[vsni.co.uk] Contact us | Site map
Forum | VSN International Forum Index
Author Message
Post new topic

<  ASReml  ~  User Created Inverse Matrix for Genetic Analysis

Michelle
Posted: Thu Aug 06, 2009 4:50 pm Reply with quote
Joined: 06 Aug 2009 Posts: 5
Hi-

I have recently installed ASReml for R in Linux (Suse 11.1) and am attempting to do a genetic analysis (mixed model analysis) of some data and have been following the information provided in Chapter 5 of the ASReml-R instruction manual.

What I would like to do, is incorporate an inverted genomic relationship matrix which I have already created instead of using an inverted matrix generated from the pedigree (created using asreml.Ainverse() as I understand it). From what I can tell, this is possible, but I am not sure if I am doing it correctly. This is what I have created so far that is generating some data:

RMATINV.giv = read.table(file.choose(), header=FALSE) #inverted matrix
testgeno = read.table(file.choose(), header=TRUE) #genotypes with phenotypes file
test.ped = read.table(file.choose(), header=TRUE) #pedigree file

testgeno$Animal<-as.factor(testgeno$Animal)

RMATINV2.giv = as.matrix(RMATINV.giv)
colnames(RMATINV2.giv) = c("9001":"9010")
rownames(RMATINV2.giv) = c("9001":"9010")

test.asr = asreml(Fert ~ ARS.BFGL.BAC.10172, random= ~ giv(Animal,var=T), control = asreml.control(ginverse=list(Animal=RMATINV2.giv)), data=testgeno)

My questions are:

1) the pedigree file does not seem to be incorporatted into the analysis- is this ok and correct?
2) the format of my relationship matrix is a 10 by 10 matrix (not the default format of asreml which likes a row column value format.) It seems that it is ok and possible to leave the matrix in this format as long as this is indicated in control = asreml.control(ginverse=list(Animal=RMATINV2.giv)) as I have done. Is this correct?
3) Also, it is noted in the ASReml-R manual that all forms of an imported inverted matrix need to have a "rowNames" indicator...I'm not sure if I have done this.

Files mentioned in the above code are attached. Thanks to anyone who can provide some information on this subject.

Michelle



RMATInv.txt
 Description:

Download
 Filename:  RMATInv.txt
 Filesize:  1.81 KB
 Downloaded:  1109 Time(s)


test.ped.txt
 Description:

Download
 Filename:  test.ped.txt
 Filesize:  144 Bytes
 Downloaded:  1095 Time(s)


testgeno.txt
 Description:

Download
 Filename:  testgeno.txt
 Filesize:  201 Bytes
 Downloaded:  1023 Time(s)

View user's profile Send private message
brian.cullis
Posted: Fri Aug 07, 2009 12:36 am Reply with quote
Guest
Dear michelle
i passed this onto Dave Butler who is cc'd in the email
his reply was

What you are doing looks ok. Reading a giv matrix in this form is a seldom used option and there may be a bug in the code that reads and converts it to sparse form. I'll check it out.

***ie watch this space



warm regards

Brian Cullis
Research Leader, Biometrics &
Senior Principal Research Scientist
NSW Department of Primary Industries
Wagga Wagga Agricultural Institute

Visiting Professorial Fellow
School of Mathematics and Applied Statistics
Faculty of Informatics
University of Wollongong

Professor,
Faculty of Agriculture, Food & Natural Resources
The University of Sydney

Adjunct Professor
School of Computing and Mathematics
Charles Sturt University


Phone: 61 2 6938 1855
Fax: 61 2 6938 1809
Mobile: 0439 448 591






This message is intended for the addressee named and may contain confidential information. If you are not the intended recipient, please delete it and notify the sender. Views expressed in this message are those of the individual sender, and are not necessarily the views of their organisation.

