Forum | VSN International
http://forum.vsni.co.uk//
The forum for discussion of VSN International software and general statistical issuesenCopyright 2000-2010 VSN International Ltd.support@vsni.co.uksupport@vsni.co.ukhttp://blogs.law.harvard.edu/tech/rss60Thu, 13 Dec 2018 21:22:21 GMTThu, 13 Dec 2018 21:22:21 GMThttp://www.vsni.co.uk/common/images/asreml145.gifForum | VSN International
http://forum.vsni.co.uk//
ASReml - RE: how to get genetic correlation in multivariate GBLUP model
http://forum.vsni.co.uk/viewtopic.php?p=5454#5454
Don't use predict to get breeding values. The breeding values are found in <div class="code">Code:<div class="inside_code">summary(modelGBLUP, all = T)$coef.random</div></div>, along with the standard errors and z-values. You will be looking for the coeffcients that begin with giv(ID).ASRemlhttp://forum.vsni.co.uk//posting.php?mode=reply&t=1887tshalevThu, 13 Dec 2018 19:29:01 GMThttp://forum.vsni.co.uk/viewtopic.php?p=5454#5454ASReml - Order of residuals in ante dependence model across time
http://forum.vsni.co.uk/viewtopic.php?p=5453#5453
I am fitting an antedependence model across harvest times (below) and I would like to do some residual plots.
<br />
There is however no indication in the manual as to what order are the model residuals are in .
<br />
Does anyone know?
<br />
<br />
m2.ante<-asreml(cbind(grain.weight.100.seeds.1,grain.weight.100.seeds.2,grain.weight.100.seeds.3)~trait*variety*environment,
<br />
random=~ blockf+at(environment,c(2,4)):bayf,
<br />
rcov=~at(environment):units:ante(trait),
<br />
data=harvest_time,na.method.X='include')
<br />
<br />
Thanks
<br />
HelenaASRemlhttp://forum.vsni.co.uk//posting.php?mode=reply&t=1890a1002973Thu, 13 Dec 2018 04:22:11 GMThttp://forum.vsni.co.uk/viewtopic.php?p=5453#5453ASReml - RE: how to get genetic correlation in multivariate GBLUP model
http://forum.vsni.co.uk/viewtopic.php?p=5452#5452
Thanks <span style="font-style: italic">tshalev, </span>
<br />
I did get some output using this corgh() in the model that produced warnings
<br />
summary(modelGBLUP)$varcomp
<br />
<br />
gamma component
<br />
giv(Genot):trait!trait.trt2:!trait.trt1.cor -0.01321706 -0.01321706
<br />
giv(Genot):trait!trait.trt3:!trait.trt1.cor 0.63034235 0.63034235
<br />
<br />
1) So these cor values then should be the genetic correlation among traits I'm looking at?
<br />
<br />
2) When I further tried obtaining the estimated BV for individual lines with:
<br />
<span style="font-weight: bold">predGBLUP<-predict(modelGBLUP,classify="Genot")$predictions$pvals</span>
<br />
<br />
Same Warning message showed up:
<br />
Abnormal termination
<br />
Nonpositive definite G matrix: 0 singularities, 1 negative pivots; order 6
<br />
Results may be erroneous
<br />
<br />
Then, the predicted.values were all 0s. So somehow why it gives the varcomp but not the EBVs?
<br />
<br />
Thanks again,
<br />
MaoASRemlhttp://forum.vsni.co.uk//posting.php?mode=reply&t=1887MaoHuangTue, 11 Dec 2018 17:36:25 GMThttp://forum.vsni.co.uk/viewtopic.php?p=5452#5452ASReml - RE: how to get genetic correlation in multivariate GBLUP model
http://forum.vsni.co.uk/viewtopic.php?p=5451#5451
In this case, it would seem that the correlation matrix is non-positive definite. Since your model is already quite simple, you could run a univariate model for each trait and then try to use the genetic variance component as your initial starting values in the multivariate model. Note that the message you get is a warning and not an error, so the model is running. You should check whether it converged; if it did, you can still use the output.ASRemlhttp://forum.vsni.co.uk//posting.php?mode=reply&t=1887tshalevTue, 11 Dec 2018 17:25:37 GMThttp://forum.vsni.co.uk/viewtopic.php?p=5451#5451ASReml - RE: how to get genetic correlation in multivariate GBLUP model
http://forum.vsni.co.uk/viewtopic.php?p=5450#5450
Hi there <span style="font-style: italic">tshalev,</span> Thank you for your reply.
