DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)

发布时间:2022-10-05 21:30

点击关注,桓峰基因

桓峰基因

生物信息分析,SCI文章撰写及生物信息基础知识学习:R语言学习,perl基础编程,linux系统命令,Python遇见更好的你

134篇原创内容

公众号

桓峰基因公众号推出基于基因组数据生信分析教程并配有视频在线教程,目前整理出来的教程目录如下:

DNA 1. Germline Mutation Vs. Somatic Mutation 傻傻分不清楚
DNA 2. SCI 文章中基因组变异分析神器之 maftools
DNA 3. SCI 文章中基因组变异分析神器之 maftools
DNA 4. SCI 文章中基因组的突变信号(maftools)
DNA 5. 基因组变异文件VCF格式详解
DNA 6. 基因组变异之绘制精美瀑布图(ComplexHeatmap)
DNA 7. 基因组拷贝数变异分析及可视化 (GISTIC2.0)
DNA 8. 癌症的突变异质性及寻找新的癌症驱动基因(MutSigCV)
DNA 9. 揭秘肿瘤异质性与TMB, MSI之间的相关性
DNA 10. 识别癌症驱动基因 (OncodriveCLUST)
DNA 11. 识别肿瘤蛋白质三维结构上突变热点(HotSpot3D)
DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)


这期介绍一下GWAS分析的可视化软件包CMplot,对于研究SNP全体分析,家系遗传等方向的非常实用,给自己的文章添点色彩感。

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第1张图片

我们来看看这个软件包能做的事情,GWAS可视化算是很OK了,记得这方面的分析可以考虑使用了!

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第2张图片

前言

全基因组关联分析(GWAS)是在某一特定人群中研究遗传突变和表型之间的相关性。GWAS的理论基础是连锁不平衡定律(linkage disequilibrium,LD),既假设观察到的SNP与真正的致病突变(causal variant)之间存在很强的LD全基因组关联分析(Genome wide association study,GWAS)是对多个个体在全基因组范围的遗传变异(标记)多态性进行检测,获得基因型,进而将基因型与可观测的性状,即表型,进行群体水平的统计学分析,根据统计量或显著性 p 值筛选出最有可能影响该性状的遗传变异(标记),挖掘与性状变异相关的基因。因此,GWAS分析需要准备的文件主要有:基因型数据和表型数据。

表型数据统计分析方法包括:

  1. 逻辑回归(表型数据为二元)

  2. 线性回归(表型数据为连续性变量)

  3. 表型数据正态分析(如果不是正态分布,需转换处理为正态分布)

  4. 表型数据均值、中值、最大值、最小值

  5. 影响因子对表型的影响分析

基因型数据一般是测序下机数据经SNP calling、基因型填补等后拿到的.vcf文件,后续根据研究者的要求和条件进行质控(Quality control),群体结构分层及亲缘关系/PCA等处理。

软件安装

我们使用CMplot软件包来实现GWAS绘制Manhattan图

if (!require(CMplot)) install.packages("CMplot")

library("CMplot")

参数说明

这个CMplot函数的参数还是蛮多的,毕竟是绘制这种复杂的图像,这也是避免不了的事情,我们大概解释一下参数。

Usage CMplot(Pmap, col=c(“#4197d8”, “#f8c120”, “#413496”, “#495226”, “#d60b6f”, “#e66519”, “#d581b7”, “#83d3ad”, “#7c162c”, “#26755d”), bin.size=1e6, bin.range=NULL, bin.legend.num=10, pch=19, type=“p”, band=1, H=1.5, ylim=NULL, cex.axis=1, lwd.axis=1.5, cex.lab=1.5, plot.type=“b”, multracks=FALSE, points.alpha=100L, cex=c(0.5,1,1), r=0.3, outward=FALSE, ylab=expression(-log[10](italic§)), ylab.pos=3, xticks.pos=1, mar = c(3,6,3,3), threshold = NULL, threshold.col=“red”, threshold.lwd=1, threshold.lty=2, amplify= TRUE, signal.cex = 1.5, signal.pch = 19, signal.col=NULL, signal.line=2, highlight=NULL, highlight.cex=1, highlight.pch=19, highlight.type=“p”, highlight.col=“red”, highlight.text=NULL, highlight.text.col=“black”, highlight.text.cex=1, highlight.text.xadj=NULL, highlight.text.yadj=NULL, highlight.text.font=3, chr.labels=NULL, chr.border=FALSE, chr.labels.angle=0, chr.den.col=“black”, chr.pos.max=FALSE, cir.band=1, cir.chr=TRUE, cir.chr.h=1.5, cir.legend=TRUE, cir.legend.cex=0.6, cir.legend.col=“black”, LOG10=TRUE, box=FALSE, conf.int=TRUE, conf.int.col=NULL, file.output=TRUE, file=c(“jpg”,“pdf”,“tiff”), dpi=300, height=NULL, width=NULL, memo=“”, main=“”, main.cex=1.5, main.font=2, trait.legend.ncol=NULL, verbose=TRUE)

