Search
Question: ChAMP filter removing good samples from different batch of EPIC arrays.
0
gravatar for sarka.vorackova
3 months ago by
sarka.vorackova0 wrote:

Hi,
I have a problem analysing EPIC arrays with ChAMP. I have a set of 7 arrays - one part of arrays was done during November 2016-January 2017 (4 arrays), second part was done in June 2017.
When I analysed the first and second portion of arrays separately, everything seems ok in both cases, all the samples pass the filtering with Failed CpG fraction around 0.03-0.001.
But when I try to analyse all the arrays together, the filter removes all samples from arrays from June according to great amount of failed probes:

                 Failed CpG fraction
MRK50 BS            0.9194669703
MRK50 oxBS          0.7961700762
Z748 BS             0.8965456313
Z748 oxBS           0.7784871085
TRO63 BS            0.9166965001
TRO63 oxBS          0.7794537127 #etc. for the 3 arrays from June - all the samples were removed by  the filter

0568 BS             0.0022115258
0568 oxBS           0.0301945450
0515 BS             0.0014966775
0515 oxBS           0.0195157047
0681 BS             0.0011225081
0681 oxBS           0.0105864528
G89 BS              0.0017611367
G89 oxBS            0.0332179259 #etc. for the other 4 arrays

Also in GenomeStudio the number of probes with detection pval<0.01 was quite similar for all samples.

Any ideas what could be the problem??

I also tried to proceed all the arrays together with minfi but I got an error:

Error in read.metharray(files, extended = extended, verbose = verbose,  :
  [read.metharray] Trying to parse IDAT files with different array size but seemingly all of the same type.

but when I set force=TRUE in

read.metharray.exp(targets = targets, extended = FALSE, recursive = FALSE, verbose = FALSE, force = TRUE)

all the samples were processed without any further problems and the quality control did not show any bad samples.
Unfortunately I don’t know how can I set force=TRUE for read.metharray.exp in ChAMP if I want to do champ.load(method="minfi").

Can anyone help?

Thanks in advance, I will appreciate any answers!!

ADD COMMENTlink modified 3 months ago by Yuan Tian70 • written 3 months ago by sarka.vorackova0
0
gravatar for Yuan Tian
3 months ago by
Yuan Tian70
London
Yuan Tian70 wrote:

Hello:

I am the developer of ChAMP, and you indeed raised a good question, when I coding ChAMP, I did not consider situation that different arrays might be mixed together. And previous I did not included parameter of minfi Force in ChAMP parameter. I will fix it now. You should see the change in 24-36 hours.

Best

Yuan Tian

ADD COMMENTlink written 3 months ago by Yuan Tian70
The many failed samples is because the data from these (later) arrays are scrambled and therefore garbage. This happens when you read epic arrays which have different number of probes in, with earlier versions of minfi. This is fixed in current release. Best, Kasper (Sent from my phone.) > On Aug 4, 2017, at 21:09, Yuan Tian [bioc] <noreply@bioconductor.org> wrote: > > Activity on a post you are following on support.bioconductor.org > User Yuan Tian wrote Answer: ChAMP filter removing good samples from different batch of EPIC arrays.: > > Hello: > > I am the developer of ChAMP, and you indeed raised a good question, when I coding ChAMP, I did not consider situation that different arrays might be mixed together. And previous I did not included parameter of minfi Force in ChAMP parameter. I will fix it now. You should see the change in 24-36 hours. > > Best > > Yuan Tian > > Post tags: minfi, champ > > You may reply via email or visit A: ChAMP filter removing good samples from different batch of EPIC arrays.
ADD REPLYlink written 3 months ago by Kasper Daniel Hansen6.3k

Great, thanks for answers and everything!!

Sarka

ADD REPLYlink written 3 months ago by sarka.vorackova0
0
gravatar for sarka.vorackova
3 months ago by
sarka.vorackova0 wrote:

Great, thanks for answers and everything!!

Sarka

ADD COMMENTlink written 3 months ago by sarka.vorackova0
0
gravatar for Yuan Tian
3 months ago by
Yuan Tian70
London
Yuan Tian70 wrote:

Hello Sarka:

Thanks Kasper for your kind explain. I just modified the code, and make sure ChAMP would do correct calculate for mix array data set. I did not encounter this situation before thus did not consider this fully enough. Pretty sorry for the bug. 

I have find a mix-array data set and tested it with ChAMP and latest version minfi (development version), they both can process mix-array and return the same fail CpG fraction value. So I think it's correct. 

You can use it after new ChAMP online, it should be version 2.8.9.

Best

Yuan Tian

ADD COMMENTlink written 3 months ago by Yuan Tian70

Thank you Yuan Tian, I just installed the new version and the data loading works perfectly!

Best, Sarka
 

ADD REPLYlink written 3 months ago by sarka.vorackova0
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.2.0
Traffic: 147 users visited in the last hour