Hello,

I am using ChAMP for my methylation analysis. In my case my data isn't normalized through the ChAMP pipeline. I am hoping to use ChAMP from section 5.6 of the tutorial.(https://bioconductor.org/packages/release/bioc/vignettes/ChAMP/inst/doc/ChAMP.html) I was able to run DMP successfully but when I ran Bumphunter and Block finder the program seems to fail. This is how I am running the DMR analysis. 

myDMR <- champ.DMR(beta = myNorm,pheno=test$Sample_Group,method="Bumphunter")

myBlock <- champ.Block(beta=myNorm,pheno=test$Sample_Group,arraytype="450K")

> class(myNorm)

[1] "matrix"

> test$Sample_Group

 [1] "Good" "Good" "Good" "Bad"  "Bad"  "Bad"  "Good" "Bad"  "Good" "Bad" 

[11] "Bad"  "Bad"  "Bad"  "Good" "Bad"  "Good" "Bad"  "Bad"  "Bad"  "Bad" 

>head(test,5)

 Sample_Name Sample_Well Sample_Plate Sample_Group Pool_ID

1  1          NA           NA         Good      NA

2  2          NA           NA         Good      NA

3  3          NA           NA         Good      NA

4  4          NA           NA          Bad      NA

5  5          NA           NA          Bad      NA

   person  age sex status  Array        Slide

1 21   F   Good R03C01 200517420015

2 22   F   Good R06C01 200517420102

3 31   M   Good R06C02 200517420085

4 23   F    Bad R01C01 200517420029

5 48   M    Bad R06C02 200517420117

    Basename

1  /methylation/idats/200517420015_R03C01

2  /methylation/idats/200517420102_R06C01

3  /methylation/idats/200517420085_R06C02

4  /methylation/idats/200517420029_R01C01

5 /methylation/idats/200517420117_R06C02

    filenames

1  /methylation/idats/200517420015_R03C01

2  /methylation/idats/200517420102_R06C01

3  /methylation/idats/200517420085_R06C02

4   /methylation/idats/200517420029_R01C01

5 /methylation/idats/200517420117_R06C02c

The out and error that I get from bumhunter is:

[===========================]

[<<<<< ChAMP.DMR START >>>>>]

-----------------------------

<< Find DMR with Bumphunter Method >>

3 cores will be used to do parallel BMIQ computing.

According to your data set, champ.DMR() detected 0 clusters contains MORE THAN 7 probes within300 maxGap. These clusters will be used to find DMR.

[bumphunterEngine] Parallelizing using 3 workers/cores (backend: doParallelMC, version: 1.0.10).

[bumphunterEngine] Computing coefficients.

[bumphunterEngine] Smoothing coefficients.

Error in FUN(X[[i]], ...) : reduce can only handle matrices or vectors

For Block analysis:

[===========================]

[<<<< ChAMP.Block START >>>>]

-----------------------------

<< Load Annotation Successfully >>

<< Get Clusters by cgi-info Successfully >>

<< Calculate Average Beta Value Successfully >>

<< Generate Block Position Successfully >>

<< New Clusters are generated for blocks >>

<< Generate information for New Clusters >>

[bumphunterEngine] Parallelizing using 3 workers/cores (backend: doParallelMC, version: 1.0.10).

[bumphunterEngine] Computing coefficients.

[bumphunterEngine] Smoothing coefficients.

[bumphunterEngine] Performing 500 permutations.

[bumphunterEngine] Computing marginal permutation p-values.

[bumphunterEngine] Smoothing permutation coefficients.

[bumphunterEngine] cutoff: NA

[bumphunterEngine] Finding regions.

[bumphunterEngine] No bumps found!

<< Run Bumphunter Successfully >>

Error in 1:nrow(Block) : argument of length 0

Hello:

Based on your message from ChAMP, I think maybe there is something with the rowname of your beta matrix. Because there is message here : "According to your data set, champ.DMR() detected 0 clusters contains MORE THAN 7 probes within300 maxGap. These clusters will be used to find DMR". Normally, bumphunter will firstly get all clusters, which is relied on your CpG probe location, but here champ.DMR() shows it can get no clusters at all, that is in correct. So based on my understanding, champ.DMR() will firstly map all probes in your beta matrix into annotation file, then get all the probe's position and chr information, I suspect the program failed here. So have you checked if your beta matrix contains CpG ID as rowname?

The latter one, champ.Block() is obviously, it shows no bumps can be detected.

Best

Yuan Tian

Thanks for replying.  You are right about the message with "0 clusters", I was using only 10 CpG's for my test run, using all CpG's solved the problem.