<br />
Before fitting this simpler version of model with corh(),
<br />
I did try using the cough()
<br />
But it gave me the below error message:
<br />
<br />
<span style="font-weight: bold">Nonpositive definite G matrix: 0 singularities, 1 negative pivots; order 6
<br />
</span>
<br />
<br />
<br />
Do you know what may cause this error?
<br />
Note: Our G matrix was already positive definite, so not sure why it gives this error.
<br />
something else may be wrong?
<br />
<br />
Thanks again.ASRemlhttp://forum.vsni.co.uk//posting.php?mode=reply&t=1887MaoHuangMon, 10 Dec 2018 16:06:32 GMThttp://forum.vsni.co.uk/viewtopic.php?p=5450#5450ASReml - ASReml-R 3 vs ASReml-R 4 question
http://forum.vsni.co.uk/viewtopic.php?p=5449#5449
Looking for ASReml-R-V4 gurus out there. In V3 I had been assigning my trait name to an object then using "get(trait)" in the ASReml call, rather than changing the trait name in each of my model statements. This does not work now in V4. Any ideas on this?ASRemlhttp://forum.vsni.co.uk//posting.php?mode=reply&t=1889scottdhaleySat, 08 Dec 2018 23:41:07 GMThttp://forum.vsni.co.uk/viewtopic.php?p=5449#5449ASReml - Different variance components, BVs between versions 3 and 4
http://forum.vsni.co.uk/viewtopic.php?p=5448#5448
Hi,
<br />
<br />
I am trying out ASREML version 4, and I have noticed that I am getting different outcomes when compared to version 3. For example, I run this model in 3:
<br />
<br />
<div class="code">Code:<div class="inside_code">asreml(data = wrc.pheno.dat,
<br />
fixed = fath ~ site,
<br />
random = ~ site:REP + idh(site):REP:SET + ped(tree),
<br />
rcov = ~ at(site):id(units),
<br />
ginverse = list(tree = wrc.ainv),
<br />
na.method.X = "include", na.method.Y = "include",
<br />
maxit = 100)</div></div>
<br />
<br />
And this one in 4:
<br />
<div class="code">Code:<div class="inside_code">asreml(data = wrc.pheno.dat,
<br />
fixed = fath ~ site,
<br />
random = ~ site:REP + idh(site):REP:SET + vm(tree, wrc.ainv),
<br />
residual = ~ dsum(~ id(units) | site),
<br />
na.action = na.method("include", "include"),
<br />
maxit = 100)</div></div>
<br />
<br />
And I am assuming that these are identical models, although I am not 100% sure. The variance components for the first model (version 3):
<br />
<div class="code">Code:<div class="inside_code"> gamma component std.error z.ratio constraint
<br />
site:REP!site.var 3515034.3 3515034.3 1116827 3.147340 Positive
<br />
site:REP:SET!site.1 7443651.3 7443651.3 2504858 2.971686 Positive
<br />
site:REP:SET!site.2 512916.7 512916.7 2367086 0.216687 Positive
<br />
site:REP:SET!site.3 10556786.5 10556786.5 2914884 3.621683 Positive
<br />
ped(tree)!ped 13674032.7 13674032.7 3596819 3.801702 Positive
<br />
site_1!variance 28316151.5 28316151.5 3338351 8.482077 Positive
<br />
site_2!variance 43404937.7 43404937.7 4282322 10.135842 Positive
<br />
site_3!variance 26182652.5 26182652.5 3353438 7.807703 Positive</div></div>
<br />
<br />
And for the second model (version 4):
<br />
<div class="code">Code:<div class="inside_code"> effect component std.error z.ratio bound X.ch vc.percent
<br />
1 site:REP 3515194.1 1117124 3.146646 P 0.0 2.6310123
<br />
2 site:REP:SET!site_1 7443850.5 2505343 2.971190 P 0.0 5.5714881
<br />
3 site:REP:SET!site_2 513183.2 2367278 0.216782 P 0.4 0.3841015
<br />
4 site:REP:SET!site_3 10556567.6 2914490 3.622098 P 0.0 7.9012591
<br />
5 vm(tree, wrc.ainv) 13673759.4 3596494 3.801970 P 0.0 10.2343792
<br />
6 site_1!R 28316194.0 3338246 8.482357 P 0.0 21.1937813
<br />
7 site_2!R 43404687.9 4281854 10.136892 P 0.0 32.4870447
