当前位置：网站首页>Piblup test report 1- pedigree based animal model

Piblup test report 1- pedigree based animal model

2022-06-27 15:00:00 【Analysis of breeding data】

PIBLUP Software github Address

PIBLUP

Download and install MKL

MKL yes Intel The library of , Need to register , Download and install

Download a software , decompression , Enter the path :

       
        cd l_mkl_2018.3.222/
       
       
        1.

Then type the following command to install

       
        ./install.sh
       
       
        1.

install MPI

       
        yum install mpich
       
       
        1.

Animal models ( Pedigree )

Model

y i j k = s e x i + b i r t t h M o n j + b i r t h W e i g h t + a k + e i j k y_{ijk} = sex_i + birtthMon_j + birthWeight_ + a_k + e_{ijk} yijk=sexi+birtthMonj+birthWeight+ak+eijk

among ,

Fixed factor : sex, birthMon

Covariant quantity : birthWeight

Random factors : Additive effect (a)

Phenotypic data ( part )

Clothes : ID, birthMonth, Sex, birthWeight, weight1, weight2

The missing value is -99, weight1 Missing values , weight2 No missing value .

       
        192243 2 2 27.50 147.35 147.35
        
192240 3 2 26.00 -99 124.91
        
192242 3 2 29.50 142.78 142.78
        
192246 3 1 30.30 143.43 143.43
        
192241 3 1 36.20 147.95 147.95
        
192251 3 2 36.00 154.93 154.93
        
192245 3 2 32.00 142.20 142.20
        
192247 3 1 31.00 148.84 148.84
        
192250 3 1 29.00 123.64 123.64
        
192249 3 2 24.80 140.96 140.96
       
       
        1.
        2.
        3.
        4.
        5.
        6.
        7.
        8.
        9.
        10.

Pedigree data ( part )

ID, Sire, Dam

       
        192243  181 4007
        
192240  3980  3762
        
192242  3980  4010
        
192246  3980  4899
        
192241  847 3525
        
192251  4597  4204
        
192245  3920  3588
        
192247  3980  4341
        
192250  4597  4464
        
192249  3980  184
       
       
        1.
        2.
        3.
        4.
        5.
        6.
        7.
        8.
        9.
        10.

PIBLUP command

       
        DATAFILE data2.txt #  Data name 
        
NTRAITS 1 #  Single sex analysis 
        
TRAITS 5 #  The traits are in the fifth column 
        
NEFFECTS 4 #  share 4 Two effect values (ID, birthMonth, birthWeight, Sex)
        
NVAR 6 #  Data is shared 6 Column 
        
MISSING -99 #  The missing value is -99
        
WEIGHT 0 #  No, weight Variable 
        
EFFECTS #  In the effect value ,  The second, the third fixed factor ,  The fourth covariate ,
        
1 2/F 3/F 4/FR,0 1/R,1,1
        
COV 1 # G Matrix variance components 0.3
        
0.3
        
RCOV # R matrix   Variance components 0.7
        
0.7
        
COVFILE 1 PED ped.txt 0 0 #  A pedigree ,  No inbreeding ,  No genetic group 
       
       
        1.
        2.
        3.
        4.
        5.
        6.
        7.
        8.
        9.
        10.
        11.
        12.
        13.
        14.

result (BLUP value )

       
        1 1 2 2 0 43.9615
        
1 1 2 3 0 26.2066
        
1 1 2 4 0 15.3291
        
1 1 2 5 0 29.9283
        
1 1 2 6 0 29.0748
        
1 2 3 2 0 25.3714
        
1 2 3 1 0 22.4129
        
1 3 4 0 0 2.74602
        
1 4 1 192243 0 2.50189
        
1 4 1 192242 0 5.17704
       
       
        1.
        2.
        3.
        4.
        5.
        6.
        7.
        8.
        9.
        10.

asreml Comparison results

       
        !WORKSPACE 100 !RENAME !outfolder !ARGS 1// !DOPART $1
        
Title: data2.
        
#192243 2 2 27.50 147.35 147.35
        
#192240 3 2 26.00 -99 124.91
        
#192242 3 2 29.50 142.78 142.78
        
#192246 3 1 30.30 143.43 143.43
        
#192241 3 1 36.20 147.95 147.95
        
 ID  !P      # 192241
        
 birth_mon  !A
        
 sex  !A       # 3
        
 birth_we        # 36.20
        
 y1 !M -99           # 147.95
        
 y2        # 147.95
        
# Check/Correct these field definitions.
        
ped.txt
        
data2.txt
        

        
!part 1
        
!sigmap
        
y1  ~ mu sex birth_mon birth_we,         # Specify fixed model
        
      !r  ID  0.3 !GF       # Specify random model
        
residual idv(units 0.7 !GF)
        
predict ID
       
       
        1.
        2.
        3.
        4.
        5.
        6.
        7.
        8.
        9.
        10.
        11.
        12.
        13.
        14.
        15.
        16.
        17.
        18.
        19.
        20.
        21.
        22.
        23.

Results comparison

       
        library(data.table)
        
library(tidyverse)
        

        
piblup <- fread("1.par.sol.0")
        
dim(piblup)
        
head(piblup,10)
        

        
piblup1 <- piblup[9:351,c(4,6)]
        
head(piblup1)
        

        
asblup <- fread("data21/data2.sln")
        
head(asblup,10)
        
dim(asblup)
        

        
asblup1 <- asblup[10:352,c(2,3)]
        

        
head(asblup1)
        
head(piblup1)
        

        
names(piblup1) <- c("Level","pi-Effect")
        

        
re <- merge(piblup1,asblup1,"Level")
        
head(re)
        
names(re) <- c("ID","PIBLUP","ASReml-blup")
        
head(re)
        
cor(re$PIBLUP,re$`ASReml-blup`)
        
plot(re$PIBLUP,re$`ASReml-blup`)
        

        
head(re)
       
       
        1.
        2.
        3.
        4.
        5.
        6.
        7.
        8.
        9.
        10.
        11.
        12.
        13.
        14.
        15.
        16.
        17.
        18.
        19.
        20.
        21.
        22.
        23.
        24.
        25.
        26.
        27.
        28.
        29.

Comparison results

       
        > head(re)
        
    ID    PIBLUP ASReml-blup
        
1:  67  4.075890      4.0760
        
2: 131  1.477880      1.4780
        
3: 140 -0.625255     -0.6255
        
4: 181  5.004880      5.0090
        
5: 184  2.872590      2.8730
        
6: 188  4.016820      4.0200
       
       
        1.
        2.
        3.
        4.
        5.
        6.
        7.
        8.

The correlation coefficient

       
        > cor(re$PIBLUP,re$`ASReml-blup`)
        
[1] 1
       
       
        1.
        2.

At two o 'clock

1, asreml in , Even if the initial value is not set , The result is the same , The variance components are determined by 0.3, rose 50, But heritability doesn't change .
2, weight1 and weight2, The results are basically the same . explain PIBLUP Can handle missing values .
3, The effect value of fixed factor is different , however BLUP Value consistent .
4, PIBLUP What is given is not predict means, It is BLUP value .

原网站

版权声明
本文为[Analysis of breeding data]所创，转载请带上原文链接，感谢
https://yzsam.com/2022/02/202202161606392597.html