Dear Norman,
While doing other things I have been thinking about your problem
which is a larger application of genetic groups than I have previously
seen in ASREML.
:::
damage 1 !-72
damage2 1 !-5184
a2age 1 !m0 !-214
a2age2 1 !m0 !-45800
wean !m0
angped !MAKE !GROUPS 84 !NEW
angdat !maxit 1
wean ~ damage damage2 a2age a2age2 sex bstat treat !r anim dam,
damcpe 85.0178 hys 670.1648
1 1 1
0 0 0 !S2==308.1997
anim 2
2 0 US 169.8740 -59.6598 120.5122
anim
I tried centering the covariates and the response variable but
that did not change much.
Then I realised that ASREML is overfitting the genetic groups.
Basically, the genetic groups relate to fixed effects in the
model and by having these fitted to animal and dam, something
quite odd is happening. My naive expectation is that all the
genetic groups for DAM would be singular. However, the singular
effects are not not appearing in such a simple way. Rather
the singular effects are for groups
anim 24 27 29 30 31 32 34 36 37 38 60 68 69 72 73 77 78
dam 17 19 20 24 29 30 31 34 36 37 38 65 68 69 72 73 74 77 78
which is only 36 singularities.
The following table shows the nominal number of records in each group.
Some of these are quite small and so maybe should be collapsed.
I notice to that some sires [as coded in the data file] are allocated to
different genetic groups in the pedigree.
1 18.875 2 19.560 3 302.744 4 342.782 5 891.693
6 1806.936 7 956.594 8 1897.674 9 2066.808 10 2074.986
11 436.842 12 764.472 13 93.617 14 298.535 15 79.650
16 106.658 17 425.128 18 324.981 19 566.314 20 392.268
21 516.347 22 321.825 23 317.934 24 176.680 25 339.812
26 156.023 27 4.625 28 65.926 29 4.969 30 5.340
31 26.979 32 6.867 33 6.840 34 8.555 35 26.010
36 5.219 37 44.930 38 2.000 39 10.832 40 49.055
41 10.906 42 27.820 43 1005.355 44 869.233 45 1664.555
46 847.311 47 937.452 48 738.801 49 1618.136 50 312.113
51 692.805 52 30.875 53 17.750 54 79.463 55 89.689
56 413.128 57 303.169 58 472.117 59 316.398 60 209.846
61 111.752 62 413.503 63 305.169 64 554.152 65 328.898
66 118.273 67 20.719 68 3.312 69 4.715 70 67.793
71 26.947 72 6.750 73 6.688 74 4.031 75 3.809
76 44.234 77 25.822 78 13.863 79 4.625 80 1.094
81 1.094 82 40.305 83 0.906 84 9.738 28739 28739.000
I tried rescaling the data (by 100) in case the problem is failure
to detect singularities. This seems not the case.
I'm now not sure that I do expect 84 singularities as the genetic
group construction is probably based on these effects being at the
animal level. I think this needs a bit of investigation on a smaller
data set where we can fit the genetic groups in a direct manner.
Unfortunately I've run out of time before heading to UK to advance this much
further.
Arthur
ASREML will encounter the singularities in an unpredictable order
so what we have no expectation as to which effects will appear singular.
If solved in a different order, the singularities wuld probably move.
<><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
Arthur Gilmour PhD mailto:Arthur.Gilmour@agric.nsw.gov.au
Principal Research Scientist (Biometrics) fax: <61> 2 6391 3899
NSW Agriculture <61> 2 6391 3922
Orange Agricultural Institute telephone work: <61> 2 6391 3815
Forest Rd, ORANGE, 2800, AUSTRALIA home: <61> 2 6362 0046
ASREML is still free by anonymous ftp from pub/aar on ftp.res.bbsrc.ac.uk
or point your web browser at ftp://ftp.res.bbsrc.ac.uk/pub/aar/
To join the asreml discussion list, send the message
subscribe
mailto:asreml-request@chiswick.anprod.CSIRO.au
To send messages to the list, mailto:asreml@chiswick.anprod.CSIRO.au
Asreml list archive: http://www.chiswick.anprod.csiro.au/lists/asreml
<> <> <> <> <> <> <>
"Why do you seek the living among the dead?
He is not here but is risen!
Remember how He spoke to you while He was still in Galilee, saying
'The Son of Man must be delivered into the hands of sinful men,
and be crucified, and THE THIRD DAY RISE AGAIN.'" angels, Luke 24:5-7
Many claim to be guides in life. Only Jesus validated His claim by
living according to His own teaching and declaring His death and
resurrection before the event.
<><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
--
Asreml mailinglist archive: http://www.chiswick.anprod.csiro.au/lists/asreml