Re: [aroma.affymetrix] unusual copy number analysis result - split copy numbers

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

Re: [aroma.affymetrix] unusual copy number analysis result - split copy numbers

Henrik Bengtsson
Hi.

On Mon, Sep 13, 2010 at 4:19 PM, Patrick <[hidden email]> wrote:

> Hi everyone,
> I'm using AROMA's implementation of the CRMA v2 method to get copy
> number estimates for cancer samples, and I'm getting a very unusual
> result.  Many of the samples have a chromosome where AROMA has called
> primarily copy number gains or losses, and the losses are mixed in
> with each other.  That is, if you plot the probes' intensities by
> their positions on the chromosome, you see large stretches (~10,000
> probes) where there are no intensities in the normal range
> (corresponding to no gain or loss), and there are intensities both
> above and below the normal range, mixed in with each other along the
> chromosome.  It is as if the plots for a chromosome with a long
> deletion and a chromosome with a long addition were laid atop each
> other.

It is not 100% clear from your description what you are observing.
Note that it is possible to attach PNGs to messages sent to this
mailing list as long as you send it as an email (not via the web
interface).  What chip type are you working on and do you look at a
public data set?  Have you used CRMAv2 on other data sets without a
problem?

FYI, Johan Staaf reported odd looking copy number results that are
reproducible and very odd. See thread 'Problems with Affymetrix 250K
Sty2 arrays after CRMAv2 analysis' on June 23-August 5, 2010, cf.
http://goo.gl/FGVe.  From the discussion in that thread, it seems to
have something to do with annotation issues, but it is still to be
solved.  Is that what you are experiencing?

> It seems implausible that a cancer sample would have copy number gains
> and losses mixed in with each other in such small intervals, over such
> large stretches of chromosome, without any loci having the usual 2
> copies, so I suspect the normalization or the affy array is the source
> of this phenomenon.  I looked at the data without using AROMA, and the
> phenomenon was not evident.  I re-normalized the data 3 times, each
> time using only one step of the AROMA normalization in isolation.  The
> base position normalization step produced the phenomenon, and the
> allele crosstalk calibration and the fragment length normalization
> steps did not.

What would help troubleshooting is if you could see other software
such as dChip of Affymetrix GTC produces the same oddities.  If they
do, we know for sure it's something odd with the annotation.

/Henrik

> Any thoughts on what I'm seeing and on how the base pair normalization
> could cause it would be very appreciated.
> Thanks,
> Patrick
>
> --
> When reporting problems on aroma.affymetrix, make sure 1) to run the latest version of the package, 2) to report the output of sessionInfo() and traceback(), and 3) to post a complete code example.
>
>
> You received this message because you are subscribed to the Google Groups "aroma.affymetrix" group with website http://www.aroma-project.org/.
> To post to this group, send email to [hidden email]
> To unsubscribe and other options, go to http://www.aroma-project.org/forum/
>

--
When reporting problems on aroma.affymetrix, make sure 1) to run the latest version of the package, 2) to report the output of sessionInfo() and traceback(), and 3) to post a complete code example.


You received this message because you are subscribed to the Google Groups "aroma.affymetrix" group with website http://www.aroma-project.org/.
To post to this group, send email to [hidden email]
To unsubscribe and other options, go to http://www.aroma-project.org/forum/
Reply | Threaded
Open this post in threaded view
|

Re: [aroma.affymetrix] unusual copy number analysis result - split copy numbers

Patrick Danaher
Hi Henrik,
Thanks for your response.  The thread you suggested ( http://goo.gl/FGVe ) describes my problem well - I'm getting a very similar intensity profile for some chromosomes in some samples.  The attached png shows the problem (red dots are intensities; the black dots are from a copy number calling problem and can be ignored).  The second figure plots the called intensities against normal reference intensities for the same loci.  
As for your specific questions: I've never used CRMAv2 before, my dataset isn't public, and it's an affy 6.0 chip.  
Do you think annotation issues would cause this problem only in a small subset of my samples?
Thanks,
Patrick

On Sun, Sep 26, 2010 at 1:36 PM, Henrik Bengtsson <[hidden email]> wrote:
Hi.

