博文
ASReml-R与协变量模型分析
||
实际工作中，常常会遇到不同年份的试验，数据如何整合一起分析。这里举个简单的例子，有一片种源试验林，其中2块为8年林，一块为9年林。把上述3片林一起分析，进而选择表现好的种源。这时，就可以将年份作为协变量，引入线性模型中，消除年份的影响，从而可以筛选出目标种源。
首先，数据的部分格式如下：
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 | > head(df) Treeid Mum Dad Provence Plustree Family Block Block1 Line 1 10001 263 <NA> 23 L60 263 B 2 8 2 10002 237 <NA> 23 L5 237 B 2 8 3 10003 234 <NA> 23 L2 234 B 2 8 4 10004 260 <NA> 23 L56 260 B 2 8 5 10005 236 <NA> 23 L4 236 B 2 8 6 10006 214 <NA> 22 22003 214 B 2 8 Linetree Trail year d10 1 1 gd 8 10.4 2 10 gd 8 NA 3 11 gd 8 5.2 4 12 gd 8 12.8 5 13 gd 8 12.2 6 14 gd 8 15.5 |
这里，使用种源、家系和个体模型，代码如下：
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 | ## 种源模型 m1<-asreml(d10~year+Block, random=~Provence+Block:Line, data=df) ## 家系模型 m2<-asreml(d10~year+Block, random=~Provence+Provence/Family+Block:Line, data=df) ## 单株模型 m3<-asreml(d10~year+Block, random=~ped(Treeid)+Block:Line, ginverse=list(Treeid=pedinv),data=df) summary(m1)$varcomp wald(m1) |
运行结果如下：
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 | ## 种源模型 > summary(m1)$varcomp gamma component std.error Provence!Provence.var 0.02190295 0.1751748 0.1139682 Block:Line!Block.var 0.20059090 1.6042807 0.2134393 R!variance 1.00000000 7.9977740 0.1953053 z.ratio constraint Provence!Provence.var 1.537050 Positive Block:Line!Block.var 7.516331 Positive R!variance 40.950106 Positive > wald(m1) Wald tests for fixed effects Response: d10 Terms added sequentially; adjusted for those above Df Sum of Sq Wald statistic Pr(Chisq) (Intercept) 1 39518 4941.1 < 2.2e-16 *** year 1 511 63.9 1.332e-15 *** Block 14 1260 157.6 < 2.2e-16 *** residual (MS) 8 --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 ## 家系模型 > summary(m2)$varcomp gamma component std.error Provence!Provence.var 0.01663476 0.1247737 0.1130592 Provence:Family!Provence.var 0.07452158 0.5589699 0.1154963 Block:Line!Block.var 0.22213146 1.6661590 0.2170419 R!variance 1.00000000 7.5007789 0.1869149 z.ratio constraint Provence!Provence.var 1.103614 Positive Provence:Family!Provence.var 4.839722 Positive Block:Line!Block.var 7.676671 Positive R!variance 40.129378 Positive > wald(m2) Wald tests for fixed effects Response: d10 Terms added sequentially; adjusted for those above Df Sum of Sq Wald statistic Pr(Chisq) (Intercept) 1 36040 4804.8 < 2.2e-16 *** year 1 444 59.3 1.388e-14 *** Block 14 1193 159.1 < 2.2e-16 *** residual (MS) 8 --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 ## 单株模型 > summary(m3)$varcomp gamma component std.error z.ratio ped(Treeid)!ped 0.4356748 2.464560 0.4825557 5.107306 Block:Line!Block.var 0.2948261 1.667796 0.2172520 7.676779 R!variance 1.0000000 5.656878 0.4210363 13.435606 constraint ped(Treeid)!ped Positive Block:Line!Block.var Positive R!variance Positive > wald(m3) Wald tests for fixed effects Response: d10 Terms added sequentially; adjusted for those above Df Sum of Sq Wald statistic Pr(Chisq) (Intercept) 1 51931 9180.1 < 2.2e-16 *** year 1 491 86.8 < 2.2e-16 *** Block 14 939 166.1 < 2.2e-16 *** residual (MS) 6 --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 |
从结果看，年份作为协变量的效应是显著的。
种源的筛选可以通过育种值来获取，将部分育种值展示如下。
1 2 3 4 5 6 7 | > summary(m1,all=T)$coef.random[1:5,] solution std error z ratio Provence_11 0.26767555 0.2443300 1.0955494 Provence_12 -0.34572466 0.2500429 -1.3826613 Provence_13 0.03588465 0.2311507 0.1552435 Provence_15 0.24067021 0.2664837 0.9031329 Provence_17 -0.19850575 0.2324755 -0.8538781 |
此外，还有家系和个体模型，根据对应的育种值，也可筛选出优良的家系或个体。
https://blog.sciencenet.cn/blog-1114360-817794.html
上一篇：ASReml-R与阈性状(threshold trait)分析
下一篇：random regression model之ASReml V4篇