参数细节如下:

Pmap 输入数据文件

col 设置颜色,可以是vector or matrix

当vector数量少于染色体数目时,循环利用;也可以对每一个性状的每一条染色体进行设置

bin.size 设置SNP密度图中的窗口大小

bin.max bin中SNP数量的阈值,当大于阈值时染色体bin颜色为同一颜色

cex 设置绘制点的大小

pch 设置绘制点的形状,同plot中的"pch"

band 设置染色体之间的间隔,当为0时染色体间无空隙,默认为1

cir.band 设置不同circle的空隙,默认为1

H 性状circle的高度设置,默认1

ylim 设置y轴的范围同plot中的"ylim"

cex.axis 设置坐标轴字体和标签字体的大小

bin.size 设置SNP密度图中的窗口大小

cex.axis 设置坐标轴字体和标签字体的大小

plot.type 设置不同的绘图类型,可以设定为 “d”, “c”, “m”, “q” or “b”

“d” 表示 SNP density plot “c” 表示 circle-Manhattan plot

“m” 表示 Manhattan plot “q” 表示 Q-Q plot

“b” 表示 circle-Manhattan, Manhattan and Q-Q plots一起绘制

plot.type=c(“m”,“q”) 表示Manhattan plot和Q-Q plot一起绘制

multracks 设置是否需要绘制多个track

cex 绘制点的大小,可是为单个数值或向量(对应同一绘图中不同的plot)

r 设置圈的半径大小

xlab 设置x轴标签

ylab 设置y轴标签

outward 设置点的朝向是否向外

threshold 设置阈值并添加阈值线

threshold.col 设置阈值线的颜色

threshold.lwd 设置阈值线的宽度

threshold.lty 设置阈值线的类型

amplify 设置是否放大显著的点

chr.labels 设置染色体的标签

signal.cex 设置显著点的大小

signal.pch 设置显著点的形状

signal.col 设置显著点的颜色

singal.line 设置显著线的宽度

chr.den.col 设置SNP密度图的颜色

cir.band 设置环状曼哈顿图中不同染色体之间的间隔

cir.chr 设置是否显示染色体的边界

cir.chr.h 设置染色体边界的高度

chr.den.col SNP密度的颜色,可以为字符,向量或空

cir.legend 设置是否显示图例

cir.legend.cex 设置图例字体的大小

cir.legend.col 设置图例的颜色

LOG10 设置是否对p-value取log10对数

box 曼哈顿是否绘制方框

conf.int.col 设置QQ图中置信区间的颜色

file.output 设置是否输出图片

file 设置输出图片的格式,可以设定为"jpg", “pdf”, “tiff”

dpi 设置输出图片的分辨度

memo 设置输出图片文件的名字

数据读取

我们这里选择使用软件包自带的两个数据集,方便大家理解。

Genotyped by pig 60k chip Description This dataset gives the results of Genome-wide association study of 3 traits, individuals were genotyped by pig 60K chip. Format A dataframe containing 3 traits’ Pvalue

data(pig60K)  #calculated p-values by MLM
head(pig60K)
##           SNP Chromosome Position    trait1     trait2     trait3
## 1 ALGA0000009          1    52297 0.7738187 0.51194318 0.51194318
## 2 ALGA0000014          1    79763 0.7738187 0.51194318 0.51194318
## 3 ALGA0000021          1   209568 0.7583016 0.98405289 0.98405289
## 4 ALGA0000022          1   292758 0.7200305 0.48887140 0.48887140
## 5 ALGA0000046          1   747831 0.9736840 0.22096836 0.22096836
## 6 ALGA0000047          1   761957 0.9174565 0.05753712 0.05753712