Post generated using Mail2Forum (http://www.mail2forum.com)
Butler, David
Posted: Fri Aug 07, 2009 2:53 am Reply with quote
Guest
Hi Michelle,

My results from your example:

Quote:
library(asreml, lib.loc="~/MyRlibrary")
Loading required package: lattice
Quote:
RMATINV.giv = read.table("rmatinv_208.txt", header=FALSE) #inverted matrix
testgeno = read.table("testgeno_180.txt", header=TRUE) #genotypes with phenotypes file
test.ped = read.table("testped_874.txt", header=TRUE) #pedigree file
testgeno$Animal<-as.factor(testgeno$Animal)
RMATINV2.giv = as.matrix(RMATINV.giv)
colnames(RMATINV2.giv) = c("9001":"9010")
rownames(RMATINV2.giv) = c("9001":"9010")

Quote:

test.asr = asreml(Fert ~ ARS.BFGL.BAC.10172,
+ random= ~ giv(Animal,var=T),
+ control = asreml.control(ginverse=list(Animal=RMATINV2.giv)), data=testgeno)

asreml(): 2.00 Library: 2.00bi Run: Fri Aug 7 12:05:03 2009
Equations: 12 (2 dense)
Initial update shrinkage factor: 0.316
LogLik S2 DF

11.0858 0.0140 8 12:05:03 (1 component(s) constrained)
11.1206 0.0152 8 12:05:03 (1 component(s) constrained)
11.1231 0.0152 8 12:05:03 (1 component(s) constrained)
11.1233 0.0152 8 12:05:03 (1 component(s) constrained)
11.1233 0.0152 8 12:05:03 (1 component(s) constrained)
11.1233 0.0152 8 12:05:03
11.1233 0.0152 8 12:05:03

Finished on: Fri Aug 7 12:05:03 2009

LogLikelihood Converged

Quote:
summary(test.asr)$varcomp
gamma component std.error z.ratio constraint
giv(Animal, var = T) 1.011929e-07 1.543191e-09 7.715957e-10 2 Boundary
R!variance 1.000000e+00 1.525000e-02 7.624999e-03 2 Positive

In answer to your questions:

1. The ginverse as supplied in matrix form looks to have been used in tha analysis, and the variance component is bounded. I also checked the "harvey" example in the manual using a full A inverse instead of the sparse form and it seemed ok.
2. While it's not necessary to wrap it with "asreml.control()", the inverse matrix in whatever form must be given in the "ginverse" argument as you have done. Asreml determines if it is a matrix or 3 column dataframe and continues accordingly. A matrix with <= 3 columns will confuse it.
3. Animals in the data must appear in the ginverse. The ginv produced by asreml.Ainverse() has an attribute "rowNames" derived from the pedigree and naming the rows of the ginverse matrix. Asreml uses this to firstly check if all names in the data are in the pedigree and secondly "relabel" the factor levels if necessary.

Dave.

From: ASReml users discussion group [mailto:ASREML-L@DPI.NSW.GOV.AU] On Behalf Of Michelle
Sent: Friday, 7 August 2009 2:51 AM
To: ASREML-L@DPI.NSW.GOV.AU
Subject: User Created Inverse Matrix for Genetic Analysis



Hi-

I have recently installed ASReml for R in Linux (Suse 11.1) and am attempting to do a genetic analysis (mixed model analysis) of some data and have been following the information provided in Chapter 5 of the ASReml-R instruction manual.

What I would like to do, is incorporate an inverted genomic relationship matrix which I have already created instead of using an inverted matrix generated from the pedigree (created using asreml.Ainverse() as I understand it). From what I can tell, this is possible, but I am not sure if I am doing it correctly. This is what I have created so far that is generating some data:

RMATINV.giv = read.table(file.choose(), header=FALSE) #inverted matrix
testgeno = read.table(file.choose(), header=TRUE) #genotypes with phenotypes file
test.ped = read.table(file.choose(), header=TRUE) #pedigree file

testgeno$Animal<-as.factor(testgeno$Animal)

RMATINV2.giv = as.matrix(RMATINV.giv)
colnames(RMATINV2.giv) = c("9001":"9010")
rownames(RMATINV2.giv) = c("9001":"9010")

test.asr = asreml(Fert ~ ARS.BFGL.BAC.10172, random= ~ giv(Animal,var=T), control = asreml.control(ginverse=list(Animal=RMATINV2.giv)), data=testgeno)