On Mon, Sep 13, 2010 at 4:19 PM, Patrick <[hidden email]> wrote:
> Hi everyone,
> I'm using AROMA's implementation of the CRMA v2 method to get copy
> number estimates for cancer samples, and I'm getting a very unusual
> result.  Many of the samples have a chromosome where AROMA has called
> primarily copy number gains or losses, and the losses are mixed in
> with each other.  That is, if you plot the probes' intensities by
> their positions on the chromosome, you see large stretches (~10,000
> probes) where there are no intensities in the normal range
> (corresponding to no gain or loss), and there are intensities both
> above and below the normal range, mixed in with each other along the
> chromosome.  It is as if the plots for a chromosome with a long
> deletion and a chromosome with a long addition were laid atop each
> other.

It is not 100% clear from your description what you are observing.
Note that it is possible to attach PNGs to messages sent to this
mailing list as long as you send it as an email (not via the web
interface).  What chip type are you working on and do you look at a
public data set?  Have you used CRMAv2 on other data sets without a
problem?

FYI, Johan Staaf reported odd looking copy number results that are
reproducible and very odd. See thread 'Problems with Affymetrix 250K
Sty2 arrays after CRMAv2 analysis' on June 23-August 5, 2010, cf.
http://goo.gl/FGVe.  From the discussion in that thread, it seems to
have something to do with annotation issues, but it is still to be
solved.  Is that what you are experiencing?

> It seems implausible that a cancer sample would have copy number gains
> and losses mixed in with each other in such small intervals, over such
> large stretches of chromosome, without any loci having the usual 2
> copies, so I suspect the normalization or the affy array is the source
> of this phenomenon.  I looked at the data without using AROMA, and the
> phenomenon was not evident.  I re-normalized the data 3 times, each
> time using only one step of the AROMA normalization in isolation.  The
> base position normalization step produced the phenomenon, and the
> allele crosstalk calibration and the fragment length normalization
> steps did not.

What would help troubleshooting is if you could see other software
such as dChip of Affymetrix GTC produces the same oddities.  If they
do, we know for sure it's something odd with the annotation.

/Henrik

> Any thoughts on what I'm seeing and on how the base pair normalization
> could cause it would be very appreciated.
> Thanks,
> Patrick
>
> --
> When reporting problems on aroma.affymetrix, make sure 1) to run the latest version of the package, 2) to report the output of sessionInfo() and traceback(), and 3) to post a complete code example.
>
>
> You received this message because you are subscribed to the Google Groups "aroma.affymetrix" group with website http://www.aroma-project.org/.
> To post to this group, send email to [hidden email]
> To unsubscribe and other options, go to http://www.aroma-project.org/forum/
>

--
When reporting problems on aroma.affymetrix, make sure 1) to run the latest version of the package, 2) to report the output of sessionInfo() and traceback(), and 3) to post a complete code example.


You received this message because you are subscribed to the Google Groups "aroma.affymetrix" group with website http://www.aroma-project.org/.
To post to this group, send email to [hidden email]
To unsubscribe and other options, go to http://www.aroma-project.org/forum/

--
When reporting problems on aroma.affymetrix, make sure 1) to run the latest version of the package, 2) to report the output of sessionInfo() and traceback(), and 3) to post a complete code example.
 
 
You received this message because you are subscribed to the Google Groups "aroma.affymetrix" group with website http://www.aroma-project.org/.
To post to this group, send email to [hidden email]
To unsubscribe and other options, go to http://www.aroma-project.org/forum/

strange intensity profile.PNG (254K) Download Attachment
intensityXreference.PNG (79K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [aroma.affymetrix] unusual copy number analysis result - split copy numbers

Henrik Bengtsson-3
Hi.

