技术：ggpubr提高作图效率

ggpubr提高作图效率。这个包R的作图省略了好多步骤，类似于爬虫的postman。

实际上，我们主要做统计分析，画太多时间画图，很不情愿。

library(ggpubr)

## Error in `library()`:
## ! there is no package called 'ggpubr'

library(tidyverse)

## Error in `library()`:
## ! there is no package called 'tidyverse'

图+均线

mtcars %>% 
  gather(key,value,everything()) %>% 
  group_by(key) %>% 
  summarise(n_distinct(value)) %>% 
  arrange(desc(`n_distinct(value)`))

## Error in `` mtcars %>% gather(key, value, everything()) %>% group_by(key) %>% summarise(
##     n_distinct(value)) %>% arrange(desc(`n_distinct(value)`)) ``:
## ! could not find function "%>%"

mtcars1 <- 
  mtcars %>% 
  mutate(am = as.factor(am))

## Error in `mtcars %>% mutate(am = as.factor(am))`:
## ! could not find function "%>%"

mtcars1 %>% 
  ggdensity(x = "qsec", # 注意变量str化
            col = "am",
            add = "mean", # 加均线
            rug = TRUE, # rug | logical value. If TRUE, add marginal rug.
            fill = "am", # 这是做频率图我经常忘了，为了好看。
            palette = c("#00AFBB", "#E7B800")) # 这个我真的不记得。

## Error in `mtcars1 %>% ggdensity(x = "qsec", col = "am", add = "mean", rug = TRUE, fill = "am",
##     palette = c("#00AFBB", "#E7B800"))`:
## ! could not find function "%>%"

mtcars1 %>% 
  gghistogram(x = "qsec", # 注意变量str化
            col = "am",
            add = "mean", # 加均线
            rug = TRUE, # rug | logical value. If TRUE, add marginal rug.
            fill = "am", # 这是做频率图我经常忘了，为了好看。
            palette = c("#00AFBB", "#E7B800")) # 这个我真的不记得。

## Error in `mtcars1 %>% gghistogram(x = "qsec", col = "am", add = "mean", rug = TRUE, fill = "am",
##     palette = c("#00AFBB", "#E7B800"))`:
## ! could not find function "%>%"

注意变量str化，因此要因子化提前mutate，这个图唯一好的，就是可以加均线而已， rug就是图下的竖线，不明白。 palette = c("#00AFBB", "#E7B800"))两种对比，推荐的配色。 ggdensity和gghistogram取决于离散程度。

ECDF

mtcars1 %>% 
ggplot(aes(qsec)) +
stat_ecdf(aes(color = am,linetype = am), 
              geom = "step", size = 1.5) +
  scale_color_manual(values = c("#00AFBB", "#E7B800"))+
  labs(y = "f(weight)")

## Error in `mtcars1 %>% ggplot(aes(qsec))`:
## ! could not find function "%>%"

不好看。

箱线图、小提琴图和p值

在前面的eda解释中，箱线图主要用于一个分类和一个连续变量的分析。

mtcars2 <- 
mtcars1 %>% 
  mutate(cyl = as.factor(cyl))

## Error in `mtcars1 %>% mutate(cyl = as.factor(cyl))`:
## ! could not find function "%>%"

mtcars2 %>% 
  ggboxplot(
    x = "cyl",
    y = "qsec",
    col = "cyl",
    palette = c("#00AFBB", "#E7B800", "#FC4E07"),
    add = "jitter") +
  stat_compare_means(
    comparisons = 
    list(
      c("4", "6"), 
      c("4", "8"),
      c("6", "8")
    ))

## Error in `mtcars2 %>% ggboxplot(x = "cyl", y = "qsec", col = "cyl", palette = c("#00AFBB",
##     "#E7B800", "#FC4E07"), add = "jitter")`:
## ! could not find function "%>%"

经过之前重复抽样的理解，这里对样本间均值差异的p值，应该有很深的理解了，因此，这里还可以比较，这样的话，直接给p值，这样不会让人看图很confuse。 stat_compare_means完成这一目标，直接放入对比pairs就好。 palette = c("#00AFBB", "#E7B800", "#FC4E07")三个比较的配色。

mtcars2 %>% 
    ggviolin(
    x = "cyl",
    y = "qsec",
    fill = "cyl",
    palette = c("#00AFBB", "#E7B800", "#FC4E07"),
    add = "boxplot",
    add.params = list(fill="white")) +
  stat_compare_means(
    comparisons = 
    list(
      c("4", "6"), 
      c("4", "8"),
      c("6", "8")
    ),
    label = "p.signif",
    label.y = 50
    )

## Error in `mtcars2 %>% ggviolin(x = "cyl", y = "qsec", fill = "cyl", palette = c("#00AFBB",
##     "#E7B800", "#FC4E07"), add = "boxplot", add.params = list(fill = "white"))`:
## ! could not find function "%>%"

这里修改add，换成内置箱形图。 add.params改变箱形图的颜色。 label = "p.signif",我也觉得数字很烦，这个比较简洁。其中ns: p > 0.05。

ns: p > 0.05
*: p <= 0.05
**: p <= 0.01
***: p <= 0.001
****: p <= 0.0001

label.y = 50定位p值图中显示位置。

条形图+不分组排序

这个我一直在ggplot里面没有实现好。可以借鉴一下。

for (i in c(TRUE,FALSE)){
print(
mtcars %>% 
  mutate(cyl = as.factor(cyl),
         ) %>% 
  rownames_to_column(var = "name") %>% 
  ggbarplot(x="name", 
            y="mpg", 
            fill = "cyl", 
            color = "white",
            palette = "jco",#杂志jco的配色
            sort.val = "desc",#下降排序
            sort.by.groups=i,#不按组排序
            x.text.angle=60)
)
}

