Hello,
I want to do an intergrated network analysis for my Differential
Expressed
genes. So could anyone tell me how can I do. Which package should I
use.
Thanks,
Abdul Rawoof
[[alternative HTML version deleted]]
Abdul,
Can you tell more about your project and what exactly you mean
by integrated network analysis. What are you integrating into your DE
genes? what organism are you working with? What are you trying to
find in
the data. Is the data RNA seq or micro array (and which platform).
bioconductor has many packages that you can use to handle networks,
but the
`best` package depends on your goals / aims / resources.
Best
On Tue, May 7, 2013 at 5:57 AM, Abdul Rawoof <abdul87.edu@gmail.com>
wrote:
> Hello,
>
> I want to do an intergrated network analysis for my Differential
Expressed
> genes. So could anyone tell me how can I do. Which package should I
use.
>
> Thanks,
> Abdul Rawoof
>
> [[alternative HTML version deleted]]
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor@r-project.org
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives:
> http://news.gmane.org/gmane.science.biology.informatics.conductor
>
--
Sam McInturf
[[alternative HTML version deleted]]
Conversation went off the list:
Dear Sam,
I am working with Microarray Data analysis for illumina's WG-DASL
assay
based Human sample.
Actually I want to know Gene-Gene interaction between my
differentially
expressed genes.
Thanks,
Abdul Rawoof
______________________________________________________________________
___________
Abdul,
When you say gene-gene interaction, what exactly do you mean? Are
you
looking for protein-protein binding data for your DE genes? Or are
you
after other genes that have been shown to (or may) be co-regulated?
There are many ways to do analysis, and everything has pros and cons,
and
if you aren't careful your analysis will mean nothing because you
applying
the wrong method/models. In order to answer your question, we must
know
exactly what you want to do and find out. Integrated network analysis
and
gene-gene interaction can have several interpretations. Please, be
verbose
Best,
On Tue, May 7, 2013 at 9:59 AM, Sam McInturf <smcinturf@gmail.com>
wrote:
> Abdul,
> Can you tell more about your project and what exactly you mean
> by integrated network analysis. What are you integrating into your
DE
> genes? what organism are you working with? What are you trying to
find in
> the data. Is the data RNA seq or micro array (and which platform).
>
> bioconductor has many packages that you can use to handle networks,
but
> the `best` package depends on your goals / aims / resources.
>
> Best
>
>
> On Tue, May 7, 2013 at 5:57 AM, Abdul Rawoof
<abdul87.edu@gmail.com>wrote:
>
>> Hello,
>>
>> I want to do an intergrated network analysis for my Differential
Expressed
>> genes. So could anyone tell me how can I do. Which package should I
use.
>>
>> Thanks,
>> Abdul Rawoof
>>
>> [[alternative HTML version deleted]]
>>
>> _______________________________________________
>> Bioconductor mailing list
>> Bioconductor@r-project.org
>> https://stat.ethz.ch/mailman/listinfo/bioconductor
>> Search the archives:
>> http://news.gmane.org/gmane.science.biology.informatics.conductor
>>
>
>
>
> --
> Sam McInturf
>
--
Sam McInturf
[[alternative HTML version deleted]]
Dear Hilary,
Your test for the 3-way interaction is correct, although 3-way
interactions are pretty hard to interpret.
However testing for the 2-way interaction in the presence of a 3-way
interaction does not make statistical sense. This is because the
parametrization of the 2-way interaction as a subset of the 3-way is
somewhat arbitrary. Before you can test the 2-way interaction
species*treatment in a meaningful way you would need to accept that
the
3-way interaction is not necessary and remove it from the model.
In general, I am of the opinion that classical statistical factorial
interation models do not usually provide the most meaningful
parametrizations for genomic experiments. In most cases, I prefer to
fit
the saturated model (a different level for each treatment combination)
and
make specific contrasts. There is some discussion of this in the
limma
User's Guide.
In your case, I guess that you might want to test for
species*treatment
interaction separately at each time point. It is almost impossible to
do
this within the classical 3-way factorial setup. However it is easy
with
the one-way approach I just mentioned, or else you could use:
~Age + Age:Species + Age:Treatment + Age:Species:Treatment
Best wishes
Gordon
> Date: Thu, 9 May 2013 14:55:46 -0400
> From: Hilary Smith <hilary.a.smith.964 at="" nd.edu="">
> To: "bioconductor at r-project.org" <bioconductor at="" r-project.org="">
> Subject: [BioC] Statistics question for multi-factor interaction
test
> in edgeR
>
> Hi. I need to generate two GLM tests of a factorial design with RNA-
Seq
> count data. I have 3 factors with 2 levels apiece (2 species X 2
> treatments X 2 times), and 4 separate replicates each (i.e., we made
a
> total of 2*2*2*4 = 32 separate libraries). Our main interest is in
the
> interaction of species*treatment, as we think species A will alter
gene
> expression in the treatment stress vs. treatment benign, whereas
> species B is expected to show little change. However, we?d like to
also
> do another test of species*treatment*time, because it is possible
that
> the ability of species A to alter gene expression in response to the
> stress treatment may differ at the 1st versus 2nd time point.
>
> I think the way to set this up, is to create a design matrix as
follows,
> with the lrt test with coef 5 giving the differentially expressed
genes
> for the species*treatment test, and coef 8 giving the the
differentially
> expressed gene for the species*treatment*time test (after calling
> topTags that is). Yet to ensure I have the statistics correct, my
> questions are: (1) is this thinking correct, as I don?t see many 3x2
> factorial models to follow, and (2) do I need to set up a reference
> somehow (which I assume would be the set of four samples with
> TreatmentBenign*SpeciesB*Time2, but I?m not fully sure if that is
> correct or needed).
>
> Many thanks in advance for your insight!
> ~Hilary
>
>> designFF <- model.matrix(~Treatment*Species*Age)
>> colnames(designFF)
> [1] "(Intercept)"
> [2] " TreatmentStress"
> [3] "SpeciesA "
> [4] "Time1"
> [5] "TreatmentStress:SpeciesA"
> [6] "TreatmentStress:Time1"
> [7] "SpeciesA:Time1"
> [8] "TreatmentStress:SpeciesA:Time1"
>
> And then to run tests with:
>> fit <- glmFit(y, designFF)
>
>> lrtInteractionStressSpecies <- glmLRT(fitFF, coef=5)
>> lrtInteractionStressSpeciesTime <- glmLRT(fitFF, coef=8)
>
______________________________________________________________________
The information in this email is confidential and
intend...{{dropped:5}}
Hi Gordon,
In a previous email on the list you said:
> testing for the 2-way interaction in the presence of a 3-way
> interaction does not make statistical sense. This is because the
> parametrization of the 2-way interaction as a subset of the 3-way is
> somewhat arbitrary. Before you can test the 2-way interaction
> species*treatment in a meaningful way you would need to accept that
> the 3-way interaction is not necessary and remove it from the model.
Does this mean that it is impossible to test for a 2-way interaction
when your model includes a 3-way interaction term? Or does it just
mean
that the parametrization provided by
"model.matrix(~1+factor1*factor2*factor3)" is such that the 2-way
interaction is not represented by any coefficient, but rather by a
complex contrast?
> I prefer to fit the saturated model (a different level for each
> treatment combination) and make specific contrasts. There is some
> discussion of this in the limma User's Guide.
If I understand correctly here, you are saying that one can fit a
model
where each coefficient represents the abundance for one specific
combination of the 3 factors, as in the Limma User's Guide section
8.5.2
"Analysing as for a Single Factor". In other words, one could do
"model.matrix(~0+factor1:factor2:factor3)" and this would be an
alternate parametrization of the same design. And with this
parametrization, all the 2- and 3-way interaction terms (and simple
pairwise comparisons) can easily be tested from the single full 3-way
interaction by specifying the appropriate contrasts. Do I understand
correctly?
-Ryan Thompson
[[alternative HTML version deleted]]
On Sun, 12 May 2013, Ryan C. Thompson wrote:
> Hi Gordon,
>
> In a previous email on the list you said:
>
>> testing for the 2-way interaction in the presence of a 3-way
>> interaction does not make statistical sense. This is because the
>> parametrization of the 2-way interaction as a subset of the 3-way
is
>> somewhat arbitrary. Before you can test the 2-way interaction
>> species*treatment in a meaningful way you would need to accept that
the
>> 3-way interaction is not necessary and remove it from the model.
