bumphunter stalls on "Smoothing coefficients"
1
0
Entering edit mode
@rasmusrydbirk-12534
Last seen 3.0 years ago

I've been using bumphunter to get DMR and Blocks as defined in the ChAMP package (champ.DMR and champ.Block functions, respectively) on a MethylationEPIC data set (37 controls, 41 patients). Unfortunately, for both functions, it's stalling at:

[bumphunterEngine] Computing coefficients. [bumphunterEngine] Smoothing coefficients.

I've been running with at least 20 cores/100 GB memory for up to 99 hours. I've been trying R v. 3.5, 3.5.3, and 3.6. How can this smoothing take so long/any advice on what I should do? The last time, I tried with R v. 3.5 using ChAMP v. 2.12.4 (bumphunter v. 1.24.5). with standard code:

champ.DMR(beta=myNorm,
          pheno=myLoad$pd$Sample_Group,
          compare.group=NULL,
          arraytype="EPIC",
          method = "Bumphunter",
          minProbes=7,
          adjPvalDmr=0.05,
          cores=20,
          ## following parameters are specifically for Bumphunter method.
          maxGap=300,
          cutoff=NULL,
          pickCutoff=TRUE,
          smooth=TRUE,
          smoothFunction=loessByCluster,
          useWeights=FALSE,
          permutations=NULL,
          B=250,
          nullMethod="bootstrap")

Output:

[ Section 2:  Run DMR Algorithm Start ]

Loading required package: IlluminaHumanMethylationEPICanno.ilm10b4.hg19
<< Find DMR with Bumphunter Method >>
20 cores will be used to do parallel Bumphunter computing.
According to your data set, champ.DMR() detected 11355 clusters contains MORE THAN 7 probes within300 maxGap. These clusters will be used to find DMR.

[bumphunterEngine] Parallelizing using 20 workers/cores (backend: doParallelMC, version: 1.0.14).
[bumphunterEngine] Computing coefficients.
[bumphunterEngine] Smoothing coefficients.
Loading required package: rngtools
Loading required package: pkgmaker
Loading required package: registry

Attaching package: 'pkgmaker'

The following object is masked from 'package:S4Vectors':

    new2

The following object is masked from 'package:base':

    isFALSE

=>> PBS: job killed: walltime 356440 exceeded limit 356400

I will add sessionInfo() later, server is busy right now.

bumphunter ChAMP • 655 views
ADD COMMENT
2
Entering edit mode
Yuan Tian ▴ 250
@yuan-tian-13904
Last seen 8 months ago
United Kingdom

Someone in my lab reported similar error before. But I can not replicate it on my server or laptop. So I eventually solved it by ask her setting cores=1, means not use parallel.

Thus, I suspect the reason is doParallel package.

Best Yuan Tian

rasmus.rydbirk [bioc] noreply@bioconductor.org 于 2019年5月20日周一 下午3:02写道:

ADD COMMENT
0
Entering edit mode

Hi Yuan

Thanks for quick reply. I will try overnight with 1 or 4 cores to see if I can get it to work. Can you give me an estimate of running time for this (both DMRs and blocks) when only using one core? Are we talking 10 min/1 h/1 day?

/Rasmus

Fra: Yuan Tian [bioc] noreply@bioconductor.org Sendt: 20. maj 2019 16:15 Til: Rasmus Rydbirk rasmus.rydbirk@regionh.dk Emne: [bioc] A: bumphunter stalls on "Smoothing coefficients"

Activity on a post you are following on support.bioconductor.orghttps://support.bioconductor.org

User Yuan Tianhttps://support.bioconductor.org/u/13904/ wrote Answer: bumphunter stalls on "Smoothing coefficients"https://support.bioconductor.org/p/121241/#121242:

Someone in my lab reported similar error before. But I can not replicate it on my server or laptop. So I eventually solved it by ask her setting cores=1, means not use parallel.

Thus, I suspect the reason is doParallel package.

Best Yuan Tian

rasmus.rydbirk [bioc] noreply@bioconductor.orgnoreply@bioconductor.org 于 2019年5月20日周一 下午3:02写道:

ADD REPLY
0
Entering edit mode

Sorry, I can't estimate it well, cause on my server and computer it works well...so I normally only use parallel settings.

Best Yuan Tian

rasmus.rydbirk [bioc] noreply@bioconductor.org 于 2019年5月20日周一 下午3:18写道:

ADD REPLY
0
Entering edit mode

I can confirm multi-thread problems: Running with 1 core took 10-25 min for both functions. Thanks!

ADD REPLY

Login before adding your answer.

Traffic: 824 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