Question

Three microarray questions: Oligo package, custom microarrays and makeContrasts

0

Entering edit mode

8.1 years ago

R.Blues ▴ 160

Hello everyone,

I am dealing with different microarray analysis in R, and I have three questions. I took a look, and I haven't found an exact answer for any of them. If there is already one, sorry :(

-Concerning the package oligo: Is there any way I could create an annotation for a microarray using CLF, PGF files and etc.? I am working with Human Gene 2.0 ST arrays and I cannot install the XPS package due to certain library problems when installing ROOT (sigh). Is there any alternative? I need to be able to annotate each probe for ulterior analysis.

-I also have to analyze custom gene expression microarrays from Agilent, using R. Is there any special way to procede? I imagine that after the DE analysis, I will have to blast the sequences of the probes to my genome after analyzing it, but I would appreciate any advice or protocol. Is there any special library I should use for the DE analysis, being it a custom microarray from Agilent?

-Also, a (maybe) stupid question. When using makeContrasts from limma for three groups, one of them a control, it does not say which group differ in what, and you have to use pairwise comparisons after that. Then... what is the meaning of using the ANOVA for these three groups?

Thank you all for your help, and sorry for bothering you. Have a nice day!!

R microarray oligo gene analysis • 2.5k views

ADD COMMENT • link updated 8.1 years ago by warren-mcgee ▴ 40 • written 8.1 years ago by R.Blues ▴ 160

score 2 · Accepted Answer · 2016-04-01

Hi R.Blues,

Here are my responses to your questions, in case you're still stuck:

1) Working with package oligo: I would recommend that you look into the pdInfoBuilder R Package. It allows you to take the CLF and PGF files for an array, and create a database that can then be used with the oligo package. See the documentation there to get you started, and certainly leave a comment if you have additional questions.

2) I'm not going to be of the greatest help here, but here is my suggestion: first place to check would be the agilp R package, which was designed for Agilent arrays, and may be helpful with a custom array. Otherwise, you can attempt to convert the Agilent files into Affymetrix-like files. CEL files for the raw data give you information about the intensity for each spot on the microarray (documentation about CEL format). From there, the CLF maps probes to the locations found in the CEL file (CLF documentation), and the PGF file maps probes to probesets (PGF documentation). Alternatively, a CDF file combines both CLF and PGF features into one file, and this is used for expression arrays rather than exon arrays (CDF documentation). If Agilent gave you information about any of these, you can see if you can convert their files into the Affymetrix formats. Those could then be easily imported into R. Otherwise, what you can do is map the probes to the genome using bowtie or BWA (or any other short-read aligner), then do a probe-level analysis.

3) the ANOVA for more than 3 groups tests the null hypothesis that none of the three categories changed. If the F test produces a significant p-value, that means that at least one of the 3 groups changed with respective to the others. MakeContrasts would then allow you to do the post-hoc testing to see which pairwise comparison was significant. I would be careful about how to correct for multiple hypotheses in scenario of more than 3 groups; see Limma's documentation for their recommendation (specifically, section 13.3 of the Limma User Guide).

Hope that helps!

Warren