My questions are:

1) the pedigree file does not seem to be incorporatted into the analysis- is this ok and correct?
2) the format of my relationship matrix is a 10 by 10 matrix (not the default format of asreml which likes a row column value format.) It seems that it is ok and possible to leave the matrix in this format as long as this is indicated in control = asreml.control(ginverse=list(Animal=RMATINV2.giv)) as I have done. Is this correct?
3) Also, it is noted in the ASReml-R manual that all forms of an imported inverted matrix need to have a "rowNames" indicator...I'm not sure if I have done this.

Files mentioned in the above code are attached. Thanks to anyone who can provide some information on this subject.

Michelle





Attachments:
RMATInv.txt
test.ped.txt
testgeno.txt





This message is intended for the addressee named and may contain confidential information. If you are not the intended recipient, please delete it and notify the sender. Views expressed in this message are those of the individual sender, and are not necessarily the views of their organisation.


********************************DISCLAIMER****************************
The information contained in the above e-mail message or messages
(which includes any attachments) is confidential and may be legally
privileged. It is intended only for the use of the person or entity
to which it is addressed. If you are not the addressee any form of
disclosure, copying, modification, distribution or any action taken
or omitted in reliance on the information is unauthorised. Opinions
contained in the message(s) do not necessarily reflect the opinions
of the Queensland Government and its authorities. If you received
this communication in error, please notify the sender immediately
and delete it from your computer system network.



This message is intended for the addressee named and may contain confidential information. If you are not the intended recipient, please delete it and notify the sender. Views expressed in this message are those of the individual sender, and are not necessarily the views of their organisation.

Post generated using Mail2Forum (http://www.mail2forum.com)
Michelle
Posted: Fri Aug 07, 2009 5:53 pm Reply with quote
Joined: 06 Aug 2009 Posts: 5
My thanks to both of you- your answers were very helpful.

Michelle
View user's profile Send private message
Marte Sodeland
Posted: Tue Feb 02, 2010 8:17 am Reply with quote
Joined: 29 Jan 2010 Posts: 4
Hi,
I'm trying to do the same thing as Michelle so I wanted to test if this code would also work for me.
I downloaded her data and used the code provided by David, but for me this didn't work. I get the error message:
"Error in model.frame.default(formula = fert ~ ARS.BFGL.BAC.10172 + Animal + variable lengths differ (found for 'T')".

Does anyone know why this is not working on my machine?
Have a windows R-2.9.0 version.

Marte
View user's profile Send private message
Butler, David
Posted: Tue Feb 02, 2010 9:17 am Reply with quote
Guest
Hi Marte,

Looks to me like you are using an earlier release of V3, and the asreml call has "var=T" in the ped() function. This is no longer required in V3 and in earlier releases it caused a curious looking error when building the model frame.
Dave.


-----Original Message-----
From: ASReml users discussion group on behalf of Marte Sodeland
Sent: Tue 02-Feb-10 6:17 PM
To: ASREML-L@DPI.NSW.GOV.AU
Subject: Re: User Created Inverse Matrix for Genetic Analysis

Hi,
I'm trying to do the same thing as Michelle so I wanted to test if this code would also work for me.
I downloaded her data and used the code provided by David, but for me this didn't work. I get the error message:
"Error in model.frame.default(formula = fert ~ ARS.BFGL.BAC.10172 + Animal + variable lengths differ (found for 'T')".

Does anyone know why this is not working on my machine?
Have a windows R-2.9.0 version.

Marte







This message is intended for the addressee named and may contain confidential information. If you are not the intended recipient, please delete it and notify the sender. Views expressed in this message are those of the individual sender, and are not necessarily the views of their organisation.

