$ head 41588_2015_BFng3406_MOESM37_ESM.csv
Trait1,Trait2,rg,se,z,p
ADHD,Age at Menarche,-0.153,0.08218,-1.858,0.063
ADHD,Age at Smoking,-0.036,0.2427,-0.147,0.883
ADHD,Alzheimer's,-0.055,0.2191,-0.249,0.803
ADHD,Anorexia,0.192,0.1162,1.649,0.099
ADHD,Autism Spectrum,-0.164,0.1438,-1.144,0.253
ADHD,BMI,0.287,0.08913,3.222,0.001
ADHD,BMI 2010,0.258,0.08606,2.993,0.003
ADHD,Bipolar,0.265,0.1539,1.722,0.085
ADHD,Birth Length,-0.055,0.1679,-0.329,0.742

install.packages("corrplot")
library(corrplot)
#读取数据
ldsc = read.csv(file = '41588_2015_BFng3406_MOESM37_ESM.csv',header=TRUE)

#提取需要画的表型
to_plot_col<-c('Age at Menarche','Alzheimer\'s','Anorexia','Autism Spectrum','BMI','Bipolar','Birth Length','Birth Weight','Childhood Obesity','Coronary Artery Disease','Crohn\'s Disease','Depression','Ever/Never Smoked','Fasting Glucose','HDL','Height','Infant Head Circumference','LDL','Rheumatoid Arthritis','Schizophrenia','T2D','Triglycerides','Ulcerative Colitis','Years of Education')

#构建一个rg矩阵，和一个p值矩阵,并命名row和col
gcm_rg<-matrix(1, nrow = length(to_plot_col), ncol = length(to_plot_col))
gcm_p<-matrix(1, nrow = length(to_plot_col), ncol = length(to_plot_col))
rownames(gcm_rg) <- to_plot_col
colnames(gcm_rg) <- to_plot_col
rownames(gcm_p) <-to_plot_col
colnames(gcm_p) <- to_plot_col

#将数据填进矩阵
for (row in 1:nrow(ldsc)) {
    if(ldsc[row, "Trait1"]%in%to_plot_col && ldsc[row, "Trait2"]%in%to_plot_col){
    Trait1 <- ldsc[row, "Trait1"]
    Trait2  <- ldsc[row, "Trait2"]
    gcm_rg[Trait1,Trait2]<-ldsc[row, "rg"]
    gcm_p[Trait1,Trait2]<-p<-ldsc[row, "p"]
    gcm_rg[Trait2,Trait1]<-ldsc[row, "rg"]
    gcm_p[Trait2,Trait1]<-p<-ldsc[row, "p"]
    }
    }

#使用corrplot绘图：（方法选择square）
#sig.level = 0.05/300 为显著水平（这里是多重检验调整的）
#gcm_rg为相关系数矩阵
#gcm_p为p值矩阵
#order="hclust" 为通过分层聚类对矩阵进行排序
#其他选项均为细节调整 可以参考 https://cran.r-project.org/web/packages/corrplot/vignettes/corrplot-intro.html 

corrplot(gcm_rg, p.mat=gcm_p, method="square",order="hclust",sig = "pch",sig.level = 0.05/300 ,pch="*",pch.cex=2,
         tl.col="black",tl.srt=45,addgrid.col="grey90")

install.packages("dplyr")
devtools::install_github("mkanai/corrplot")

#读取数据
ldsc = read.csv(file = '41588_2015_BFng3406_MOESM37_ESM.csv',header=TRUE)

#计算FDR调整后p值
ldsc$fdr<-p.adjust(ldsc$p, method ="fdr")

to_plot_col<-c('Age at Menarche','Alzheimer\'s','Anorexia','Autism Spectrum','BMI','Bipolar','Birth Length','Birth Weight','Childhood Obesity','Coronary Artery Disease','Crohn\'s Disease','Depression','Ever/Never Smoked','Fasting Glucose','HDL','Height','Infant Head Circumference','LDL','Rheumatoid Arthritis','Schizophrenia','T2D','Triglycerides','Ulcerative Colitis','Years of Education')

gcm_rg<-matrix(NA, nrow = length(to_plot_col), ncol = length(to_plot_col))
gcm_p<-matrix(NA, nrow = length(to_plot_col), ncol = length(to_plot_col))
rownames(gcm_rg) <- to_plot_col
colnames(gcm_rg) <- to_plot_col
rownames(gcm_p) <-to_plot_col
colnames(gcm_p) <- to_plot_col

#p值矩阵改为fdr矩阵
for (row in 1:nrow(ldsc)) {
    if(ldsc[row, "Trait1"]%in%to_plot_col && ldsc[row, "Trait2"]%in%to_plot_col){
    Trait1 <- ldsc[row, "Trait1"]
    Trait2  <- ldsc[row, "Trait2"]
    gcm_rg[Trait1,Trait2]<-ldsc[row, "rg"]
    gcm_p[Trait1,Trait2]<-ldsc[row, "fdr"]
    gcm_rg[Trait2,Trait1]<-ldsc[row, "rg"]
    gcm_p[Trait2,Trait1]<-ldsc[row, "fdr"]
    }
    }

#事先对矩阵进行排序（此修改的版本直接使用order="hclust"时有bug）
gcm_rg.hclust<-corrMatOrder(gcm_rg,order ="hclust")
gcm_rg_hclust<-gcm_rg[gcm_rg.hclust,gcm_rg.hclust]
gcm_p_hclust<-gcm_p[gcm_rg.hclust,gcm_rg.hclust]

options(repr.plot.width=15, repr.plot.height=15)
#使用修改后的corrplot绘图：（注意！！方法选择 psquare）
corrplot(gcm_rg_hclust ,p.mat=gcm_p_hclust, diag=FALSE,method="psquare",sig="pch",insig="label_sig",sig.level = 0.01 ,pch.cex=2,tl.col="black",tl.srt=45,addgrid.col="grey90")