Experiences of Joint Nordic Genetic Evaluation

1
Experiences of Joint Nordic Genetic Evaluation
Nordic Cattle Genetic Evaluation

• Jarmo Juga, Martin Lidauer, Jukka Pösö, Jörn
  Pedersen, Ole Maagaard Pedersen, Anders Fogh,
  Anki Roth, Bjorg Heringstad

2
Nordic Cattle Genetic Evaluationstarted 1.1.2002

• Dansk Kvaeg, Denmark
• Geno, Norway
• FABA, Finland
• Svensk Mjölk, Sweden
• Budget about 5 million DKK

3
Development projects

• Milk traits
• Denmark and Finland
• Fertility traits
• Denmark and Sweden
• Mastitis
• Norway -gt ?
• Type traits
• Denmark

4
Experiences

• The timetables of the projects were too
  optimistic
• The models are running, but need fine tuning
• Milk traits heterosis, genetic groups, modelling
  later lactations in Denmark
• Fertility modelling the fixed effects
• Mastitis under estimation of the trend
• Type traits standardisation across countries

5
Heterozygosity and breed proportions in Denmark.
Friesian cattle (BW), Holstein (US)
6
Heterozygosity and breed proportions in Sweden.
Friesian cattle (BW), Holstein (US)
7
Heterozygosity and breed proportions in Finland.
Friesian cattle (BW), Holstein (US)
8
Test day model 305 records

• Unified model Esa Mäntysaari (2002)
• 305 days records are used as a special case of a
  test day record -gt unified model

Hol gt 70 milj obs., Reds gt 40 milj. obs
9
(No Transcript)
10
Trait groups in fertility
NRR Non return rate ICF Interval from calving
to first insemination IFL Interval from first to
last insemination HST Heat strength AIS Number
of ins. Per pregnancy FT Fertility disorders
(treatments)
11
Summary statistics for the joint Nordic HOL
dataset for mastitis.
12
Models tested in mastitis

• Model R
• Ma50 CxAge1 CMY1 hy1 sire
• Ma51_300 CxAge1 CMY1 hy1 sire
• Ma2_150 CxAge2 CMY2 hy2 sire
• Ma3_150 CxAge3 CMY3 hy3 sire
• Model FR
• Ma50 CxAge1 CMY1 H5Y1 hy1 sire
• Ma51_300 CxAge1 CMY1 H5Y1 hy1 sire
• Ma2_150 CxAge2 CMY2 H5Y2 hy2 sire
• Ma3_150 CxAge3 CMY3 H5Y3 hy3 sire
• Model F
• Ma50 CxAge1 CMY1 HY1 sire
• Ma51_300 CxAge1 CMY1 HY1 sire
• Ma2_150 CxAge2 CMY2 HY2 sire
• Ma3_150 CxAge3 CMY3 HY3 sire

• CxAge1 Fixed effect of CountryxAge of first
  calving, age at first calving in months (16
  classes). The effect had a total of 48 classes.
• CxAge2 Fixed effect of CountryxAge of second
  calving,
• CxAge3 Fixed effect of CountryxAge of third
  calving,
• CMY1 Fixed effect of CountryxMonthxYear of
  first calving,
• CMY2 Fixed effect of CountryxMonthxYear of
  second calving,
• CMY3 Fixed effect of CountryxMonthxYear of
  third calving,
• H5Y1 Fixed effect of Herdx5-Year-period of
  first calving,
• H5Y2 Fixed effect of Herdx5-Year-period of
  second calving,
• H5Y3 Fixed effect of Herdx5-Year-period of
  third calving,
• HY1 Fixed effect of HerdxYear of first calving,
• HY2 Fixed effect of HerdxYear of second
  calving,
• HY3 Fixed effect of HerdxYear of third calving,
• hy1 Random effect of HerdxYear of first
  calving,
• hy2 Random effect of HerdxYear of first
  calving,
• hy3 Random effect of HerdxYear of first
  calving, and
• sire Random effect of sire of the cow.

13
Genetic trends given as mean PTA by birth year of
sire, for sires born from 1979 to 1998, for the
four mastitis traits in Nordic Holstein.
14
Validation criteria

• Comparison to national models and current
  official EBVs (PTAs)
• Trend, ranking of animals (correlation), s.d. of
  EBVs
• Comparison to current Interbull evaluations
• Trend, genetic level, ranking of bulls,
  correlation
• Interbull tests 1, 2 and 3, where applicable
• Improving IB2 for test day models
• Predictive ability

15
Evaluation scheme
X
16
NAVcluster in Foulum, Denmark

• Parallel computing
• 12 nodes
• 26 Gb RAM
• 1 Tb Disk95 Gb per node
• Metamodel in two days
• All RD
• All service runs

17
Testing the NAV cluster with Holstein data
18
Conclusions

• We can work it out!
• The goal is to start the joint evaluation with
  first traits in autumn 2004