使用admixture软件做祖先成分分析小实例

参考文章

• 宇宙实验媛 公众号文章 《群体结构分析三种常用方法（下篇）》

admixture软件下载

http://software.genetics.ucla.edu/admixture/download.html
不需要安装，解压出来即可使用

数据

使用这篇文章中获得的 https://www.jianshu.com/p/5938ca3b6725 vcf 文件

第一步：plink格式转化

plink --vcf ../KiwifruitPathogenFiltered.recode.vcf --make-bed --out KiwifruitPathogen --allow-extra-chr

这里用到的plink版本是1.9

第二步：admixture软件做祖先成分分析

这里K随便选择一个5

admixture --cv KiwifruitPathogen.bed 5 | tee log5.out

遇到报错

****                   ADMIXTURE Version 1.3.0                  ****
****                    Copyright 2008-2015                     ****
****           David Alexander, Suyash Shringarpure,            ****
****                John  Novembre, Ken Lange                   ****
****                                                            ****
****                 Please cite our paper!                     ****
****   Information at www.genetics.ucla.edu/software/admixture  ****

Cross-validation will be performed.  Folds=5.
Random seed: 43
Point estimation method: Block relaxation algorithm
Convergence acceleration algorithm: QuasiNewton, 3 secant conditions
Point estimation will terminate when objective function delta < 0.0001
Estimation of standard errors disabled; will compute point estimates only.
Invalid chromosome code!  Use integers.

搜索最后一行报错Invalid chromosome code! Use integers找到解决办法
https://www.biostars.org/p/236704/
有人说将第一步plink生成的.bim文件第一列的内容改成整数
我这里的数据用到的是细菌单倍体，我将第一列内容统一改为1
使用vim编辑器批量替换字符串
https://www.cnblogs.com/nkwy2012/p/6365714.html
:%s/A/B/g:替换所有行中的A为B
再次运行以上命令就成功了，输出结果
意外的是将bim文件中的CM000染色体编号替换为数字1后，使用smartpca做主成分分析也成功了！

****                   ADMIXTURE Version 1.3.0                  ****
****                    Copyright 2008-2015                     ****
****           David Alexander, Suyash Shringarpure,            ****
****                John  Novembre, Ken Lange                   ****
****                                                            ****
****                 Please cite our paper!                     ****
****   Information at www.genetics.ucla.edu/software/admixture  ****

Cross-validation will be performed.  Folds=5.
Random seed: 43
Point estimation method: Block relaxation algorithm
Convergence acceleration algorithm: QuasiNewton, 3 secant conditions
Point estimation will terminate when objective function delta < 0.0001
Estimation of standard errors disabled; will compute point estimates only.
Size of G: 21x22612
Performing five EM steps to prime main algorithm
1 (EM)  Elapsed: 0.032  Loglikelihood: -68682.8 (delta): 609659
2 (EM)  Elapsed: 0.027  Loglikelihood: -64365.8 (delta): 4316.97
3 (EM)  Elapsed: 0.028  Loglikelihood: -62115.1 (delta): 2250.72
4 (EM)  Elapsed: 0.027  Loglikelihood: -60418   (delta): 1697.11
5 (EM)  Elapsed: 0.028  Loglikelihood: -59017.1 (delta): 1400.89
Initial loglikelihood: -59017.1
Starting main algorithm
1 (QN/Block)    Elapsed: 0.117  Loglikelihood: -31056.4 (delta): 27960.7
2 (QN/Block)    Elapsed: 0.117  Loglikelihood: -24590.8 (delta): 6465.61
3 (QN/Block)    Elapsed: 0.15   Loglikelihood: -21524.3 (delta): 3066.49
4 (QN/Block)    Elapsed: 0.134  Loglikelihood: -20417.1 (delta): 1107.22
5 (QN/Block)    Elapsed: 0.136  Loglikelihood: -20102.6 (delta): 314.425
6 (QN/Block)    Elapsed: 0.159  Loglikelihood: -19876   (delta): 226.682
7 (QN/Block)    Elapsed: 0.147  Loglikelihood: -19726.1 (delta): 149.867
8 (QN/Block)    Elapsed: 0.132  Loglikelihood: -19708.2 (delta): 17.8477
9 (QN/Block)    Elapsed: 0.148  Loglikelihood: -19707.1 (delta): 1.15903
10 (QN/Block)   Elapsed: 0.131  Loglikelihood: -19707.1 (delta): 0.000143719
11 (QN/Block)   Elapsed: 0.139  Loglikelihood: -19707.1 (delta): 1.3288e-05
Summary: 
Converged in 11 iterations (1.755 sec)
Loglikelihood: -19707.085953
Fst divergences between estimated populations: 
    Pop0    Pop1    Pop2    Pop3    
Pop0    
Pop1    0.774   
Pop2    0.746   0.829   
Pop3    0.865   0.773   0.894   
Pop4    0.780   0.705   0.793   0.811   
CV error (K=5): 0.15979
Writing output files.

但是如何对结果进行解读还得继续仔细看文献呀！

欢迎大家关注我的公众号小明的数据分析笔记本

公众号二维码.jpg

如果大家需要文章中用到的vcf文件的话，关注公众号后留言 KiwiFruit就可以啦！