********************************DISCLAIMER****************************
The information contained in the above e-mail message or messages
(which includes any attachments) is confidential and may be legally
privileged. It is intended only for the use of the person or entity
to which it is addressed. If you are not the addressee any form of
disclosure, copying, modification, distribution or any action taken
or omitted in reliance on the information is unauthorised. Opinions
contained in the message(s) do not necessarily reflect the opinions
of the Queensland Government and its authorities. If you received
this communication in error, please notify the sender immediately
and delete it from your computer system network.


This message is intended for the addressee named and may contain confidential information. If you are not the intended recipient, please delete it and notify the sender. Views expressed in this message are those of the individual sender, and are not necessarily the views of their organisation.

Post generated using Mail2Forum (http://www.mail2forum.com)
Marte Sodeland
Posted: Wed Feb 03, 2010 6:04 pm Reply with quote
Joined: 29 Jan 2010 Posts: 4
I'm using the latest ASReml-R version available to me (from 4th nov 2009). When I remove 'var=T' I get the error message:
"Error in buffer[3] <- -Ainv$locgiv[match(names(gcov.mdl[[Gn]][[i]]$struc)[3], :
replacement has length zero".

Marte
View user's profile Send private message
Butler, David
Posted: Thu Feb 04, 2010 2:14 am Reply with quote
Guest
Marte,

Sorry, I changed the way pedigrees are stored internally and there seems to be a lingering bug with giv(). I'll attend to it. Try ped().

Dave.


From: ASReml users discussion group [mailto:ASREML-L@DPI.NSW.GOV.AU] On Behalf Of Marte Sodeland
Sent: Thursday, 4 February 2010 4:05 AM
To: ASREML-L@DPI.NSW.GOV.AU
Subject: Re: User Created Inverse Matrix for Genetic Analysis



I'm using the latest ASReml-R version available to me (from 4th nov 2009). When I remove 'var=T' I get the error message:
"Error in buffer[3] <- -Ainv$locgiv[match(names(gcov.mdl[[Gn]][[i]]$struc)[3], :
replacement has length zero".

Marte







This message is intended for the addressee named and may contain confidential information. If you are not the intended recipient, please delete it and notify the sender. Views expressed in this message are those of the individual sender, and are not necessarily the views of their organisation.


********************************DISCLAIMER****************************
The information contained in the above e-mail message or messages
(which includes any attachments) is confidential and may be legally
privileged. It is intended only for the use of the person or entity
to which it is addressed. If you are not the addressee any form of
disclosure, copying, modification, distribution or any action taken
or omitted in reliance on the information is unauthorised. Opinions
contained in the message(s) do not necessarily reflect the opinions
of the Queensland Government and its authorities. If you received
this communication in error, please notify the sender immediately
and delete it from your computer system network.

Post generated using Mail2Forum (http://www.mail2forum.com)
Craig Hardner
Posted: Thu Feb 04, 2010 2:30 am Reply with quote
Guest
Sorry Dave just to hassle you again, any progress with the Legendre multipliers?

Thanks Craig

Dr Craig Hardner
Research Fellow
School of Land, Crop and Food Sciences
University of Queensland
St Lucia 4072 Queensland, Australia
ph: +61 7 3346 9465
email: craig.hardner@uq.edu.au (craig.hardner@uq.edu.au)


From: ASReml users discussion group [mailto:ASREML-L@DPI.NSW.GOV.AU] On Behalf Of Butler, David
Sent: Thursday, 4 February 2010 12:12 PM
To: ASREML-L@DPI.NSW.GOV.AU
Subject: Re: User Created Inverse Matrix for Genetic Analysis



Marte,


Sorry, I changed the way pedigrees are stored internally and there seems to be a lingering bug with giv(). I'll attend to it. Try ped().



Dave.



From: ASReml users discussion group [mailto:ASREML-L@DPI.NSW.GOV.AU] On Behalf Of Marte Sodeland
Sent: Thursday, 4 February 2010 4:05 AM
To: ASREML-L@DPI.NSW.GOV.AU
Subject: Re: User Created Inverse Matrix for Genetic Analysis
I'm using the latest ASReml-R version available to me (from 4th nov 2009). When I remove 'var=T' I get the error message:
"Error in buffer[3] <- -Ainv$locgiv[match(names(gcov.mdl[[Gn]][[i]]$struc)[3], :
replacement has length zero".