On Tue, Sep 28, 2010 at 11:54 AM, Patrick Danaher
<[hidden email]> wrote:
> Hi Henrik,
> Thanks for your response.  The thread you suggested ( http://goo.gl/FGVe )
> describes my problem well - I'm getting a very similar intensity profile for
> some chromosomes in some samples.  The attached png shows the problem (red
> dots are intensities; the black dots are from a copy number calling problem
> and can be ignored).  The second figure plots the called intensities against
> normal reference intensities for the same loci.
> As for your specific questions: I've never used CRMAv2 before, my dataset
> isn't public, and it's an affy 6.0 chip.

What's your chip type?  The other thread reported problems on
Mapping250K_Sty though labelled as "Sty 2" and I don't know what the
"2" means there.

> Do you think annotation issues would cause this problem only in a small
> subset of my samples?

When I say annotation issues, I really mean that if the CDF for the
chip type is not the correct one, you might pick up the wrong probe
signals, especially for SNPs, e.g. PM_A may get the value of a total
CN probe once in a while, say.  It could be a software/annotation bug
in the Affymetrix DAT to CEL file conversion and so on.  That's why it
is crucial to know more about the chip used.

I also recommend that you try dChip and/or Affymetrix GTC, if possible.

/Henrik

> Thanks,
> Patrick
> On Sun, Sep 26, 2010 at 1:36 PM, Henrik Bengtsson <[hidden email]>
> wrote:
>>
>> Hi.
>>
>> On Mon, Sep 13, 2010 at 4:19 PM, Patrick <[hidden email]>
>> wrote:
>> > Hi everyone,
>> > I'm using AROMA's implementation of the CRMA v2 method to get copy
>> > number estimates for cancer samples, and I'm getting a very unusual
>> > result.  Many of the samples have a chromosome where AROMA has called
>> > primarily copy number gains or losses, and the losses are mixed in
>> > with each other.  That is, if you plot the probes' intensities by
>> > their positions on the chromosome, you see large stretches (~10,000
>> > probes) where there are no intensities in the normal range
>> > (corresponding to no gain or loss), and there are intensities both
>> > above and below the normal range, mixed in with each other along the
>> > chromosome.  It is as if the plots for a chromosome with a long
>> > deletion and a chromosome with a long addition were laid atop each
>> > other.
>>
>> It is not 100% clear from your description what you are observing.
>> Note that it is possible to attach PNGs to messages sent to this
>> mailing list as long as you send it as an email (not via the web
>> interface).  What chip type are you working on and do you look at a
>> public data set?  Have you used CRMAv2 on other data sets without a
>> problem?
>>
>> FYI, Johan Staaf reported odd looking copy number results that are
>> reproducible and very odd. See thread 'Problems with Affymetrix 250K
>> Sty2 arrays after CRMAv2 analysis' on June 23-August 5, 2010, cf.
>> http://goo.gl/FGVe.  From the discussion in that thread, it seems to
>> have something to do with annotation issues, but it is still to be
>> solved.  Is that what you are experiencing?
>>
>> > It seems implausible that a cancer sample would have copy number gains
>> > and losses mixed in with each other in such small intervals, over such
>> > large stretches of chromosome, without any loci having the usual 2
>> > copies, so I suspect the normalization or the affy array is the source
>> > of this phenomenon.  I looked at the data without using AROMA, and the
>> > phenomenon was not evident.  I re-normalized the data 3 times, each
>> > time using only one step of the AROMA normalization in isolation.  The
>> > base position normalization step produced the phenomenon, and the
>> > allele crosstalk calibration and the fragment length normalization
>> > steps did not.
>>
>> What would help troubleshooting is if you could see other software
>> such as dChip of Affymetrix GTC produces the same oddities.  If they
>> do, we know for sure it's something odd with the annotation.
>>
>> /Henrik
>>
>> > Any thoughts on what I'm seeing and on how the base pair normalization
>> > could cause it would be very appreciated.
>> > Thanks,
>> > Patrick
>> >
>> > --
>> > When reporting problems on aroma.affymetrix, make sure 1) to run the
>> > latest version of the package, 2) to report the output of sessionInfo() and
>> > traceback(), and 3) to post a complete code example.
>> >
>> >
>> > You received this message because you are subscribed to the Google
>> > Groups "aroma.affymetrix" group with website http://www.aroma-project.org/.
>> > To post to this group, send email to [hidden email]
>> > To unsubscribe and other options, go to
>> > http://www.aroma-project.org/forum/
>> >
>>
>> --
>> When reporting problems on aroma.affymetrix, make sure 1) to run the
>> latest version of the package, 2) to report the output of sessionInfo() and
>> traceback(), and 3) to post a complete code example.
>>
>>
>> You received this message because you are subscribed to the Google Groups
>> "aroma.affymetrix" group with website http://www.aroma-project.org/.
>> To post to this group, send email to [hidden email]
>> To unsubscribe and other options, go to
>> http://www.aroma-project.org/forum/
>
> --
> When reporting problems on aroma.affymetrix, make sure 1) to run the latest
> version of the package, 2) to report the output of sessionInfo() and
> traceback(), and 3) to post a complete code example.
>
>
> You received this message because you are subscribed to the Google Groups
> "aroma.affymetrix" group with website http://www.aroma-project.org/.
> To post to this group, send email to [hidden email]
> To unsubscribe and other options, go to http://www.aroma-project.org/forum/
>

