Question

Is there any tutorial to perform Mouse/Human Transcriptome microarrays 1.0 ?

0

Entering edit mode

jacorvar ▴ 40

@jacorvar-8972

Last seen 6 months ago

European Union

Dear bioC community,

I have spent more than a whole day searching for tutorials/manuals to analyse mouse transcriptome arrays 1.0. I want to perform a DE analysis gene-wise but considering only exons, and there's apparently no information in the web. And what's the difference between mta10stprobeset and mta10sttranscriptioncluster? Does one consider the probes and the other the probesets? In that case, how can I distinguish probes mapping to exons, junctions and others? Is there any function for that? When I use mta10sttrancriptioncluster and limma to analyse DE genes, about 50% are DE, which makes no sense. Could any kind soul help me please?

Thanks in advance

mta10 tutorial affy • 1.7k views

ADD COMMENT • link updated 8.6 years ago by James W. MacDonald 65k • written 8.6 years ago by jacorvar ▴ 40

score 3 · Accepted Answer · 2015-10-13

3

Entering edit mode

James W. MacDonald 65k

@james-w-macdonald-5106

Last seen 47 minutes ago

United States

If you summarize your array data doing something like

dat <- read.celfiles(filenames = list.celfiles())
eset <- rma(dat)

Then you will be summarizing the data at the 'core' or transcript level. In other words, you will be taking all probes that align to any exonic region (not junction probes) for each gene, and then summarizing them using RMA. If you do this, then you are treating your HTA array as if it were a Gene ST array, in which case any tutorial you find for analyzing microarray data will be directly applicable. If you are using limma, then the limma User's Guide is your friend.

The mta10stprobeset.db and mta10sttranscriptcluster.db packages are for when you summarize at the 'probeset' and 'core' levels, respectively. So you want the latter.

As for your analysis with 50% DE, without knowing more about your study and how you modeled the data, that may or may not make sense to me.

ADD COMMENT • link 8.6 years ago James W. MacDonald 65k

0

Entering edit mode

So "core" means I am considering only probes at exon or coding transcript level, i.e., ignoring probes falling at introns? In that case, are the expression values for the probes aggregated per exon?

ADD REPLY • link 8.6 years ago jacorvar ▴ 40

1

Entering edit mode

No, they are aggregated at the gene level. And yes, ignoring introns. That's what I meant when I said taking all probes that align to any exonic region (not junction probes) for each gene, and then summarizing them.

The only intronic probes on an Affy array are in control probesets, and those AFAIK are never combined into anything but the control probesets.

You cannot really summarize at the exon level, with any Affy array. What people usually consider to be the exon level is what Affy calls the 'probeset' level, which is defined as being within a 'probe set region' or PSR. The PSR may in fact cover the entirety of an exon, but it may not, especially if the exon is large, in which case there may be multiple probesets that interrogate different portions of the exon. You could hypothetically combine all the probesets for each exon into a table similar to the core_mps table in the pd.mta.1.0.st package, and then summarize at the exon level if you really cared about such things.

ADD REPLY • link 8.6 years ago James W. MacDonald 65k

1

Entering edit mode

@jacorvar
This image gives you a visual example of why a Probe Selection Region (PSR) may not correspond to an exon.
PSR0100000619.mm and PSR0100000620.mm do have a one to one relationship of PSR -> Exon. However, PSR0100000624.mm and PSR0100000625.mm do not have a simple one to one relationship to a single full exon. PSR0100000624.mm overlaps one full exon [EX010000598.mm] and part of another exon [EX010000599.mm]. PSR0100000625.mm targets the remaining portion of [EX0100000599.mm]

http://i.imgur.com/dCqusbt.png

ADD REPLY • link 8.6 years ago Jason Hubbard ▴ 20