R ggplot：如何在不考虑变量数量的情况下绘制热图

使用下面的

data.frame

数据

df <- read.table(text = c("
NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA
                          0.4748    NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA
                          0.905 0.5362  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA
                          0.0754    0.0118  0.0614  NA  NA  NA  NA  NA  NA  NA  NA  NA
                          0.8768    0.3958  0.7952  0.1034  NA  NA  NA  NA  NA  NA  NA  NA
                          0.5708    0.2056  0.4984  0.2356  0.6736  NA  NA  NA  NA  NA  NA  NA
                          0.2248    0.6204  0.268   0.0014  0.183   0.0768  NA  NA  NA  NA  NA  NA
                          0.483 0.9824  0.5314  0.0114  0.3906  0.1968  0.6308  NA  NA  NA  NA  NA
                          0.697 0.732   0.7604  0.0264  0.594   0.3334  0.416   0.7388  NA  NA  NA  NA
                          0.2918    0.7286  0.3382  0.003   0.2386  0.1122  0.8712  0.7266  0.509   NA  NA  NA
                          0.5904    0.8352  0.6704  0.0188  0.4966  0.273   0.5192  0.8328  0.8736  0.5914  NA  NA
                          0.3838    0.8768  0.4476  0.0042  0.3148  0.1498  0.7288  0.873   0.6178  0.8276  0.7432  NA
                          "), header = F)

colnames(df) <- c( "TK1",   "TK2",  "TK3",  "TK4"   , "TK5",    "TK6",  "TK7",  "TK8",  "TK9",  "TK10", "TK11", "TK12")
rownames(df) <- c( "TK1",   "TK2",  "TK3",  "TK4"   , "TK5",    "TK6",  "TK7",  "TK8",  "TK9",  "TK10", "TK11", "TK12")

df
#        TK1    TK2    TK3    TK4    TK5    TK6    TK7    TK8    TK9   TK10   TK11 TK12
#TK1      NA     NA     NA     NA     NA     NA     NA     NA     NA     NA     NA   NA
#TK2  0.4748     NA     NA     NA     NA     NA     NA     NA     NA     NA     NA   NA
#TK3  0.9050 0.5362     NA     NA     NA     NA     NA     NA     NA     NA     NA   NA
#TK4  0.0754 0.0118 0.0614     NA     NA     NA     NA     NA     NA     NA     NA   NA
#TK5  0.8768 0.3958 0.7952 0.1034     NA     NA     NA     NA     NA     NA     NA   NA
#TK6  0.5708 0.2056 0.4984 0.2356 0.6736     NA     NA     NA     NA     NA     NA   NA
#TK7  0.2248 0.6204 0.2680 0.0014 0.1830 0.0768     NA     NA     NA     NA     NA   NA
#TK8  0.4830 0.9824 0.5314 0.0114 0.3906 0.1968 0.6308     NA     NA     NA     NA   NA
#TK9  0.6970 0.7320 0.7604 0.0264 0.5940 0.3334 0.4160 0.7388     NA     NA     NA   NA
#TK10 0.2918 0.7286 0.3382 0.0030 0.2386 0.1122 0.8712 0.7266 0.5090     NA     NA   NA
#TK11 0.5904 0.8352 0.6704 0.0188 0.4966 0.2730 0.5192 0.8328 0.8736 0.5914     NA   NA
#TK12 0.3838 0.8768 0.4476 0.0042 0.3148 0.1498 0.7288 0.8730 0.6178 0.8276 0.7432   NA

结果

问题

我想让热图的绘制成为动态的。那么，不管变量和观测值的数量如何，都可以绘制热图，而无需更改不同变量数量的代码

有什么办法吗

library(ggplot2)
library(tidyr)
library(dplyr)

无论有多少列和行，此代码块都可以工作

df <-
  df %>%
  mutate(Relationship = rownames(.)) %>% #Replaces trial$Relationship <- rownames(df) 
  select(Relationship, everything()) %>% #Replaces trial1 <- subset(trial, select = c(13, 1, 2, 3,4,5,6,7,8,9,10,11,12))
  gather('Relation1', 'Strength', -1) %>% #Replaces df2 <- gather(trial1, "Relation1", "Strength", 2:13)
  mutate(Strength = round(Strength, digits = 2))

df%
mutate（Relationship=rownames（.））%>%#替换试用版$Relationship%#替换trial1%#替换df2%
不同的（）%>%
提取（关系，放入=c（“TK”，“num”），“（\\D+）（\\D+）”，
删除=假，转换=真）%>%
排列（数量）%>%
选择（关系）
df%
变异（关系=因子（关系，级别=因子级别$关系），
关系1=系数（关系1，级别=修订版（系数级别$关系）））

修改的绘图代码

ggheatmap <- ggplot(df, aes(Relationship, Relation1,  fill = Strength))+
  geom_tile(color = "white")+
  scale_fill_gradient2(low = "red", high = "green", mid = "lightgreen", 
                       midpoint = 0.5, limit = c(0,1), space = "Lab", 
                       name="Correlation") + theme_minimal()

ggheatmap + 
  geom_text(aes(Relationship, Relation1, label = Strength), color = "black", size = 4) +
  labs(x = expression(""), 
       y=expression(""))

---

ggheatmap在这种情况下，我觉得你的方法很迂回（我提到过）。这段代码只需要使用colnames（df）和rownames（df）

库（2）；图书馆（GG2）
df2@MirHenglin因为您正在对
factorLevels
中的因子级别进行排序，在调用
factor
时也设置
ordered=TRUE
对您来说有意义吗？@steveb我认为如果您使用变量进行建模，
TK
变量使得
TK3
大于
TK2
大于
TK1
，则使用
ordered=TRUE
更重要，因为在建模时，有序因子和无序因子的处理方式可能非常不同。排列因子的级别只是告诉函数获取变量的顺序；它不会改变因子的值。
# Order Relatinoship variables by numeric suffix
# Since its a square matrix you only have to do it once for both columns

factorLevels <-
  df %>%
  select(Relationship) %>%
  distinct() %>%
  extract(Relationship, into = c("TK", "num"), "(\\D+)(\\d+)",
          remove = FALSE, convert=TRUE) %>%
  arrange(num) %>%
  select(Relationship)

df <-
  df %>%
  mutate(Relationship = factor(Relationship, levels = factorLevels$Relationship),
         Relation1 = factor(Relation1, levels = rev(factorLevels$Relationship)))

ggheatmap <- ggplot(df, aes(Relationship, Relation1,  fill = Strength))+
  geom_tile(color = "white")+
  scale_fill_gradient2(low = "red", high = "green", mid = "lightgreen", 
                       midpoint = 0.5, limit = c(0,1), space = "Lab", 
                       name="Correlation") + theme_minimal()

ggheatmap + 
  geom_text(aes(Relationship, Relation1, label = Strength), color = "black", size = 4) +
  labs(x = expression(""), 
       y=expression(""))

library(reshape2); library(ggplot2)

df2 <- melt(as.matrix(df), id.var = names(df)[1])    # as.matrix() fixes colnames of long df.
df2$Var2 <- with(df2, factor(Var2, levels=rev(levels(Var2))))

ggheatmap <- ggplot(df2, aes(Var1, Var2, fill=value)) +
  geom_tile(color = "white")+
  scale_fill_gradient2(low = "red", high = "green", mid = "lightgreen", 
                       midpoint = 0.5, limit = c(0,1), space = "Lab", 
                       name="Correlation") + theme_minimal()

ggheatmap + 
  geom_text(aes(label = round(value, 2)), color = "black", size = 4) + 
  labs(x = expression(""), y=expression(""))