初始数据类似如下：
image.png
填充下缺失值
1 2 3 4 5 | data[data==0] <- NA data[is.na(data)] <- min(data,na.rm = T)*0.01 pheatmap(log10(data)) pheatmap(data,scale = "row") |
直接取log绘制不报错，但做
1 2 | Error in hclust(d, method = method) : NA/NaN/Inf in foreign function call (arg 11) |
原因是数据中有标准差sd为0的行，做归一化时报错。pheatmap中做归一化采用的是scale方法：
image.png
因此如果要做scale，需要将这样的行去掉：
1 2 | test <- data[apply(data, 1, function(x) sd(x)!=0),] pheatmap(test,scale = "row") |
Ref:
https://stackoverflow.com/questions/30350438/pheatmap-scale-row-giving-error-in-hclustd-method-method-na-nan-inf-in
https://zhuanlan.zhihu.com/p/30518877