>
> Does this mean that it is impossible to test for a 2-way interaction
> when your model includes a 3-way interaction term?
It is mathematically possible but has no scientific meaning. This is
called the marginality principle in linear models:
http://en.wikipedia.org/wiki/Principle_of_marginality
> Or does it just mean that the parametrization provided by
> "model.matrix(~1+factor1*factor2*factor3)" is such that the 2-way
> interaction is not represented by any coefficient, but rather by a
> complex contrast?
The same principle applies regardless of the parametrization.
>> I prefer to fit the saturated model (a different level for each
treatment
>> combination) and make specific contrasts. There is some discussion
of this
>> in the limma User's Guide.
> If I understand correctly here, you are saying that one can fit a
model
> where each coefficient represents the abundance for one specific
> combination of the 3 factors, as in the Limma User's Guide section
8.5.2
> "Analysing as for a Single Factor". In other words, one could do
> "model.matrix(~0+factor1:factor2:factor3)" and this would be an
> alternate parametrization of the same design. And with this
> parametrization, all the 2- and 3-way interaction terms (and simple
> pairwise comparisons) can easily be tested from the single full
3-way
> interaction by specifying the appropriate contrasts. Do I understand
> correctly?
Yes, I am recommending the group mean parametrization, as in the limma
User's Guide Section 8.5.2 or edgeR User's Guide Section 3.3.1.
I recommend this parametrization because each contrast that is drawn
has
an explicit meaning in terms of comparisons of groups and can be
interpretted on its own terms.
The original poster did what I intended.
Best
Gordon
> -Ryan Thompson
>
______________________________________________________________________
The information in this email is confidential and
intend...{{dropped:4}}
Dear Ryan,
The marginality principle is most easily understood in a 2-way
factorial
model. Suppose you have a 2x2 factorial experiment with two genotypes
and
two treatments (active vs control).
If a two-way interaction exists, then this means that the treatment
effect
is different for the genotypes. It makes no sense to test for a
"treatment effect" in this situation (even though mathematical models
allow you to do so) because there is no consistent treatment effect
without specifying the genotype.
On the other hand, it always meaningful to test for a treatment effect
in
the two genotypes separately, and then to ask whether the two
treatment
effects are consistent or different.
In a 3-way factorial model, a 3-way interaction means that the
experiment
cannot be reduced to 2-way marginals in any meaningful way.
Best wishes
Gordon
On Mon, 13 May 2013, Gordon K Smyth wrote:
> On Sun, 12 May 2013, Ryan C. Thompson wrote:
>
>> Hi Gordon,
>>
>> In a previous email on the list you said:
>>
>>> testing for the 2-way interaction in the presence of a 3-way
interaction
>>> does not make statistical sense. This is because the
parametrization of
>>> the 2-way interaction as a subset of the 3-way is somewhat
arbitrary.
>>> Before you can test the 2-way interaction species*treatment in a
>>> meaningful way you would need to accept that the 3-way interaction
is not
>>> necessary and remove it from the model.
>>
>> Does this mean that it is impossible to test for a 2-way
interaction when
>> your model includes a 3-way interaction term?
>
> It is mathematically possible but has no scientific meaning. This
is called
> the marginality principle in linear models:
>
> http://en.wikipedia.org/wiki/Principle_of_marginality
>
>> Or does it just mean that the parametrization provided by
>> "model.matrix(~1+factor1*factor2*factor3)" is such that the 2-way
>> interaction is not represented by any coefficient, but rather by a
complex
>> contrast?
>
> The same principle applies regardless of the parametrization.
>
>>> I prefer to fit the saturated model (a different level for each
treatment
>>> combination) and make specific contrasts. There is some discussion
of this
>>> in the limma User's Guide.
>
>> If I understand correctly here, you are saying that one can fit a
model
>> where each coefficient represents the abundance for one specific
>> combination of the 3 factors, as in the Limma User's Guide section
8.5.2
>> "Analysing as for a Single Factor". In other words, one could do
>> "model.matrix(~0+factor1:factor2:factor3)" and this would be an
alternate
>> parametrization of the same design. And with this parametrization,
all the
>> 2- and 3-way interaction terms (and simple pairwise comparisons)
can easily
>> be tested from the single full 3-way interaction by specifying the
>> appropriate contrasts. Do I understand correctly?