## Error in `mtcars %>% mutate(cyl = as.factor(cyl), ) %>% rownames_to_column(var = "name") %>%
##     ggbarplot(x = "name", y = "mpg", fill = "cyl", color = "white", palette = "jco",
##       sort.val = "desc", sort.by.groups = i, x.text.angle = 60)`:
## ! could not find function "%>%"

参考ggplot2中实现for循环。

我之前一直没有很容易的放在一张图比较。

x.text.angle=60这个更加简单了。 theme(axis.text.x = element_text(angle = 70, hjust = 1)) 比较复杂。

偏差图

这些美化，都是我需要提前做好的。

mtcars %>% 
  rownames_to_column(var = "name") %>% 
  mutate(cyl = as.factor(cyl),
         mpg_z = (mpg-mean(mpg))/sd(mpg),
         mpg_grp = case_when(
           mpg_z<0 ~ "low",
           TRUE ~ "high"
         ),
         mpg_grp = as.factor(mpg_grp),
         mpg_grp = fct_relevel(mpg_grp,"low", "high")
         ) %>% 
  ggbarplot(
    x="name", 
    y="mpg_z", 
    fill = "mpg_grp",
    color = "white",
    palette = "jco",
    sort.val = "asc",
    sort.by.groups = FALSE,
    x.text.angle=60,
    ylab = "MPG z-score",
    xlab = FALSE,
    legend.title="MPG Group"
    )

## Error in `mtcars %>% rownames_to_column(var = "name") %>% mutate(cyl = as.factor(cyl),
##   mpg_z = (mpg - mean(mpg)) / sd(mpg), mpg_grp = case_when(mpg_z < 0 ~ "low",
##   TRUE ~ "high"), mpg_grp = as.factor(mpg_grp), mpg_grp = fct_relevel(mpg_grp,
##     "low", "high")) %>% ggbarplot(x = "name", y = "mpg_z", fill = "mpg_grp",
##     color = "white", palette = "jco", sort.val = "asc", sort.by.groups = FALSE,
##     x.text.angle = 60, ylab = "MPG z-score", xlab = FALSE, legend.title = "MPG Group")`:
## ! could not find function "%>%"

这里其实使用了标准化处理$\frac{x-\mu}{\sigma}$

mpg_z = (mpg-mean(mpg))/sd(mpg),
mpg_grp = case_when(
 mpg_z<0 ~ "low",
 TRUE ~ "high"
),
mpg_z = fct_relevel(mpg_z,"low", "high")

棒棒糖图(Lollipop chart)

这个之前当初我搞了很久了。终于出了简单的方法了。

for (i in c(TRUE,FALSE)) {
mtcars %>% 
  rownames_to_column(var = "name") %>% 
  mutate(cyl = as.factor(cyl)) %>% 
  ggdotchart(
    x="name", 
    y="mpg", 
    color = "cyl", 
    palette = c("#00AFBB", "#E7B800", "#FC4E07"), 
    sorting = "ascending", 
    add = "segments", 
    rotate=i,
    ggtheme = theme_minimal()
    )
}

## Error in `mtcars %>% rownames_to_column(var = "name") %>% mutate(cyl = as.factor(cyl)) %>%
##     ggdotchart(x = "name", y = "mpg", color = "cyl", palette = c("#00AFBB",
##       "#E7B800", "#FC4E07"), sorting = "ascending", add = "segments", rotate = i,
##     ggtheme = theme_minimal())`:
## ! could not find function "%>%"

rotate决定是否旋转，比ggplot简单多了，coord_flip。

mtcars %>% 
  rownames_to_column(var = "name") %>% 
  mutate(cyl = as.factor(cyl)) %>% 
  ggdotchart(
    x="name", 
    y="mpg", 
    color = "cyl", 
    palette = c("#00AFBB", "#E7B800", "#FC4E07"), 
    sorting = "ascending", 
    add = "segments", 
    rotate=TRUE,
    group = "cyl", 
    dot.size = 6, 
    label = round(mtcars$mpg),
    font.label = list(color="white", size=9, vjust=0.5),
    
    ggtheme = theme_minimal()
  )

## Error in `mtcars %>% rownames_to_column(var = "name") %>% mutate(cyl = as.factor(cyl)) %>%
##     ggdotchart(x = "name", y = "mpg", color = "cyl", palette = c("#00AFBB",
##       "#E7B800", "#FC4E07"), sorting = "ascending", add = "segments", rotate = TRUE,
##     group = "cyl", dot.size = 6, label = round(mtcars$mpg), font.label = list(
##       color = "white", size = 9, vjust = 0.5), ggtheme = theme_minimal())`:
## ! could not find function "%>%"

label = round(mtcars$mpg),这是目前这个包的一个bug，必须这么做，但是呢，已经非常出色了。

技术：ggpubr提高作图效率

技术系列导航

图+均线

ECDF

箱线图、小提琴图和p值

条形图+不分组排序

偏差图

棒棒糖图(Lollipop chart)

技术系列导航

技术：ggpubr提高作图效率

技术 系列导航

图+均线

ECDF

箱线图、小提琴图和p值

条形图+不分组排序

偏差图

棒棒糖图(Lollipop chart)

技术 系列导航

技术系列导航

技术系列导航