--
When reporting problems on aroma.affymetrix, make sure 1) to run the latest version of the package, 2) to report the output of sessionInfo() and traceback(), and 3) to post a complete code example.


You received this message because you are subscribed to the Google Groups "aroma.affymetrix" group with website http://www.aroma-project.org/.
To post to this group, send email to [hidden email]
To unsubscribe and other options, go to http://www.aroma-project.org/forum/
Reply | Threaded
Open this post in threaded view
|

Re: [aroma.affymetrix] unusual copy number analysis result - split copy numbers

Henrik Bengtsson
On Tue, Sep 28, 2010 at 12:03 PM, hb <[hidden email]> wrote:

> Hi.
>
> On Tue, Sep 28, 2010 at 11:54 AM, Patrick Danaher <[hidden email]> wrote:
>> Hi Henrik,
>> Thanks for your response.  The thread you suggested ( http://goo.gl/FGVe )
>> describes my problem well - I'm getting a very similar intensity profile for
>> some chromosomes in some samples.  The attached png shows the problem (red
>> dots are intensities; the black dots are from a copy number calling problem
>> and can be ignored).  The second figure plots the called intensities against
>> normal reference intensities for the same loci.
>> As for your specific questions: I've never used CRMAv2 before, my dataset
>> isn't public, and it's an affy 6.0 chip.
>
> What's your chip type? The other thread reported problems on Mapping250K_Sty though labelled as "Sty 2" and I don't know what the "2" means there.

Woops, I read it as it was *not* a GenomeWideSNP_6 chip in your
"...and it's an affy 6.0 chip" note.  So, it is GenomeWideSNP_6.

>
>> Do you think annotation issues would cause this problem only in a small
>> subset of my samples?
>
> When I say annotation issues, I really mean that if the CDF for the chip type is not the correct one, you might pick up the wrong probe signals, especially for SNPs, e.g. PM_A may get the value of a total CN probe once in a while, say. It could be a software/annotation bug in the Affymetrix DAT to CEL file conversion and so on. That's why it is crucial to know more about the chip used.
>
> I also recommend that you try dChip and/or Affymetrix GTC, if possible.

Since it is GenomeWideSNP_6, you should be able to try it on Affymetrix GTC.

