[BioC] ArrayQualityMetrics: Problem with labels in plots. A bug??
lohith madireddy
lohithreddym at gmail.com
Tue Nov 3 18:37:12 CET 2009
Hi Audrey,
thanks for the addition of the "spatial" argument. Yes, it is the only
function that requires a lot of memory. Other functions are very quick
once we get the aqm.prepdata object. I agree with you that it is not
necessary to make arguments for other plots and will most definitely
confuse the users.
The MA plot as a separate function works fine. I did make a silly
mistake earlier. Do you recommend not to have loess line on these
plots due to the fact that the software reports the abnormal arrays in
the summary already and showing a loess curve might be repetitive and
may be confusing for new users??
Thank you Audrey!
With Best Regards,
Lohith Madireddy
On Tue, Nov 3, 2009 at 4:08 AM, Audrey Kauffmann <audrey at ebi.ac.uk> wrote:
> Hi Lohith,
>
> Thank you for your feedback and suggestions.
> It is true that spatial distributions require a large memory and cause the
> function to fail on some machines (note that with large memory it is most
> of the time possible to run it without problems). This is why I added an
> argument "spatial" that you can set to FALSE if you do not want these
> plots. This is a new feature of the package.
> I am not sure I want to create such an argument for each of the plots as
> it would, I think, make the arrayQualityMetrics function less easy to use,
> which was one the primary goal of it. But as I was aware that the spatial
> distributions would fail on large arrays, I made this one available.
>
> Regarding the failing of the MA plot outside from the wrapper. I do not
> understand why it would not work if it works in the arrayQualityMetrics
> function as the two are the exact same ones. Can you please send me the
> lines of code you used to create ratDTH.ab and the ones you used to call
> the wrapper, so that I am sure of what you are doing?
>
>
> Best wishes,
> Audrey
>
>> Dear Audrey,
>>
>> the labeling works good now. Thanks for the new version..
>> There are certainly memory issues performing spatial plots. I just
>> have 69 arrays with around 9000 features on each array. I can imagine
>> its impossible to perform these plots for large number of arrays of
>> larger sizes.The software mostly never produces spatial plots since it
>> can not allocate some 'xyz' MB. The error is as show below: I have a
>> max of 4095MB allocated to R on a 64 bit system.
>> Error : cannot allocate vector of size 150.1 Mb
>> In addition: There were 50 or more warnings (use warnings() to see the
>> first 50)
>> In addition: Warning messages:
>> 1: In arrayQualityMetrics(ratDTH.ab, do.logtransform = TRUE, outdir =
>> "QA_BeforeRMA_69Chips2", :
>> Cannot draw spatial distribution of intensities
>> 2: In rep.int(1L:ncol(x), rep.int(nrow(x), ncol(x))) :
>> Reached total allocation of 3583Mb: see help(memory.size)
>> 3: In rep.int(1L:ncol(x), rep.int(nrow(x), ncol(x))) :
>> Reached total allocation of 3583Mb: see help(memory.size)
>> 4: In inherits(x, "data.frame") :
>> Reached total allocation of 3583Mb: see help(memory.size)
>> 5: In inherits(x, "data.frame") :
>> Reached total allocation of 3583Mb: see help(memory.size)
>>
>>
>> So, I tried to do spatial plots and MA plots separately creating 'aqm'
>> objects and generating report for these two separately (performing
>> garbage collection in between). I was successful with all except the
>> MAplots. It gives the following error. I can not exactly see what's
>> wrong with the error it gave. Are there any NULL values?? I do not
>> think so. There were no errors when I used the 'arrayQualityMetrics'
>> wrapper function though. Also, as a suggestion it would be very useful
>> also to have a loess curve in each of the MAplots as we get using the
>> affy or affyPLM packages.
>>
>>> maplot.aqm<-aqm.maplot(ratDTH.ab)
>> Error in limits.and.aspect(prepanel.default.xyplot, prepanel = prepanel,
>> :
>> need at least one panel
>> In addition: Warning messages:
>> 1: In is.na(x) : is.na() applied to non-(list or vector) of type 'NULL'
>> 2: In is.na(x) : is.na() applied to non-(list or vector) of type 'NULL'
>>
>> One more request is with the Wrapper function. I felt the option of
>> customizing the report is a good thing as the user can plot only what
>> he needs. But, it would be easier if the wrapper function itself has
>> an option(s) to not plot some of the plots the user is not interested
>> in. In my case, I would not want the 'spatial distribution of feature
>> intensities' plot as it almost always fails to draw all or few of the
>> 69 arrays and I would prefer to do it separately.
>>
>> Awaiting your reply,
>> Thank you
>> Lohith
>>
>>
>>
>> On Fri, Oct 30, 2009 at 1:51 AM, Audrey Kauffmann <audrey at ebi.ac.uk>
>> wrote:
>>>
>>> Dear Jeremy,
>>>
>>> Sorry I did not think about this, I think it takes a couple of days
>>> between the moment the developer modifies the version of the package and
>>> the moment you can actually install it using biocLite. You should be
>>> able to install version 2.4.1 after the week-end.
