preprocessNoob + preprocessQuantile Error handling preprocessMethod
2
0
Entering edit mode
@andrewjskelton73-7074
Last seen 4 months ago
United Kingdom

On the advice of this post, I tried the preprocessNoob() -> preprocessQuantile() route of normalisation in the minfi package.  Using the resulting object from preprocessNoob, when I run preprocessQuantile I get the error:

Error in if ((is(object, "MethylSet") || is(object, "GenomicMethylSet")) &&  : 
  missing value where TRUE/FALSE needed

Which was odd. Checking the code for preprocessQuantile, we can see that it runs this line:

  if ((is(object, "MethylSet") || is(object, "GenomicMethylSet")) && 
        preprocessMethod(object)["rg.norm"] != "Raw (no normalization or bg correction)") 
        warning("preprocessQuantile has only been tested with 'preprocessRaw'")

Which looks for the preprocess method, and the rg.norm column. Which doesn't exist. Running the code: 

preprocessMethod(lumi.norm.bgcorr)

Gives the result:

             mu.norm 
"Noob, dyeCorr=TRUE" 

...and mu.norm != rg.norm - Any advice?

minfi • 2.3k views
ADD COMMENT
1
Entering edit mode
@kasper-daniel-hansen-2979
Last seen 13 months ago
United States
This is a bug. I have travel and a couple of other deadlines, so it might take some time to fix properly. Note that we have never tested preprocessQuantile on output from preprocessNoob. In general you cannot assume that combining two different methods are better than the best of the individual methods. Kasper
ADD COMMENT
0
Entering edit mode

Thanks Kasper. So while preprocessFunnorm and preprocessIllumina both have implementations for background correction prior to   the normalisation, I know that preprocessQuantile and preprocessSWAN don't. If I remember right, preprocessFunnorm is an extension of the quantile method, so would it not make sense to background and dye bias correct prior to normalisation? (I realise it isn't tested, but I just wanted to be sure my logic was sound)

ADD REPLY
0
Entering edit mode
I am not disagreeing with you that it could be a good idea to combine Noob and Quantile, and that this is different to combining (say) two different between-array normalization methods (note that Noob does do some between array stuff). I am just pointing out to you that the onerous is on you to check whether you actually get "better" results; you cannot assume it based on first principles. Thanks for submitted this issue to the gitHub tracker. Best, Kasper On Thu, Aug 20, 2015 at 4:41 AM, andrew.j.skelton73 [bioc] < noreply@bioconductor.org> wrote: > Activity on a post you are following on support.bioconductor.org > > User andrew.j.skelton73 <https: support.bioconductor.org="" u="" 7074=""/> wrote Comment: > preprocessNoob + preprocessQuantile Error handling preprocessMethod > <https: support.bioconductor.org="" p="" 71234="" #71264="">: > > Thanks Kasper. So while preprocessFunnorm and preprocessIllumina both > have implementations for background correction prior to the > normalisation, I know that preprocessQuantile and preprocessSWAN don't. > If I remember right, preprocessFunnorm is an extension of the quantile > method, so would it not make sense to background and dye bias correct prior > to normalisation? (I realise it isn't tested, but I just wanted to be sure > my logic was sound) > > ------------------------------ > > Post tags: minfi > > You may reply via email or visit > C: preprocessNoob + preprocessQuantile Error handling preprocessMethod >
ADD REPLY
0
Entering edit mode

Cheers Kasper - I will do some testing between preprocessQuantile and preprocessNoob + preprocessQuantile, once the bug is sorted out!

ADD REPLY
0
Entering edit mode

Hi Andrew,

I am trying to apply some normalization method on EPIC array and came to this topic. As you discussed above what is your opinion about applying  preprocessQuantile and preprocessNoob + preprocessQuantile? Which method worked better for your data normalization?

Thanks.

ADD REPLY
1
Entering edit mode
@kasper-daniel-hansen-2979
Last seen 13 months ago
United States
It would be great if you could add this as an issue on the minfi Github page. Then you'll get notified when it is fixed.
ADD COMMENT
0
Entering edit mode

Issue has been raised here

ADD REPLY

Login before adding your answer.

Traffic: 585 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6