Genotyped by Bovine50K chip Description This dataset gives the effects of all SNPs using rrBLUP algorithm for 3 traits, individuals were genotyped by Bovine50K chip. Format A dataframe containing 3 traits’ SNPs effects

data(cattle50K)  #calculated SNP effects by rrblup
head(cattle50K)
##    SNP chr    pos Somatic cell score  Milk yield Fat percentage
## 1 SNP1   1  59082        0.000244361 0.000484255    0.001379210
## 2 SNP2   1 118164        0.000532272 0.000039800    0.000598951
## 3 SNP3   1 177246        0.001633058 0.000311645    0.000279427
## 4 SNP4   1 236328        0.001412865 0.000909370    0.001040161
## 5 SNP5   1 295410        0.000090700 0.002202973    0.000351394
## 6 SNP6   1 354493        0.000110681 0.000342628    0.000105792

实例操作

这里的例子都是来之文档说明,我就当一次搬运工,有需要使用自己数据分析的,搞不定可以找桓峰基因,我们随时在线等!话不多说,上代码就可以了。

使用例子:https://github.com/YinLiLin/CMplot

1. SNP-density plot

绘制SNP密度图

CMplot(pig60K, type = "p", plot.type = "d", bin.size = 1e+06, chr.den.col = c("darkgreen",
    "yellow", "red"), file = "jpg", memo = "", dpi = 300, main = "SNP-density plot",
    file.output = TRUE, verbose = TRUE, width = 9, height = 6)
##  SNP-Density Plotting.
##  Plots are stored in: F:/demo script/基因组分析系列/CMplot

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第3张图片

2. Circular-Manhattan plot

绘制圆形Manhattan图

(1) Genome-wide association study(GWAS)

CMplot(pig60K, type = "p", plot.type = "c", chr.labels = paste("Chr", c(1:18, "X",
    "Y"), sep = ""), r = 0.4, cir.legend = TRUE, outward = FALSE, cir.legend.col = "black",
    cir.chr.h = 1.3, chr.den.col = "black", file = "jpg", memo = "", dpi = 300, file.output = TRUE,
    verbose = TRUE, width = 10, height = 10)
##  Circular-Manhattan Plotting trait1.
##  Circular-Manhattan Plotting trait2.
##  Circular-Manhattan Plotting trait3.
##  Plots are stored in: F:/demo script/基因组分析系列/CMplot

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第4张图片

添加色带的Manhattan图

CMplot(pig60K, type = "p", plot.type = "c", r = 0.4, col = c("grey30", "grey60"),
    chr.labels = paste("Chr", c(1:18, "X", "Y"), sep = ""), threshold = c(1e-06,
        1e-04), cir.chr.h = 1.5, amplify = TRUE, threshold.lty = c(1, 2), threshold.col = c("red",
        "blue"), signal.line = 1, signal.col = c("red", "green"), chr.den.col = c("darkgreen",
        "yellow", "red"), bin.size = 1e+06, outward = FALSE, file = "jpg", memo = "",
    dpi = 300, file.output = TRUE, verbose = TRUE, width = 10, height = 10)
##  Circular-Manhattan Plotting trait1.
##  Circular-Manhattan Plotting trait2.
##  Circular-Manhattan Plotting trait3.
##  Plots are stored in: F:/demo script/基因组分析系列/CMplot

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第5张图片

(2) Genomic Selection/Prediction(GS/GP)