>
> Yes, I am recommending the group mean parametrization, as in the
limma User's
> Guide Section 8.5.2 or edgeR User's Guide Section 3.3.1.
>
> I recommend this parametrization because each contrast that is drawn
has an
> explicit meaning in terms of comparisons of groups and can be
interpretted on
> its own terms.
>
> The original poster did what I intended.
>
> Best
> Gordon
>
>> -Ryan Thompson
>>
>
______________________________________________________________________
The information in this email is confidential and
intend...{{dropped:4}}
Dear Gordon,
I understand now. The 3-way experiment cannot be reduced to 2-way
marginals, but one can still test for 2-way interactions within each
level of the 3rd factor. That matches with my own previous
understanding, and my confusion was just one of terminology. Thank you
for the clarification!
-Ryan
On Sun May 12 18:29:20 2013, Gordon K Smyth wrote:
> Dear Ryan,
>
> The marginality principle is most easily understood in a 2-way
> factorial model. Suppose you have a 2x2 factorial experiment with
two
> genotypes and two treatments (active vs control).
>
> If a two-way interaction exists, then this means that the treatment
> effect is different for the genotypes. It makes no sense to test
for
> a "treatment effect" in this situation (even though mathematical
> models allow you to do so) because there is no consistent treatment
> effect without specifying the genotype.
>
> On the other hand, it always meaningful to test for a treatment
effect
> in the two genotypes separately, and then to ask whether the two
> treatment effects are consistent or different.
>
> In a 3-way factorial model, a 3-way interaction means that the
> experiment cannot be reduced to 2-way marginals in any meaningful
way.
>
> Best wishes
> Gordon
>
>
> On Mon, 13 May 2013, Gordon K Smyth wrote:
>
>> On Sun, 12 May 2013, Ryan C. Thompson wrote:
>>
>>> Hi Gordon,
>>>
>>> In a previous email on the list you said:
>>>
>>>> testing for the 2-way interaction in the presence of a 3-way
>>>> interaction does not make statistical sense. This is because the
>>>> parametrization of the 2-way interaction as a subset of the 3-way
>>>> is somewhat arbitrary. Before you can test the 2-way interaction
>>>> species*treatment in a meaningful way you would need to accept
that
>>>> the 3-way interaction is not necessary and remove it from the
model.
>>>
>>> Does this mean that it is impossible to test for a 2-way
interaction
>>> when your model includes a 3-way interaction term?
>>
>> It is mathematically possible but has no scientific meaning. This
is
>> called the marginality principle in linear models:
>>
>> http://en.wikipedia.org/wiki/Principle_of_marginality
>>
>>> Or does it just mean that the parametrization provided by
>>> "model.matrix(~1+factor1*factor2*factor3)" is such that the 2-way
>>> interaction is not represented by any coefficient, but rather by a
>>> complex contrast?
>>
>> The same principle applies regardless of the parametrization.
>>
>>>> I prefer to fit the saturated model (a different level for each
>>>> treatment combination) and make specific contrasts. There is some
>>>> discussion of this in the limma User's Guide.
>>
>>> If I understand correctly here, you are saying that one can fit a
>>> model where each coefficient represents the abundance for one
>>> specific combination of the 3 factors, as in the Limma User's
Guide
>>> section 8.5.2 "Analysing as for a Single Factor". In other words,
>>> one could do "model.matrix(~0+factor1:factor2:factor3)" and this
>>> would be an alternate parametrization of the same design. And with
>>> this parametrization, all the 2- and 3-way interaction terms (and
>>> simple pairwise comparisons) can easily be tested from the single
>>> full 3-way interaction by specifying the appropriate contrasts. Do
I
>>> understand correctly?
>>
>> Yes, I am recommending the group mean parametrization, as in the
>> limma User's Guide Section 8.5.2 or edgeR User's Guide Section
3.3.1.
>>
>> I recommend this parametrization because each contrast that is
drawn
>> has an explicit meaning in terms of comparisons of groups and can
be
>> interpretted on its own terms.
>>
>> The original poster did what I intended.
>>
>> Best
>> Gordon
>>
>>> -Ryan Thompson
>>>
>>
>
>
______________________________________________________________________
> The information in this email is confidential and
inte...{{dropped:6}}
