Is there any recommended pipeline or procedure for Illumina methylation 450K array analysis. and What we can do as a analyst when we obtain some methylation 450K array for some paired case and control. such as
1, differential methylation analysis
2, Pathway and Go analysis
I would start with the IDAT files and run methylumi in R (http://www.bioconductor.org/packages/release/bioc/html/methylumi.html). After that depends on what you are looking for. You could use some tools for differential subgroup discover and differential methyaltion on GenePattern (NMF, SOM, etc.), ConsensusClusterPlus in R (http://www.bioconductor.org/packages/release/bioc/html/ConsensusClusterPlus.html), SAM analysis in MeV (http://www.tm4.org/mev.html).
As for pathways, you could always use some classic ones like DAVID or Ingenuity once you have the deferentially methylated genes, but there is also a neat package in R called Functional Epigenetic Modules (FEM: http://www.bioconductor.org/packages/release/bioc/html/FEM.html), which is made to combine 450K methylation data with gene expression profiling but I think you could use its network enrichment tool without the combination. Also, I believe the FEM package can do differential methylation analysis.
There are also a few potentially useful 450K specific packages here: http://www.bioconductor.org/packages/2.14/bioc/
There are some packages from bioconductor e.g. "methylumi", "lumi", "minfi". There is one more which is "IMA".
I do not like to use IMA , because it demands Beadstudio exported file in specific format, for all kinds of microarray data I always start with raw files ( .CEL, .DAT, in your case its IDAT format), it takes 3-5 minutes to process the data.
You can just go through methyllumi and lumi R packages from bioconductor ( my favourites)
They create a R object based on the eSet class, but they all use different methods for normalization (just go through it),
in methyllumi there is function normalizeMethyLumiSet() and in lumi has function for color bias correction, background adjustment and normalization. Unfortunately, both of these methods are criticized in some ways that
they don't seem to be so great to process Illumina 450K Data.
Consequently, after following some bioinformatics forum and discussing with colleagues in my institute, I prefer "minfi" package, it does normalization accounting for differences in type I/type II probes
its simple commands are following in R:
######
setwd("")
raw.dat
To define differentially methylated regions (DMRs) you could use my colleagues tool, DMRcate. We find it very effective. It has just been published in Epigenetics and Chromatin and is available on Bioconductor. http://www.epigeneticsandchromatin.com/content/8/1/6/abstract
For gene ontology analysis, it pays to heed the advice summarised in this paper: http://www.ncbi.nlm.nih.gov/pubmed/23732277
I tried to run this function sitetest to perform Site-level Differential Methylation Analysis using IMA package but I got error message.
sitetestALL = sitetest(dataf,gcase="KO",gcontrol="WT",testmethod ="wilcox" ,Padj="BH", rawpcut = NULL,adjustpcut =NULL,betadiffcut = NULL,paired = FALSE) and I got this error message: Error in wilcox.test.default(x[1:length(lev1)], x[(length(lev1) + 1):(length(lev1) + : not enough (finite) 'x’ observations
Can you help me to solve this problem?
A great place to start is RnBeads:
http://rnbeads.mpi-inf.mpg.de/index.php
another great resource is Marmal-aid:
http://marmal-aid.org/
to assess deferential methylation across cell types fro any CpGs of interest.
Hi everyone,
It's worth joining this forum if your looking for advice or ideas. Can be very helpful.
A
https://groups.google.com/forum/#!forum/epigenomicsforum
Hi,
All the above comments are really helpful..
Thanks for sharing.
F
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics
- Bioinformatic Tools