基因组选择或者预测

CMplot(cattle50K, type = "p", plot.type = "c", LOG10 = FALSE, outward = TRUE, col = matrix(c("#4DAF4A",
    NA, NA, "dodgerblue4", "deepskyblue", NA, "dodgerblue1", "olivedrab3", "darkgoldenrod1"),
    nrow = 3, byrow = TRUE), chr.labels = paste("Chr", c(1:29), sep = ""), threshold = NULL,
    r = 1.2, cir.chr.h = 1.5, cir.legend.cex = 0.5, cir.band = 1, file = "jpg", memo = "",
    dpi = 300, chr.den.col = "black", file.output = TRUE, verbose = TRUE, width = 10,
    height = 10)
##  Circular-Manhattan Plotting Somatic cell score.
##  Circular-Manhattan Plotting Milk yield.
##  Circular-Manhattan Plotting Fat percentage.
##  Plots are stored in: F:/demo script/基因组分析系列/CMplot

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第6张图片

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第7张图片

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第8张图片

3. Single_track Rectangular-Manhattan plot

1. Genome-wide association study(GWAS)

CMplot(pig60K, type = "p", plot.type = "m", LOG10 = TRUE, threshold = NULL, file = "jpg",
    memo = "", dpi = 300, file.output = TRUE, verbose = TRUE, width = 14, height = 6,
    chr.labels.angle = 45)
##  Rectangular-Manhattan Plotting trait1.
##  Rectangular-Manhattan Plotting trait2.
##  Rectangular-Manhattan Plotting trait3.
##  Plots are stored in: F:/demo script/基因组分析系列/CMplot

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第9张图片

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第10张图片

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第11张图片

2. Amplify signals on pch, cex and col

CMplot(pig60K, plot.type="m", col=c("grey30","grey60"), LOG10=TRUE, ylim=c(2,12), threshold=c(1e-6,1e-4),
        threshold.lty=c(1,2), threshold.lwd=c(1,1), threshold.col=c("black","grey"), amplify=TRUE,
        chr.den.col=NULL, signal.col=c("red","green"), signal.cex=c(1.5,1.5),signal.pch=c(19,19),
        file="jpg",memo="",dpi=300,file.output=TRUE,verbose=TRUE,width=14,height=6)

3. Attach chromosome density on the bottom of Manhattan plot

CMplot(pig60K, plot.type = "m", LOG10 = TRUE, ylim = NULL, threshold = c(1e-06, 1e-04),
    threshold.lty = c(1, 2), threshold.lwd = c(1, 1), threshold.col = c("black",
        "grey"), amplify = TRUE, bin.size = 1e+06, chr.den.col = c("darkgreen", "yellow",
        "red"), signal.col = c("red", "green"), signal.cex = c(1.5, 1.5), signal.pch = c(19,
        19), file = "jpg", memo = "", dpi = 300, file.output = TRUE, verbose = TRUE,
    width = 14, height = 6)
##  Rectangular-Manhattan Plotting trait1.
##  Rectangular-Manhattan Plotting trait2.
##  Rectangular-Manhattan Plotting trait3.
##  Plots are stored in: F:/demo script/基因组分析系列/CMplot

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第12张图片

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第13张图片

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第14张图片

4. Highlight a group of SNPs on pch, cex, type, and col

signal <- pig60K$Position[which.min(pig60K$trait2)]
SNPs <- pig60K$SNP[pig60K$Chromosome == 13 & pig60K$Position < (signal + 1e+06) &
    pig60K$Position > (signal - 1e+06)]
CMplot(pig60K, plot.type = "m", LOG10 = TRUE, col = c("grey30", "grey60"), highlight = SNPs,
    highlight.col = "green", highlight.cex = 1, highlight.pch = 19, file = "jpg",
    memo = "", chr.border = TRUE, dpi = 300, file.output = TRUE, verbose = TRUE,
    width = 14, height = 6)
##  Rectangular-Manhattan Plotting trait1.
##  Rectangular-Manhattan Plotting trait2.
##  Rectangular-Manhattan Plotting trait3.
##  Plots are stored in: F:/demo script/基因组分析系列/CMplo
SNPs <- pig60K[pig60K$trait2 < 1e-04, 1]
CMplot(pig60K, type = "h", plot.type = "m", LOG10 = TRUE, highlight = SNPs, highlight.type = "p",
    highlight.col = NULL, highlight.cex = 1.2, highlight.pch = 19, file = "jpg",
    memo = "", dpi = 300, file.output = TRUE, verbose = TRUE, width = 14, height = 6,
    band = 0.6)
##  Rectangular-Manhattan Plotting trait1.
##  Rectangular-Manhattan Plotting trait2.
##  Rectangular-Manhattan Plotting trait3.
##  Plots are stored in: F:/demo script/基因组分析系列/CMplot
SNPs <- pig60K[pig60K$trait2 < 1e-04, 1]
CMplot(pig60K, type = "p", plot.type = "m", LOG10 = TRUE, highlight = SNPs, highlight.type = "h",
    col = c("grey30", "grey60"), highlight.col = "darkgreen", highlight.cex = 1.2,
    highlight.pch = 19, file = "jpg", dpi = 300, file.output = TRUE, verbose = TRUE,
    width = 14, height = 6)
##  Rectangular-Manhattan Plotting trait1.
##  Rectangular-Manhattan Plotting trait2.
##  Rectangular-Manhattan Plotting trait3.
##  Plots are stored in: F:/demo script/基因组分析系列/CMplot