Marte







This message is intended for the addressee named and may contain confidential information. If you are not the intended recipient, please delete it and notify the sender. Views expressed in this message are those of the individual sender, and are not necessarily the views of their organisation.


********************************DISCLAIMER****************************
The information contained in the above e-mail message or messages
(which includes any attachments) is confidential and may be legally
privileged. It is intended only for the use of the person or entity
to which it is addressed. If you are not the addressee any form of
disclosure, copying, modification, distribution or any action taken
or omitted in reliance on the information is unauthorised. Opinions
contained in the message(s) do not necessarily reflect the opinions
of the Queensland Government and its authorities. If you received
this communication in error, please notify the sender immediately
and delete it from your computer system network.

Post generated using Mail2Forum (http://www.mail2forum.com)
Butler, David
Posted: Thu Feb 04, 2010 4:00 am Reply with quote
Guest
sorry Craig, got distracted. I'll try very hard to do it next week.

dave.

From: ASReml users discussion group [mailto:ASREML-L@DPI.NSW.GOV.AU] On Behalf Of Craig Hardner
Sent: Thursday, 4 February 2010 12:28 PM
To: ASREML-L@DPI.NSW.GOV.AU
Subject: Re: User Created Inverse Matrix for Genetic Analysis




Sorry Dave just to hassle you again, any progress with the Legendre multipliers?

Thanks Craig

Dr Craig Hardner
Research Fellow
School of Land, Crop and Food Sciences
University of Queensland
St Lucia 4072 Queensland, Australia
ph: +61 7 3346 9465
email: craig.hardner@uq.edu.au (craig.hardner@uq.edu.au)


From: ASReml users discussion group [mailto:ASREML-L@DPI.NSW.GOV.AU] On Behalf Of Butler, David
Sent: Thursday, 4 February 2010 12:12 PM
To: ASREML-L@DPI.NSW.GOV.AU
Subject: Re: User Created Inverse Matrix for Genetic Analysis



Marte,


Sorry, I changed the way pedigrees are stored internally and there seems to be a lingering bug with giv(). I'll attend to it. Try ped().



Dave.



From: ASReml users discussion group [mailto:ASREML-L@DPI.NSW.GOV.AU] On Behalf Of Marte Sodeland
Sent: Thursday, 4 February 2010 4:05 AM
To: ASREML-L@DPI.NSW.GOV.AU
Subject: Re: User Created Inverse Matrix for Genetic Analysis
I'm using the latest ASReml-R version available to me (from 4th nov 2009). When I remove 'var=T' I get the error message:
"Error in buffer[3] <- -Ainv$locgiv[match(names(gcov.mdl[[Gn]][[i]]$struc)[3], :
replacement has length zero".

Marte







This message is intended for the addressee named and may contain confidential information. If you are not the intended recipient, please delete it and notify the sender. Views expressed in this message are those of the individual sender, and are not necessarily the views of their organisation.


********************************DISCLAIMER****************************
The information contained in the above e-mail message or messages
(which includes any attachments) is confidential and may be legally
privileged. It is intended only for the use of the person or entity
to which it is addressed. If you are not the addressee any form of
disclosure, copying, modification, distribution or any action taken
or omitted in reliance on the information is unauthorised. Opinions
contained in the message(s) do not necessarily reflect the opinions
of the Queensland Government and its authorities. If you received
this communication in error, please notify the sender immediately
and delete it from your computer system network.




********************************DISCLAIMER****************************
The information contained in the above e-mail message or messages
(which includes any attachments) is confidential and may be legally
privileged. It is intended only for the use of the person or entity
to which it is addressed. If you are not the addressee any form of
disclosure, copying, modification, distribution or any action taken
or omitted in reliance on the information is unauthorised. Opinions
contained in the message(s) do not necessarily reflect the opinions
of the Queensland Government and its authorities. If you received
this communication in error, please notify the sender immediately
and delete it from your computer system network.

