Question

我正在尝试自动标记manhattan plot上的某些数据点。

对于那些不知道manhattan plot是什么的人来说，这无关紧要。

重写ggplot2中的旧式R代码对我来说似乎更具挑战性。

由于我想要添加到脚本的唯一功能是能够很好地自动标记一些数据点（使用ggrepel），所以我想也许我可以覆盖绘制的图表上的标签旧代码。

我目前的尝试如下：

library(ggplot2);
library(ggrepel);

d=read.table("a.txt",header=T,fill=T, sep=" ");
dmht='';

dmht<-data.frame(chrom=d[,9], txStart = d[,11], "-log10(PValue)" = -log10(d[,5]))
# sort it
o<-order(dmht[,1],dmht[,2]);
dmht<-data.frame(dmht[o,]);
names(dmht)<-c("chrom", "txStart", "-log10(PValue)");
attach(dmht);
chrs<-c('chr1','chr2','chr3','chr4','chr5','chr6','chr7','chr8','chr9','chr10','chr11','chr12','chr13','chr14','chr15','chr16','chr17','chr18','chr19','chrX','chrY');
chrLabels = c('1', '2', '3', '4', '5', '6', '7', '8', '9', '10', '11', '12', '13', '14', '15', '16', '17', '18', '19', 'X', 'Y');
chrLen<-rep(0,length(chrs));
chrMin<-rep(0,length(chrs));
chrMax<-rep(0,length(chrs));
totalLen = 0;

for(i in 1:length(chrs)) {
    dchr<-subset(dmht, chrom == chrs[i]);
    chrMin[i]<-min(as.numeric(dchr[,2]),na.rm=T);
    chrMax[i]<-max(as.numeric(dchr[,2]),na.rm=T);
    chrLen[i]<-chrMax[i] - chrMin[i] + 1;
    totalLen = totalLen + chrMax[i];
}


ds=read.table("selected2label.txt",header=T,fill=T,sep="    ");
dsel='';

dsel<-data.frame(ds[,9], ds[,11], -log10(ds[,5]) , ds[,1])
ymin = min(as.numeric(dmht[,3], dsel[,3]), na.rm=T);
ymax = max(as.numeric(dmht[,3], dsel[,3]), na.rm=T);
# chr start positions
chrStart<-rep(0,length(chrs));

for(i in 1:length(chrs)) {
    if(i == 1) {
    chrStart[i] = 1;
    }
    else {
    chrStart[i] = chrStart[i-1] + chrMax[i-1] + 1;
    }
}

#dmht=subset(dmht,dmht[,1] < 24); # remove mitochondra snps
colors <- rep(c("blue", "green", "cyan"),9);

png("result/test/mhtTest.png" , width=1600);
par(las=3, lab=c(length(chrLabels),5,7))

# draw the dots
dchr<-subset(dmht, chrom == chrs[1]);
plot(as.numeric(dchr[,2])+chrStart[1], as.numeric(dchr[,3]), col=colors[1], ylim=c(ymin,ymax),xlim=c(chrMin[1],totalLen),axes=F,ylab="-log10(PValue)", xlab="Chromosome", main="");
for(i in 2:length(chrs)) {
    dchr<-subset(dmht, chrom == chrs[i]);
    points(as.numeric(dchr[,2])+chrStart[i], as.numeric(dchr[,3]), col=colors[i]);
}

axis(side=1,labels=chrLabels,at=chrStart);
axis(side=2);

# draw the quantiles
quants<-quantile(as.numeric(dmht[,3]), p=c(),na.rm=T);

for(q in quants) {
    abline(h=q);
}

# draw the abs values
abss<-c();
for(a in abss) {
    abline(h=a);
}

# sort it
#os<-order(dsel[,1],dsel[,2]);
#dsel<-data.frame(dsel[os,]);
#dsel
#dsel<-data.frame(dsel);
#dsel
colnames(dsel)<-c("chrom", "txStart", "-log10(PValue)" , 'GENE_ID');
detach(dmht);
attach(dsel);

# highlight the selected dots
for(i in 1:length(chrs)) {
    dchr<-subset(dsel, chrom == chrs[i] | paste("chr", chrom, sep="") == chrs[i]);
    if(length(dchr[,2]) > 0) {
    print(dchr)
    # this is the new code
    geom_label_repel(data = dchr, aes(label = GENE_ID, x=as.numeric(txStart)+chrStart[i], y = as.numeric(dchr[,3])), size = 5, box.padding = unit(0.35, "lines"), point.padding = unit(0.5, "lines")) 
    # replacing the line in the old script:
    # text(as.numeric(dchr[,2])+chrStart[i], as.numeric(dchr[,3]), dchr[,4])
    }
}


dev.off();

我替换的唯一一行（除了添加libary(ggplot2)＆amp; library(ggrepel)）是：

    # this is the new code
    geom_label_repel(data = dchr, aes(label = GENE_ID, x=as.numeric(txStart)+chrStart[i], y = as.numeric(dchr[,3])), size = 5, box.padding = unit(0.35, "lines"), point.padding = unit(0.5, "lines")) 
    # replacing the line in the old script:
    # text(as.numeric(dchr[,2])+chrStart[i], as.numeric(dchr[,3]), dchr[,4])

问题在于标签根本没有出现。

非常感谢您的帮助。

Answer 1

如果您正在尝试使用带有基础R图的ggrepel，那么简短的答案是您不能。包ggplot2（ggrepel是一个附加组件）基于网格图形系统（格子绘图包）。不可能将ggplot的元素与基本R图（你描述为旧式代码）混合和匹配。

Answer 2

此代码可能有效，也可能无效，我只是简单地描绘了你的情节代码。

ggplot(aes(x=as.numeric(dchr[,2]) + chrStart[1], y=as.numeric(dchr[,3]), color = dchr[, 2]) + 
    geom_point() +
    ylim(c(ymin,ymax)) +
    xlim(c(chrMin[1],totalLen)) +
    ylab("-log10(PValue)") +
    xlab("Chromosome")+
    ggtitle("") + 
        geom_label_repel(data = dchr, inherit.aes=FALSE,
        aes(label = GENE_ID, x=as.numeric(txStart)+chrStart[i], y = as.numeric(dchr[,3])), size = 5, box.padding = unit(0.35, "lines"), point.padding = unit(0.5, "lines"))

将R ggplot2与旧式绘图混合

2 个答案: