[BioC] Error when applying ComBat in SCAN.UPC

Peter Langfelder peter.langfelder at gmail.com
Sun Jun 8 18:57:30 CEST 2014


On Sun, Jun 8, 2014 at 3:31 AM, Joel Ma <jzma at unimelb.edu.au> wrote:
> Hi all
>
> I have 48 Affymetrix mogene data and my batch text file is as follows:
>
> Name
>         Batch
>
> Affy1.CEL
>         1
>
> Affy2.CEL
>         1
>
> Affy3.CEL
>         1
>
> AffyA1.CEL
>         2
>
> AffyA2.CEL
>         2
>
>
>
>
>
>
>
>
>>library(SCAN.UPC)
>> Sdata <- SCAN("*.CEL", convThreshold = 0.01, batchFilePath = "batchtargets1.txt", verbose = TRUE)
>
> After completing the iterations, I get this message.
>
> No batch covariates have been specified.
> Getting ready to perform batch adjustment.
> Found 2 batches
> Found 0  categorical covariate(s)
> Standardizing Data across genes
> Fitting L/S model and finding priors
> Finding parametric adjustments
> Error in while (change > conv) { : missing value where TRUE/FALSE needed
>
> I eyeballed my Sdata that I did previously (using the following) but did not spot any NaN or missing values in the normalised values.
>
>> Sdata <- SCAN("*.CEL", convThreshold = 0.01, verbose = TRUE)
>
> I am not sure what went wrong. Can someone enlighten me?


I have seen this problem in my data as well - in my case it was caused
by genes/probesets with zero variance in at least on of the batches.
Unfortunately my knowledge of the scan.upc package is non-existent so
I can't tell you how to fix the workflow for SCAN.

Best,

Peter



More information about the Bioconductor mailing list