Post generated using Mail2Forum (http://www.mail2forum.com)
Marte Sodeland
Posted: Tue Feb 09, 2010 8:30 am Reply with quote
Joined: 29 Jan 2010 Posts: 4
Thanks! :)
I would be very greatful if you could show me how a "command file" doing the same thing should look like.

Marte
View user's profile Send private message
sergio.roman
Posted: Thu Jun 02, 2011 12:52 pm Reply with quote
Joined: 02 Jun 2011 Posts: 3
Hello,

I am using ASREML on Win 64 bits and ASREML-R.

Actually I am working on ASREML-R following this example
RMATINV.giv = read.table("rmatinv_208.txt", header=FALSE) #inverted matrix
testgeno = read.table("testgeno_180.txt", header=TRUE) #genotypes with phenotypes file
test.ped = read.table("testped_874.txt", header=TRUE) #pedigree file
testgeno$Animal<-as.factor(testgeno$Animal)
RMATINV2.giv = as.matrix(RMATINV.giv)
colnames(RMATINV2.giv) = c("9001":"9010")
rownames(RMATINV2.giv) = c("9001":"9010")
test.asr = asreml(Fert ~ ARS.BFGL.BAC.10172,
+ random= ~ giv(Animal,var=T),
+ control = asreml.control(ginverse=list(Animal=RMATINV2.giv)), data=testgeno)

In my case I have two relatiomship matrices. The first is the Ainverse matrix derived from pedigree, and the second is derived from genomic relationship matrix.

I already produce the follow sintaxis. any suggestions ?

model3<-asreml(fixed=milk ~ 1
+ random= ~ ped(anim,var=T,init=1) #Ainverse matrix derived from pedigree
+ random= ~ giv(anim,var=T) #Ginverse matrix derived from markers
+ control = asreml.control(ginverse=list(Animal=IGG)), data=data)

Could you provide a single example how to do it ? Becuae in the page 53 of the manual is just mentioned, but any example is provoded !!

regards,
View user's profile Send private message Send e-mail
sergio.roman
Posted: Thu Jun 02, 2011 5:04 pm Reply with quote
Joined: 02 Jun 2011 Posts: 3
Hello,

Today I am editing my post becuase i already solve my problem.

test.asr<-asreml(milk ~1, random= ~ped(anim,var=T) + giv(id,var=T),
data=data,
control=asreml.control(ginverse=list(anim=ainv, id=gmatrix.giv )))

I am running this code on several traits.

One asr output is:

asreml(): 3.0.1 Library: 3.01gl X86_64 Run: Fri Jun 03 13:27:42 2011
LogLik S2 DF
1472.6940 0.0206 1088 13:27:59
1505.2097 0.0167 1088 13:28:12
1544.7458 0.0111 1088 13:28:26
1575.3727 0.0051 1088 13:28:39
1584.9584 0.0022 1088 13:28:53
1587.0849 0.0010 1088 13:29:06
1587.4634 0.0006 1088 13:29:20
1587.5037 0.0005 1088 13:29:33
1587.5045 0.0005 1088 13:29:47
1587.5046 0.0005 1088 13:30:00
Finished on: Fri Jun 03 13:30:00 2011
LogLikelihood Converged
> summary(fatp.asr)$varcomp
gamma component std.error z.ratio constraint
ped(anim, var = T)!ped 22.85673 0.0109424487 0.0021800603 5.019333 Positive
giv(id, var = T).giv 33.79923 0.0161810660 0.0020622599 7.846279 Positive
R!variance 1.00000 0.0004787407 0.0007904041 0.605691 Positive

In ONE of SIX i obtain this message: I already try with maxiter=100 in the as.control(), but the message remains.

