Entering edit mode
Heidi Dvinge
★
2.0k
@heidi-dvinge-2195
Last seen 10.2 years ago
Hi Mala,
> Hi Heidi,
>
> We have some question regarding the ddCt and Fold change value
> calculation in HTqPCR. On creating contrasts, if we have Mt-Wt
(Mt=mutant,
> Wt= WildType), does the +ve fold change suggest that it is higher in
the
> Mutant or Wild Type.
I assume you're using limmaCtData. In this case Mt-Wt literally means
mutant minus wild type. I.e. Wt is used as the "Calibrator" and mutant
as
the "Target" that form part of the output from limmaCtData.
If in doubt, it's always good to check manually. For example:
> # Load example preprocessed data
> data(qPCRpros)
> samples <- read.delim(file.path(system.file("exData",
package="HTqPCR"),
"files.txt"))
> # Define design and contrasts
> design <- model.matrix(~0+samples$Treatment)
> colnames(design) <- c("Control", "LongStarve", "Starve")
> contrasts <- makeContrasts(LongStarve-Control, LongStarve-Starve,
Starve-Control, levels=design)
> # The actual test
> diff.exp <- limmaCtData(qPCRpros, design=design,
contrasts=contrasts)
> # Some of the results
> diff.exp[["LongStarve - Control"]][1:3,]
genes feature.pos t.test p.value adj.p.value ddCt
FC meanTarget
72 Gene47 C24 -8.330294 0.0004927066 0.06306645 -8.320716
319.73136 26.61973
251 Gene130 K11 -10.426733 0.0001758778 0.06306645 -6.569095
94.94993 28.37136
334 Gene165 N22 -8.348398 0.0004878740 0.06306645 -6.621950
98.49306 23.84328
meanCalibrator categoryTarget categoryCalibrator
72 34.94045 OK Undetermined
251 34.94045 OK Undetermined
334 30.46523 OK OK
> # Checking the actual expression values for example gene
> getCt(qPCRpros)["Gene47", samples$Treatment=="Control"]
sample1 sample4
34.90896 34.97194
> getCt(qPCRpros)["Gene47", samples$Treatment=="LongStarve"]
sample2 sample3
27.40480 25.83467
> # These correspond to the values in meanTarget and meanCalibrator
ddCt is actually the log2 fold change as calculated by limma (the
"logFC"
column if you're familiar with that package. As mentioned in the
documentation, FC=2^(-ddCt). (Personally, I'm not a big fan of using
this
FC value, but some of my collaborators initially requested that it be
included.)
HTH
\Heidi
> And how is ddCT values used in fold change
> calculation?
>
> Thanks.
> Mala
>
> -----Original Message-----
> From: Heidi Dvinge [mailto:heidi at ebi.ac.uk]
> Sent: Thursday, May 31, 2012 5:44 AM
> To: Sinha, Mala
> Cc: bioconductor at r-project.org
> Subject: RE: HTqPCR
>
> Hi Mala,
>
>> Hi,
>>
>> Will the results change for the Contrast Analysis from limFit and
>> limmaCtData?
>>
> It should be the same, provided you use the same input data (for
example
> with ndups if you have multiple Ct values for each gene).
>
> If you're in doubt, you can always type limmaCtData, without the
brackets,
> to check the code for the function and see how lmFit and eBayes are
being
> used.
>
> Best,
> \Heidi
>
>> Thanks.
>> Mala
>>
>> -----Original Message-----
>> From: Sinha, Mala
>> Sent: Tuesday, May 29, 2012 9:07 AM
>> To: 'Heidi Dvinge'
>> Subject: RE: HTqPCR
>>
>> Thanks, will try using limFit with your suggestion.
>>
>> Mala
>>
>> -----Original Message-----
>> From: Heidi Dvinge [mailto:heidi at ebi.ac.uk]
>> Sent: Monday, May 28, 2012 3:38 AM
>> To: Sinha, Mala
>> Cc: bioconductor at r-project.org
>> Subject: Re: HTqPCR
>>
>> Dear Mala,
>>
>>> Hello Dr. Dvinge,
>>>
>>> I have been using HTqPCR 1.8.0 to analyze 96 well qPCR arrays. My
>>> experiment has 2 time point 3hrs, 18hrs and 2 treatments CT, TC
and
>>> TCATP
>>> (2X3 factorial design). I was able to run the analysis using
>>> limmaCtData and export the output file for each contrast.
>>>
>>> Question 2
>>> *************************************************************
>>> Is there an equivalent method like TopTable in limma to extract
>>> results which shows the genes that are changing across the
>>> experimental condition (like F-statistics in ANOVA or F.p.value in
>>> limma). We can get this information using the write.fit() in
limma.
>>>
>> Actually no, I never thought about incorporating this. The standard
>> output gives a (slightly modified) version of the results from
>> topTable for each of the coefficients in contrast, as well as
>> decideTests. But not the output from topTableF/topTable with
coef=NULL.
>>
>> If this is of interest, I can add it to future versions of HTqPCR?
>>
>> Alternatively, you can just extract the Ct data (using exprs or
getCt)
>> and run lmFit/eBayes/topTableF directly. Although this of course
won't
>> take the quality of the Ct data into account, i.e. you won't get
the
>> columns indicating whether the target and the calibrator samples
are
>> OK or Undetermined.
>>
>> Best,
>> \Heidi
>>
>>> Runnning limmaCtData without contrast didn't help. See below
>>>
>>>>DE.limmaNoCtr <- limmaCtData(quan_undtrmnfilter2, design=design,
>>>>ndups = 1, spacing = 1)
>>>
>>>> write.table(qDE.limmaNoCtr, file="qDE_limmafitALL.txt", sep="\t")
>>> '
>>> Attached file for this: RCode_HTqPCR.txt ;
>>> limmaCtData_output_no_contrast.xls
>>>
>>>
*********************************************************************
>>> *
>>> *
>>> Thanks for all your help.
>>>
>>> Mala Sinha
>>> Systems Analyst
>>> University of Texas Medical Branch
>>> Galveston, Texas 77573
>>> masinha at utmb.edu<mailto:masinha at="" utmb.edu="">
>>> (409) 747 6872
>>>
>>>
>>
>>
>>
>
>
>