dbGetQuery(con, paste("select distinct core_mps.transcript_cluster_id, type from featureSet inner join core_mps", : EXPR must be a length 1 vector
Could you shed light into this problem? Thanks in advance.
Thank you for your help.
Is it mandatory to include "level = "core"" in the "getMainProbes(input, level = "core")?
In the "help" menu, the argument "level" refers to "the summarization level used when calling rma". I'm not sure I understand it well on how to apply this argument in getMainProbes.
If a function has a default, then by definition it isn't necessary to supply a value for that argument. But perhaps you are actually asking what value you should use? Again, the answer to your question can be found by reading the help page, this time for
rma
:If you didn't specify a target argument for
rma
, then you by definition used the default, which matches the default forgetMainProbes
.Dear James, thank you for your previous reply. We have used the function getMainProbes using the default as in rma, however instead of the 67,528 main probes that should have been retained, the result was 67,480 main probes. What would cause this difference? Thanks in advance.
I think your math is off.
So there are 67480 main probesets out of the expected 67516, or 36 missing probesets. Where might those probesets be?
So there are 36 probesets that are both 'main' and 'NA'. If we look closer, it's sort of weird
So for whatever reason, Affy is calling at least one probe per probeset an NA type, so the probesets are getting called both main and 'NA' and that is causing them to be excluded. Given that this only affects 36 out of 67528 probesets (and our core won't use these arrays anyway) it has never been worth it to me to find out any more about it than that.
Thank you a lot for this. The response from Affy was that and I quote "67528 Probe sets listed: 44699 Coding and 22829 NonCoding". The categories used as controls are sum up to 2,995. The is no clear understanding why Affy is calling at least one probe per probeset an NA type. But we will take your response and their response for consideration before moving forward with the analysis. Thank you once again.
And the other reason I never pursued it any further is that ThermoFisherSigmaAldrichAffyABI is like, not going to help figure it out anyway.