In asreml.call(data, asr.inter, asr.struc, asr.glm, asr.predict$predict, :
At least one parameter changed by more than 1% on the last iteration[/b][/b][/color]

The completeasr output is:

asreml(): 3.0.1 Library: 3.01gl X86_64 Run: Fri Jun 03 13:25:21 2011

LogLik S2 DF
-3780.5590 322.0731 1088 13:25:38
-3670.2805 223.5859 1088 13:25:51
-3556.2858 121.2947 1088 13:26:06
-3485.0495 41.8471 1088 13:26:19
-3469.6693 13.6372 1088 13:26:33
-3467.3067 6.2171 1088 13:26:46
-3466.8833 4.4978 1088 13:27:00
-3466.8661 4.7957 1088 13:27:14
-3466.8653 4.6248 1088 13:27:27
-3466.8653 4.7088 1088 13:27:40

Finished on: Fri Jun 03 13:27:40 2011

LogLikelihood Converged
Change(%)
ped(anim, var = T)!ped 1.01
Mensajes de aviso perdidos
In asreml.call(data, asr.inter, asr.struc, asr.glm, asr.predict$predict, :
At least one parameter changed by more than 1% on the last iteration

> summary(fatk.asr)$varcomp
gamma component std.error z.ratio constraint
ped(anim, var = T)!ped 25.13203 118.34198 20.549475 5.7588808 Positive
giv(id, var = T).giv 38.12391 179.51826 21.698305 8.2733772 Positive
R!variance 1.00000 4.70881 6.291613 0.7484265 Positive

Kindest regards,

Sergio
View user's profile Send private message Send e-mail
Yves Rousselle
Posted: Fri Mar 23, 2012 11:59 am Reply with quote
Joined: 05 Mar 2012 Posts: 9
Dear all,

I also try to apply this method to my data that are basically composed by a genotypes file (containing the marq variable), a phenotypes file (containing the car variable) and a inverted kinship matrix (mat_inv) obtained by applying solve() on a molecular kinship matrix.

When I try to apply in the following model :
Code:

asreml(fixed = car~marq,random = ~ giv(code),ginverse=list(code=mat_inv),data=test,na.method.X="omit")

it produces the following messages :
Code:

asreml(): 3.0.1 Library: 3.01gl X86_64  Run: Fri Mar 23 15:45:12 2012

     LogLik         S2      DF
   -138.0904      1.6763   167  15:45:12
   -132.0462      1.3163   167  15:45:12
   -125.7831      0.8527   167  15:45:12
   -121.7657      0.4017   167  15:45:12
   -120.5736      0.1672   167  15:45:12
   -120.2326      0.0569   167  15:45:12
   -120.1352      0.0122   167  15:45:12
   -120.1143      0.0009   167  15:45:12  (1 component(s) restrained)
   -120.1129      0.0001   167  15:45:12  (1 component(s) restrained)
   -120.1128      0.0000   167  15:45:12
1 singularities inAverage Information Matrix
Exit status: -1 - Singularity in Average Information Matrix                                       

Finished on: Fri Mar 23 15:45:12 2012
 
Messages d'avis :
1: In asreml.control(...) :
  Missing row component of dimnames for ginverse matrix

2: In asreml.control(ginverse = list(code = 2 * mat_inv)) :
  Missing row component of dimnames for ginverse matrix

3: Abnormal termination
Singularity in Average Information Matrix
Results may be erroneous


Do you know what does it mean ?

Reading the other messages of this discussion, I've seen this one from David Butler on February 2010:
Quote:

Sorry, I changed the way pedigrees are stored internally and there seems to be a lingering bug with giv(). I'll attend to it. Try ped().

Can we use giv() for now ? How I use ped() with a molecular kinship matrix ?

Yves
View user's profile Send private message
Yves Rousselle
Posted: Mon Mar 26, 2012 11:38 am Reply with quote
Joined: 05 Mar 2012 Posts: 9
Dear all,

As this problem of singularity is widely discussed on this forum, I had posted a link to this discussion in an other one and Arthur Gilmour answered me. So if you want to see what he said you have to go there :
http://www.vsni.co.uk/forum/viewtopic.php?p=2832#2832

Yves
View user's profile Send private message

Display posts from previous:  

All times are GMT
Page 1 of 1
Post new topic

Jump to:  

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You can attach files in this forum
You can download files in this forum