5. Visualize only one chromosome

CMplot(pig60K[pig60K$Chromosome == 13, ], plot.type = "m", LOG10 = TRUE, col = c("grey60"),
    highlight = SNPs, highlight.col = "green", highlight.cex = 1, highlight.pch = 19,
    file = "jpg", memo = "", threshold = c(1e-06, 1e-04), threshold.lty = c(1, 2),
    threshold.lwd = c(1, 2), width = 9, height = 6, threshold.col = c("red", "blue"),
    amplify = FALSE, dpi = 300, file.output = TRUE, verbose = TRUE)
##  Rectangular-Manhattan Plotting trait1.
##  Rectangular-Manhattan Plotting trait2.
##  Rectangular-Manhattan Plotting trait3.
##  Plots are stored in: F:/demo script/基因组分析系列/CMplot

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第15张图片

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第16张图片

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第17张图片

6. add genes or SNP names around the highlighted SNPs

SNPs <- pig60K[pig60K[, 5] < (0.05/nrow(pig60K)), 1]
genes <- paste("GENE", 1:length(SNPs), sep = "_")
set.seed(666666)
CMplot(pig60K[, c(1:3, 5)], plot.type = "m", LOG10 = TRUE, col = c("grey30", "grey60"),
    highlight = SNPs, highlight.col = c("red", "blue", "green"), highlight.cex = 1,
    highlight.pch = c(15:17), highlight.text = genes, highlight.text.col = c("red",
        "blue", "green"), threshold = 0.05/nrow(pig60K), threshold.lty = 2, amplify = FALSE,
    file = "jpg", memo = "", dpi = 300, file.output = TRUE, verbose = TRUE, width = 14,
    height = 6)
##  Rectangular-Manhattan Plotting trait2.
##  Plots are stored in: F:/demo script/基因组分析系列/CMplot

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第18张图片

7. Genomic Selection/Prediction(GS/GP) or other none p-values

CMplot(cattle50K, plot.type = "m", band = 0.5, LOG10 = FALSE, ylab = "SNP effect",
    threshold = 0.015, threshold.lty = 2, threshold.lwd = 1, threshold.col = "red",
    amplify = TRUE, width = 14, height = 6, signal.col = NULL, chr.den.col = NULL,
    file = "jpg", memo = "", dpi = 300, file.output = TRUE, verbose = TRUE, cex = 0.8)
##  Rectangular-Manhattan Plotting Somatic cell score.
##  Rectangular-Manhattan Plotting Milk yield.
##  Rectangular-Manhattan Plotting Fat percentage.
##  Plots are stored in: F:/demo script/基因组分析系列/CMplot

cattle50K[, 4:ncol(cattle50K)] <- apply(cattle50K[, 4:ncol(cattle50K)], 2, function(x) x *
    sample(c(1, -1), length(x), rep = TRUE))
CMplot(cattle50K, type = "h", plot.type = "m", band = 0.5, LOG10 = FALSE, ylab = "SNP effect",
    ylim = c(-0.02, 0.02), threshold.lty = 2, threshold.lwd = 1, threshold.col = "red",
    amplify = FALSE, cex = 0.6, chr.den.col = NULL, file = "jpg", memo = "", dpi = 300,
    file.output = TRUE, verbose = TRUE)
##  (warning: all phenotypes will use the same ylim.)
##  Rectangular-Manhattan Plotting Somatic cell score.
##  Rectangular-Manhattan Plotting Milk yield.
##  Rectangular-Manhattan Plotting Fat percentage.
##  Plots are stored in: F:/demo script/基因组分析系列/CMplot

