关于Echidna软件
Echidna Mixed Models Software. Developed free for non-commercial use. Echidna performs statistical analysis by fitting linear mixed models utilising the Average Information algorithm to estimate variance parameters. Echidna is targeted for use in animal and plant breeding contexts.
程序示例
!RML !DEBUG !LOG !OUT !REN !ARG 1 3 # 2 !VIEW
Lamb data from ARG thesis page 177-8 !DOPART $1
CYR GRP 5
SEX SIRE !I
TOT
L5 !/V5 # Express as a proportion
L4 LS LR
SEC 2 !=V0 !>30 !+1
# 1 1 1 101 39 33 6 6 1
lamb.dat !skip 1 !DDF -1 !SLN #!continue
TABULATE L5 ~ GRP !STATs !DECIMAL 1
!PART 1
L5 !bin !TOTAL=TOT !DISP 1 ~ mu SEX GRP !r SIRE .16783
predict SIRE
VPREDICT
F Total Residual*3.289 + SIRE
F Genetic SIRE*4
H Heritability Genetic Total
!PART 2
L5 !bin !PROBIT !TOTAL=TOT !DISP 1 ~ mu SEX GRP !r SIRE .16783
!PART 3
L5 !bin !COMPLOGLOG !TOTAL=TOT !DISP 1 ~ mu SEX GRP !r SIRE .16783
residual at(SEC).id(units)
predict GRP
读入R中
# Echidna
setwd("D:/Echidna/Jobs/lamb3")
res<-esRT(trace=T)
读入过程
> res<-esRT(trace=T)
Loading lamb1
Thu Dec 12 18:13:04 2019
Iteration LogL NEDF
1 1 -10.396 62
2 2 -9.425 62
3 3 -9.116 62
4 4 -9.120 62
5 5 -9.121 62
6 6 -9.121 62
Thu Dec 12 18:13:05 2019 LogL Converged
Sussessfully input Echidna runs results into R.
生成的R对象特征
> names(res)
[1] "jobname" "Version" "StartTime"
[4] "Title" "Iterations" "Record.Count"
[7] "LogLikelihood" "Residual.DF" "ICparameter.Count"
[10] "AkaikeIC" "BayesianIC" "Traits"
[13] "Wald.F.table" "Components" ""
[16] "FinishAt" "Converge" "coeff"
[19] "yht" "pred" "evp"
> class(res)
[1] "esR"
模型诊断
// A code block
plot(res)
固定效应检验
> wald.esR(res,rmS=c(-3,-6))
Wald tests for fixed effects.
Source NumDF F-inc F-con
1 mu 1 58.86 58.86
2 SEX 1 4.07 3.62
3 GRP 4 21.29 21.29
模型IC
> IC.esR(res)
DF LogL AIC BIC
1 62 -9.120831 20.24166 22.3688
模型方差分量
> Var.esR(res)
Term Level Gamma Sigma std.err Z.ratio %ch const
1 SIRE 18 0.0678388 0.0678388 0.03832701 1.77 0 P
2 Residual_units 68 1.0000000 1.0000000 Inf 0.00 0
遗传参数估算
// An highlighted block
> vpin(res)
Term Component SEcomp
1 Residual 1.000000 0.000000
2 SIRE 0.067839 0.038413
3 Total 3.356800 0.038413
4 Genetic 0.271360 0.153650
5 Heritability 0.080840 0.044850
Notice: The parameter estimates are followed by
their approximate standard errors. 固定效应解
// An highlighted block
> coef(res)$fixed
Term Level Solution StndError
1 GRP 1 0.0000000 0.0000000
2 GRP 2 -0.7396206 0.1446669
3 GRP 3 -1.8166377 0.2270613
4 GRP 4 -1.2381634 0.1897600
5 GRP 5 -1.0310191 0.1729074
6 SEX 1 -0.1638923 0.0861600
7 mu 1 1.5806513 0.1519141
随机效应解
// An highlighted block
> coef(res)$random
Term Level Solution StndError
1 SIRE 1 0.2714009 0.1604427
2 SIRE 2 -0.1085139 0.1579248
3 SIRE 3 -0.0643852 0.1593922
4 SIRE 4 -0.2162157 0.1674885
5 SIRE 5 -0.0902404 0.1671876
6 SIRE 6 -0.3787981 0.1608487
7 SIRE 7 -0.0544110 0.1591641
8 SIRE 8 0.2731381 0.1562769
9 SIRE 9 0.1625571 0.1674780
...
模型预测
// An highlighted block
> mm<-predict(res)
> mm$pred
Predicted_Value Stnd_Error Ecode SIRE
1 0.80502 0.14047 E 1
2 0.42510 0.13822 E 2
3 0.46923 0.13979 E 3
4 0.31740 0.15837 E 4
5 0.44338 0.15806 E 5
6 0.15482 0.14908 E 6
7 0.47921 0.15657 E 7
8 0.80676 0.16027 E 8
9 0.69617 0.17317 E 9
...
模型运行结果
一个简单的表格是这么创建的：
| 项目 | Value |
|---|---|
| .esr | $1600 |
| .ess | $12 |
| .esy | $1 |
| .epv | $1 |
| .evp | $1 |
| .esy | $1 |
脚的文本。1
注脚的解释 ↩︎