<br />
8 site_3!R 26182707.7 3353337 7.807955 P 0.0 19.5969338</div></div>
<br />
Not sure why these are different? The breeding values are also different, although they are very strongly correlated (99%).ASRemlhttp://forum.vsni.co.uk//posting.php?mode=reply&t=1888tshalevWed, 05 Dec 2018 22:46:44 GMThttp://forum.vsni.co.uk/viewtopic.php?p=5448#5448ASReml - RE: how to get genetic correlation in multivariate GBLUP model
http://forum.vsni.co.uk/viewtopic.php?p=5447#5447
If you use corh() you will only get one correlation, as you assume that all traits are equally correlated. If you want a genetic correlation between each pair of traits, you need to use the corgh() structure.ASRemlhttp://forum.vsni.co.uk//posting.php?mode=reply&t=1887tshalevWed, 05 Dec 2018 22:20:20 GMThttp://forum.vsni.co.uk/viewtopic.php?p=5447#5447ASReml - how to get genetic correlation in multivariate GBLUP model
http://forum.vsni.co.uk/viewtopic.php?p=5446#5446
Hi all,
<br />
I am trying to fit multi-trait GBLUP model with the known the ginverse matrix termed as <span style="font-style: italic"><span style="font-weight: bold">gmatrix</span></span> (the inverse of additive relationship matrix calculated from marker data).
<br />
I used the below model:
<br />
<span style="font-weight: bold">modelGBLUP<-asreml(fixed=cbind(tr1,tr2,tr3)~trait,rcov=~units:us(trait),random=~corh(trait):giv(ID),ginverse=list(ID=<span style="font-style: italic">gmatrix</span>),na.method.Y="include",data=Y)</span>
<br />
<br />
I then obtained:
<br />
<span style="font-weight: bold">summary(modelGBLUP)$varcomp</span>
<br />
<br />
I got output with
<br />
gamma component std_error Z.ratio constraint
<br />
tr1 tr1 1.672322483. 1.672322483
<br />
tr2 tr1. -1.15232172. -1.15232172
<br />
tr2 tr2. 4.952745793. 4.952745793
<br />
tr3 tr1. -0.280542184. -0.280542184
<br />
tr3 tr2
<br />
tr3 tr3
<br />
<br />
is this variance values referring to the genetic correlation (variance due to additive genetic effects) among traits?
<br />
How come there's negative values and values >1
<br />
<br />
If this is not the genetic correlation. Then how should I proceed to get the genetic correlation for these traits?
<br />
<br />
ThanksASRemlhttp://forum.vsni.co.uk//posting.php?mode=reply&t=1887MaoHuangWed, 05 Dec 2018 17:09:26 GMThttp://forum.vsni.co.uk/viewtopic.php?p=5446#5446ASReml - Modifying the residual in ASREML R-4 for multivariate models
http://forum.vsni.co.uk/viewtopic.php?p=5445#5445
Hello,
<br />
<br />
I am trying to modify the residual in ASREML R version 4 when using a multivariate model, with little success. There arenâ€™t very many examples for multivariate models in the manual, and as far as I can tell, only the use of residual = ~id(units):us(trait) works. I am trying to include a site effect in the residual to partition the residual by site and trait, something which seemed to work in version 3. Site is encoded as a factor and I have tried either residual = ~site:id(units):us(trait), residual = ~diag(site):id(units):us(trait) and residual = ~idh(site):id(units):us(trait), all of which cause the R session to abort (crashes). I have also tried residual = ~ dsum(~ id(units):us(trait) | site), which gives the error: Error in Rparam(data): Section has 520 observations but the residual model implies 1560.
<br />
<br />
Is it possible to do this operation, or has this option been removed in version 4?
<br />
<br />
Thanks!ASRemlhttp://forum.vsni.co.uk//posting.php?mode=reply&t=1886tshalevTue, 04 Dec 2018 23:51:30 GMThttp://forum.vsni.co.uk/viewtopic.php?p=5445#5445