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第19张图片

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第20张图片

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第21张图片

4. Multi_tracks Rectangular-Manhattan plot

SNPs <- list(pig60K$SNP[pig60K$trait1 < 1e-06], pig60K$SNP[pig60K$trait2 < 1e-06],
    pig60K$SNP[pig60K$trait3 < 1e-06])
CMplot(pig60K, plot.type = "m", multracks = TRUE, threshold = c(1e-06, 1e-04), threshold.lty = c(1,
    2), threshold.lwd = c(1, 1), threshold.col = c("black", "grey"), amplify = TRUE,
    bin.size = 1e+06, chr.den.col = c("darkgreen", "yellow", "red"), signal.col = c("red",
        "green", "blue"), signal.cex = 1, file = "jpg", memo = "", dpi = 300, file.output = TRUE,
    verbose = TRUE, highlight = SNPs, highlight.text = SNPs, highlight.text.cex = 1.4)
##  Multracks-Manhattan Plotting trait1.
##  Multracks-Manhattan Plotting trait2.
##  Multracks-Manhattan Plotting trait3.
##  Multraits-Rectangular Plotting...(finished 13%)
##  Plots are stored in: F:/demo script/基因组分析系列/CMplot

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第22张图片

5. Q-Q plot

1. Single_track Q-Q plot

CMplot(pig60K, plot.type = "q", box = FALSE, file = "jpg", memo = "", dpi = 300,
    conf.int = TRUE, conf.int.col = NULL, threshold.col = "red", threshold.lty = 2,
    file.output = TRUE, verbose = TRUE, width = 5, height = 5)
##  QQ Plotting trait1.
##  QQ Plotting trait2.
##  QQ Plotting trait3.
##  Plots are stored in: F:/demo script/基因组分析系列/CMplot

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第23张图片

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第24张图片

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第25张图片

2. Multi_tracks Q-Q plot

pig60K$trait1[sample(1:nrow(pig60K), round(nrow(pig60K) * 0.8))] <- NA
pig60K$trait2[sample(1:nrow(pig60K), round(nrow(pig60K) * 0.25))] <- NA
CMplot(pig60K, plot.type = "q", col = c("dodgerblue1", "olivedrab3", "darkgoldenrod1"),
    threshold = 1e-06, ylab.pos = 2, signal.pch = c(19, 6, 4), signal.cex = 1.2,
    signal.col = "red", conf.int = TRUE, box = FALSE, multracks = TRUE, cex.axis = 2,
    file = "jpg", memo = "", dpi = 300, file.output = TRUE, verbose = TRUE, ylim = c(0,
        8), width = 5, height = 5)
##  (warning: all phenotypes will use the same ylim.)
##  Multracks-QQ Plotting trait1.
##  Multracks-QQ Plotting trait2.
##  Multracks-QQ Plotting trait3.
##  Multraits-QQ Plotting trait1.
##  Multraits-QQ Plotting trait2.
##  Multraits-QQ Plotting trait3.
##  Plots are stored in: F:/demo script/基因组分析系列/CMplot

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第26张图片

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第27张图片

References:

Yin, L. et al. rMVP: A Memory-efficient, Visualization-enhanced, and Parallel-accelerated tool for Genome-Wide Association Study, Genomics, Proteomics & Bioinformatics (2021), doi: 10.1016/j.gpb.2020.10.007.

DNA 12. SCI 文章绘图之全基因组关联分析可视化(GWAS)_第28张图片

ItVuer - 免责声明 - 关于我们 - 联系我们

本网站信息来源于互联网,如有侵权请联系:561261067@qq.com

桂ICP备16001015号