Hi Gordon
I've received and error message with limma that I haven't come across
before
but I only receive it with data that has been normalised between
arrays:
cor <- dupcor.series(NWAresults$M, design, ndups=2, spacing=840)
works fine but if I try:
nbaResults <- normalizeBetweenArrays(NWAresults)
then:
cor <- dupcor.series(nbaResults$M, design, ndups=2, spacing=840)
It returns:
*Error in chol(dinfo) : the leading minor of order 2 is not positive
definite.*
there are only two slides in this particular experiment could this be
part of the problem ?
Thanks
Jason
--
--------------------------------
Jason Skelton
Pathogen Microarrays
Wellcome Trust Sanger Institute
Hinxton
Cambridge
CB10 1SA
Tel +44(0)1223 834244 Ext 7123
Fax +44(0)1223 494919
--------------------------------
[[alternative HTML version deleted]]
At 12:44 AM 16/03/2004, Jason Skelton wrote:
>Hi Gordon
>
>I've received and error message with limma that I haven't come across
before
>but I only receive it with data that has been normalised between
arrays:
>
>cor <- dupcor.series(NWAresults$M, design, ndups=2, spacing=840)
>works fine but if I try:
>
>
>nbaResults <- normalizeBetweenArrays(NWAresults)
>then:
>cor <- dupcor.series(nbaResults$M, design, ndups=2, spacing=840)
>
>It returns:
>
>*Error in chol(dinfo) : the leading minor of order 2 is not positive
>definite.*
I've never seen this error either. What version of R/limma are you
using?
What do you see if you type
summary(nbaResults$M)?
>there are only two slides in this particular experiment could this be
>part of the problem ?
It's probably part of the problem, but we often use dupcor.series on
just
two arrays without any problems. Give me more details of your version
etc
and we will follow it up. I am guessing that you have a lot of missing
values in your data.
Gordon
>Thanks
>
>Jason
>
>--
>--------------------------------
>Jason Skelton
>Pathogen Microarrays
>Wellcome Trust Sanger Institute
>
>
>>
>> *Error in chol(dinfo) : the leading minor of order 2 is not
positive
>> definite.*
>
>
> I've never seen this error either. What version of R/limma are you
> using? What do you see if you type
>
> summary(nbaResults$M)?
Hi Gordon
Using limma version 1.5.1
& R version 1.8.0
If I look at the normalized within Arrays data (Which works with
dupcor.series)
> summary(nwam0m2MA$M)
B737.10.03.2004.m0m2.5um.0019m0m2 B737.10.03.2004.m2m0.5um.0046m0m2
Min. :-1.417e+01 Min. :-1.638e+01
1st Qu.:-1.224e+00 1st Qu.:-6.466e-01
Median :-2.273e-03 Median : 8.657e-03
Mean :-2.393e-01 Mean :-1.095e-02
3rd Qu.: 1.105e+00 3rd Qu.: 6.254e-01
Max. : 7.943e+00 Max. : 8.213e+00
NA's : 5.811e+03 NA's : 1.685e+03
The normalised between arrays data looks like this
I can't see that much difference but the quality of the second array
(B737.10.03.2004.m2m0.5um.0046m0m2)
wasn't that great in the first place.
> summary(nbamom2MA$M)
B737.10.03.2004.m0m2.5um.0019m0m2 B737.10.03.2004.m2m0.5um.0046m0m2
Min. :-1.048e+01 Min. : -22.15456
1st Qu.:-9.056e-01 1st Qu.: -0.87431
Median :-1.681e-03 Median : 0.01171
Mean :-1.770e-01 Mean : -0.01481
3rd Qu.: 8.174e-01 3rd Qu.: 0.84567
Max. : 5.874e+00 Max. : 11.10602
NA's : 5.811e+03 NA's :1685.00000
Cheers
Jason
--
--------------------------------
Jason Skelton
Pathogen Microarrays
Wellcome Trust Sanger Institute
Hinxton
Cambridge
CB10 1SA
Tel +44(0)1223 834244 Ext 7123
Fax +44(0)1223 494919
Jason,
I am a bit dubious about the merits of between array normalization
when
there are so many missing values (nearly 6000 on one of the arrays),
but it
shouldn't in principle cause an error to the duplicateCorrelation
function.
So there may be a bug in the function. Would you be prepared to send
me the
data matrix so that I can try it out myself?
Gordon
At 08:38 PM 16/03/2004, Jason Skelton wrote:
>>>*Error in chol(dinfo) : the leading minor of order 2 is not
positive
>>>definite.*
>>
>>I've never seen this error either. What version of R/limma are you
using?
>>What do you see if you type
>>
>>summary(nbaResults$M)?
>
>Hi Gordon
>
>Using limma version 1.5.1
>& R version 1.8.0
>
>If I look at the normalized within Arrays data (Which works with
>dupcor.series)
>
> > summary(nwam0m2MA$M)
>B737.10.03.2004.m0m2.5um.0019m0m2 B737.10.03.2004.m2m0.5um.0046m0m2
>Min. :-1.417e+01 Min. :-1.638e+01
>1st Qu.:-1.224e+00 1st Qu.:-6.466e-01
>Median :-2.273e-03 Median : 8.657e-03
>Mean :-2.393e-01 Mean :-1.095e-02
>3rd Qu.: 1.105e+00 3rd Qu.: 6.254e-01
>Max. : 7.943e+00 Max. : 8.213e+00
>NA's : 5.811e+03 NA's : 1.685e+03
>
>The normalised between arrays data looks like this
>I can't see that much difference but the quality of the second array
>(B737.10.03.2004.m2m0.5um.0046m0m2)
>wasn't that great in the first place.
>
> > summary(nbamom2MA$M)
>B737.10.03.2004.m0m2.5um.0019m0m2 B737.10.03.2004.m2m0.5um.0046m0m2
>Min. :-1.048e+01 Min. : -22.15456
>1st Qu.:-9.056e-01 1st Qu.: -0.87431
>Median :-1.681e-03 Median : 0.01171
>Mean :-1.770e-01 Mean : -0.01481
>3rd Qu.: 8.174e-01 3rd Qu.: 0.84567
>Max. : 5.874e+00 Max. : 11.10602
>NA's : 5.811e+03 NA's :1685.00000
>
>Cheers
>
>Jason