/Henrik

>
> /Henrik
>
>> Thanks,
>> Patrick
>> On Sun, Sep 26, 2010 at 1:36 PM, Henrik Bengtsson <[hidden email]>
>> wrote:
>>>
>>> Hi.
>>>
>>> On Mon, Sep 13, 2010 at 4:19 PM, Patrick <[hidden email]>
>>> wrote:
>>> > Hi everyone,
>>> > I'm using AROMA's implementation of the CRMA v2 method to get copy
>>> > number estimates for cancer samples, and I'm getting a very unusual
>>> > result.  Many of the samples have a chromosome where AROMA has called
>>> > primarily copy number gains or losses, and the losses are mixed in
>>> > with each other.  That is, if you plot the probes' intensities by
>>> > their positions on the chromosome, you see large stretches (~10,000
>>> > probes) where there are no intensities in the normal range
>>> > (corresponding to no gain or loss), and there are intensities both
>>> > above and below the normal range, mixed in with each other along the
>>> > chromosome.  It is as if the plots for a chromosome with a long
>>> > deletion and a chromosome with a long addition were laid atop each
>>> > other.
>>>
>>> It is not 100% clear from your description what you are observing.
>>> Note that it is possible to attach PNGs to messages sent to this
>>> mailing list as long as you send it as an email (not via the web
>>> interface).  What chip type are you working on and do you look at a
>>> public data set?  Have you used CRMAv2 on other data sets without a
>>> problem?
>>>
>>> FYI, Johan Staaf reported odd looking copy number results that are
>>> reproducible and very odd. See thread 'Problems with Affymetrix 250K
>>> Sty2 arrays after CRMAv2 analysis' on June 23-August 5, 2010, cf.
>>> http://goo.gl/FGVe.  From the discussion in that thread, it seems to
>>> have something to do with annotation issues, but it is still to be
>>> solved.  Is that what you are experiencing?
>>>
>>> > It seems implausible that a cancer sample would have copy number gains
>>> > and losses mixed in with each other in such small intervals, over such
>>> > large stretches of chromosome, without any loci having the usual 2
>>> > copies, so I suspect the normalization or the affy array is the source
>>> > of this phenomenon.  I looked at the data without using AROMA, and the
>>> > phenomenon was not evident.  I re-normalized the data 3 times, each
>>> > time using only one step of the AROMA normalization in isolation.  The
>>> > base position normalization step produced the phenomenon, and the
>>> > allele crosstalk calibration and the fragment length normalization
>>> > steps did not.
>>>
>>> What would help troubleshooting is if you could see other software
>>> such as dChip of Affymetrix GTC produces the same oddities.  If they
>>> do, we know for sure it's something odd with the annotation.
>>>
>>> /Henrik
>>>
>>> > Any thoughts on what I'm seeing and on how the base pair normalization
>>> > could cause it would be very appreciated.
>>> > Thanks,
>>> > Patrick
>>> >
>>> > --
>>> > When reporting problems on aroma.affymetrix, make sure 1) to run the
>>> > latest version of the package, 2) to report the output of sessionInfo() and
>>> > traceback(), and 3) to post a complete code example.
>>> >
>>> >
>>> > You received this message because you are subscribed to the Google
>>> > Groups "aroma.affymetrix" group with website http://www.aroma-project.org/.
>>> > To post to this group, send email to [hidden email]
>>> > To unsubscribe and other options, go to
>>> > http://www.aroma-project.org/forum/
>>> >
>>>
>>> --
>>> When reporting problems on aroma.affymetrix, make sure 1) to run the
>>> latest version of the package, 2) to report the output of sessionInfo() and
>>> traceback(), and 3) to post a complete code example.
>>>
>>>
>>> You received this message because you are subscribed to the Google Groups
>>> "aroma.affymetrix" group with website http://www.aroma-project.org/.
>>> To post to this group, send email to [hidden email]
>>> To unsubscribe and other options, go to
>>> http://www.aroma-project.org/forum/
>>
>> --
>> When reporting problems on aroma.affymetrix, make sure 1) to run the latest
>> version of the package, 2) to report the output of sessionInfo() and
>> traceback(), and 3) to post a complete code example.
>>
>>
>> You received this message because you are subscribed to the Google Groups
>> "aroma.affymetrix" group with website http://www.aroma-project.org/.
>> To post to this group, send email to [hidden email]
>> To unsubscribe and other options, go to http://www.aroma-project.org/forum/
>>
>
>

--
When reporting problems on aroma.affymetrix, make sure 1) to run the latest version of the package, 2) to report the output of sessionInfo() and traceback(), and 3) to post a complete code example.


You received this message because you are subscribed to the Google Groups "aroma.affymetrix" group with website http://www.aroma-project.org/.
To post to this group, send email to [hidden email]
To unsubscribe and other options, go to http://www.aroma-project.org/forum/