>>>
>>> Best wishes,
>>> Audrey
>>>
>>> jeremy wilson wrote:
>>>>
>>>> Dear Audrey,
>>>>
>>>> thanks for your reply. I tried re-installing the package from bioc. It
>>>> gave the same results. I later installed 2.5 version of the software.
>>>> This also gave the same plots. There is still the same problem with
>>>> ordering of colors.
>>>> I see you asked me to use version 2.4.1 but I could not find it in the
>>>> release or development version. Am I looking at the wrong places?
>>>> Please let me know and thanks for your time...
>>>>
>>>> Thank you
>>>> Lohith
>>>>
>>>> P.S: I know this might be just a simple bug in the program and I wanted
>>>> to see if I can successfully tweak the R script for the software but I
>>>> did not find the R script in both the release and development
>>>> versions..It would be great to see the code..
>>>>
>>>>
>>>> On Thu, Oct 29, 2009 at 2:41 AM, Audrey Kauffmann <audrey at ebi.ac.uk
>>>> <mailto:audrey at ebi.ac.uk>> wrote:
>>>>
>>>> Dear Jeremy,
>>>>
>>>> Thank you for reporting this. You were right there was a problem
>>>> in the code for the ordering of the colours.
>>>> I have now corrected it, so arrayQualityMetrics version 2.4.1
>>>> should be working fine.
>>>> Please let me know if you keep experiencing problems.
>>>>
>>>> Best wishes,
>>>> Audrey
>>>>
>>>> jeremy wilson wrote:
>>>>
>>>> Dear BioConductors,
>>>>
>>>> I am using arrayQualityMetrics software package to do the QA.
>>>> I very much
>>>> like all the QA plots it produces including the latest PCA
>>>> plot in one
>>>> report. I am using the latest version of R(2.10) and BioC
>>>> (2.5). I am facing
>>>> a problem with inappropriate labeling of covariates. Please
>>>> see the attached
>>>> plots.
>>>> I have total of 69 samples. I asked the software to label the
>>>> plots based on
>>>> a covariate, the summary of which is as below:
>>>>
>>>> Notation total number of samples in order
>>>> V 23
>>>> G 38
>>>> N 8
>>>>
>>>> The problem is that it does not label the graphs correctly in
>>>> the plots!! In
>>>> the plots, the first set of samples correspond to V, the next
>>>> 38 samples for
>>>> G and the last 8 for N. In the boxplot for example, the G was
>>>> labelled RED
>>>> and this should correspond to 38 samples but in the plot, the
>>>> boxplots
>>>> corresponding to these 38 samples were colored GREEN. The ones
>>>> that were
>>>> labeled RED correspond to 8 'N' samples which should be
>>>> represented in blue
>>>> according to the label at the top of plot. Likewise for the
>>>> other two. I
>>>> think it can be better understood at the first glance of the
>>>> plots.
>>>>
>>>> The phenotype of the samples was given correctly.
>>>> Following is the code I used for generating the aqm report.
>>>>
>>>> library(arrayQualityMetrics)
>>>> arrayQualityMetrics(rat.ab, do.logtransform=TRUE, outdir="QA",
>>>> intgroup="Material", grouprep=TRUE)
>>>>
>>>> I have attached only a few plots. I can produce all of them if
>>>> it helps...
>>>> It might take some time for the post to reach the mailing list
>>>> as the
>>>> earlier one I posted with a few more plots was held for
>>>> moderator's approval
>>>> as the size of the message body is over the limit!
>>>>
>>>> Please let me know what is wrong.
>>>> Thanks in advance!
>>>>
>>>> ------------------------------------------------------------------------
>>>>
>>>>
>>>>
>>>> ------------------------------------------------------------------------
>>>>
>>>>
>>>> ------------------------------------------------------------------------
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> Bioconductor mailing list
>>>> Bioconductor at stat.math.ethz.ch
>>>> <mailto:Bioconductor at stat.math.ethz.ch>
>>>> https://stat.ethz.ch/mailman/listinfo/bioconductor
>>>> Search the archives:
>>>>
>>>> http://news.gmane.org/gmane.science.biology.informatics.conductor
>>>>
>>>>
>>>> -- Audrey Kauffmann
>>>> EMBL - EBI
>>>> Cambridge UK
>>>> +44 (0) 1223 492 631 http://www.ebi.ac.uk/~audrey
>>>> <http://www.ebi.ac.uk/%7Eaudrey>
>>>>
>>>>
>>>
>>> --
>>> Audrey Kauffmann
>>> EMBL - EBI
>>> Cambridge UK
>>> +44 (0) 1223 492 631 http://www.ebi.ac.uk/~audrey
>>>
>>
>
>
> Audrey Kauffmann
> EMBL - EBI
> Cambridge UK
> +44 (0) 1223 492 631
> http://www.ebi.ac.uk/~audrey
>
>
>
More information about the Bioconductor
mailing list