blupf90 VS Plink VS GCTA 基因型数据PCA分析

邓飞

发布于 2019-06-13 20:21:35

1.3K0

发布于 2019-06-13 20:21:35

文章被收录于专栏：育种数据分析之放飞自我

这篇是软件测试. 主要是用同一批数据, 测试不同软件和方法对结果的影响, 不同方法有:

BLUPF90构建G矩阵, 计算PCA
BLUPF90构建H矩阵, 计算PCA
PLINK构建G矩阵, 计算PCA
GCTA构建G矩阵, 计算PCA

结论: BLUPF90构建G, H, Plink构建G矩阵, 结果一致. GCTA画的PCA有出入, 怀疑是默认参数的问题.

模拟一套数据, 5个世代, 最后三代有基因型数据, 每个世代400个个体, SNP为50K.

1. blupf90构建G矩阵的PCA

blupf90如果想要进行GBLUP分析, 不写系谱信息即可, 示例par文件:

DATAFILE
dat_f90.txt
TRAITS
10         # This is column 10 (phenotype) from QMSim data file
FIELDS_PASSED TO OUTPUT
1          # This will copy the ID number to the renf90.dat data file
WEIGHT(S)  # WARNING: ALWAYS PUT AN EMPTY LINE AFTER THIS!!!!!

RESIDUAL_VARIANCE
0.9                # add starting values for residual variance
EFFECT
4 cross alpha    # Fit generation as a fixed effect, 'cross alpha' is a class in SAS
EFFECT
1 cross alpha    # Fit animal effect
RANDOM
animal           # Fit animal effect (A matrix) for the effect directly above it (column 1, animal)
#FILE
#ped_f90.txt  #  pedigree file (animal, sire, dam), 0's are missing always!!!
#FILE_POS
#1 2 3 0 0        # indicates that column 1 = Animal, column 2 = Sire, column 3 = Dam
SNP_FILE
yM.txt
(CO)VARIANCES    
0.1               # add starting values for additive animal effect
OPTION alpha_size 25               # Equal to the max number of characters within a column
OPTION max_string_readline 800  # maximum number of characters in one line of data file
OPTION max_field_readline 100   # maximum number of columns in the dataset
#OPTION saveHinvOrig
#OPTION saveHinv
#OPTION sol se
#OPTION use_yams
#OPTION missing -999
OPTION plotpca

运行preGSf90后, 会生成pc1vspc2文件, 里面包括PC1和PC2两列, 增加世代为pop, 然后使用R画图:

pca = read.table(("pc1vspc2"))
head(pca)
names(pca) = c("PC1","PC2")
pca$pop = rep(c("A","B","C"),each=400)
library(ggplot2)
p <- ggplot(pca, aes(x=PC1, y=PC2, colour=pop)) 
p <- p + geom_point(size=2)
p <- p + stat_ellipse(level = 0.95, size = 1)
p <- p + geom_hline(yintercept = 0) 
p <- p + geom_vline(xintercept = 0) 
p <- p + theme_bw()
p

结果:

2. blupf90构建H矩阵的PCA

需要定义系谱和基因型, 示例par文件:

DATAFILE
dat_f90.txt
TRAITS
10         # This is column 10 (phenotype) from QMSim data file
FIELDS_PASSED TO OUTPUT
1          # This will copy the ID number to the renf90.dat data file
WEIGHT(S)  # WARNING: ALWAYS PUT AN EMPTY LINE AFTER THIS!!!!!

RESIDUAL_VARIANCE
0.9                # add starting values for residual variance
EFFECT
4 cross alpha    # Fit generation as a fixed effect, 'cross alpha' is a class in SAS
EFFECT
1 cross alpha    # Fit animal effect
RANDOM
animal           # Fit animal effect (A matrix) for the effect directly above it (column 1, animal)
FILE
ped_f90.txt  #  pedigree file (animal, sire, dam), 0's are missing always!!!
FILE_POS
1 2 3 0 0        # indicates that column 1 = Animal, column 2 = Sire, column 3 = Dam
SNP_FILE
yM.txt
(CO)VARIANCES    
0.1               # add starting values for additive animal effect
OPTION alpha_size 25               # Equal to the max number of characters within a column
OPTION max_string_readline 800  # maximum number of characters in one line of data file
OPTION max_field_readline 100   # maximum number of columns in the dataset
#OPTION saveHinvOrig
#OPTION saveHinv
#OPTION sol se
#OPTION use_yams
#OPTION missing -999
OPTION plotpca

运行preGSf90后, 会生成pc1vspc2文件, 里面包括PC1和PC2两列, 增加世代为pop, 然后使用R画图:

pca = read.table(("pc1vspc2"))
head(pca)
names(pca) = c("PC1","PC2")
pca$pop = rep(c("A","B","C"),each=400)
library(ggplot2)
p <- ggplot(pca, aes(x=PC1, y=PC2, colour=pop)) 
p <- p + geom_point(size=2)
p <- p + stat_ellipse(level = 0.95, size = 1)
p <- p + geom_hline(yintercept = 0) 
p <- p + geom_vline(xintercept = 0) 
p <- p + theme_bw()
p

3. plink根据G矩阵做PCA

代码:

plink --file b --pca 3

结果生成:

plink.eigenval  plink.eigenvec  plink.log  plink.nosex

R语言作图:

library(ggplot2)
head(dd)
p <- ggplot(dd, aes(x=PC1, y=PC2, colour=pop)) 
p <- p + geom_point(size=2)
p <- p + stat_ellipse(level = 0.95, size = 1)
# p <- p + scale_color_manual(values = cols)
p <- p + geom_hline(yintercept = 0) 
p <- p + geom_vline(xintercept = 0) 
p <- p + theme_bw()
p

4. gcta64根据G矩阵做PCA

将ped文件转化为bed文件

plink --file b --make-bed --out c

生成grm文件

gcta64 --bfile c --autosome --make-grm --out grm

生成pca文件

gcta64 --grm grm --pca 3

根据PCA信息作图

结论

blupf90的G矩阵, H矩阵, plink的PCA结果一致. GCTA构建的PCA结果不太一致, 怀疑是参数默认的有问题, 回头查看一下.

本文参与腾讯云自媒体同步曝光计划，分享自微信公众号。

原始发表：2019-04-28，如有侵权请联系 cloudcommunity@tencent.com 删除

数据分析

编程算法

本文分享自育种数据分析之放飞自我微信公众号，前往查看

如有侵权，请联系 cloudcommunity@tencent.com 删除。

本文参与腾讯云自媒体同步曝光计划，欢迎热爱写作的你一起参与！

数据分析

编程算法

登录后参与评论

0 条评论

热度