如何使用与相对丰度图相对应的颜色标签制作复合图例？

Question

我想从下面的paper 复制一个图。

它在分离 X1 列时卡住了。我想使用正则表达式，但不知道如何使用。 我计划用下划线分隔符分隔每个单词（我有一个列表），然后将 [-tes & -ria] 和 [-ceae] 后缀的单词分别提取到 Phylum 和 Family 中。在那之后，家庭之后的词应该被收集到属中。为了准确起见，可能会将“未分类”和少于5个字符的单词分组到前面的单词中。

另外，是否可以为每个家庭添加对应于相对丰度图的小颜色标签？

library(tidyverse)
james <- read_csv("tableS2a.csv")
james <- james %>% mutate(
    Cecum = rowSums(select(james, contains("Caecum"))),
    Transverse = rowSums(select(james, contains("Transv"))),
    Sigmoid = rowSums(select(james, contains("Sigmoi")))
  )
james2 <- james %>% 
  select(X1, Cecum, Transverse, Sigmoid) 

james.tab <- james2 %>%
  mutate(meanAbundance = 
           rowMeans(
             column_to_rownames(james2, var = "X1")
             )
         ) %>%
  arrange(desc(meanAbundance)) %>%
  top_n(30, meanAbundance) # extract top30

write.csv2(james.tab, "jamestab.csv")

james.tab2 <- 
  as.data.frame(
    apply(
      select(
        james.tab, 
        Cecum, 
        Transverse, 
        Sigmoid), 2, 
      function(x) x / sum(x) * 100)
    )

james.tab3 <-
  bind_cols(
    as.data.frame(
      select(james.tab, X1)), 
    as.data.frame(james.tab2)
    )

james.X1 <- select(james.tab3, X1)

# Separate X1 to Phylum(-tes/-ria), Family (-ceae), and genus
james.list <- strsplit(pull(james.X1, X1), "_")
james.class <-
  if_else(grepl("(ceae)", james.X1) == T,
          mutate(james.X1, Family =
                   grep(
                     "[[:alpha:]]ceae(_)", 
                     strsplit(pull(james.X1, X1), "_"), 
                     value = T
                   )))

我是 R 新手，上面的代码大部分是从我以前的工作中粘贴的。如果代码效率低下，请原谅我。数据集：Original table -> Top30 csv (pastebin)

追加

这是最近的结果 我没有成功实现ggtext包，可能是主题地址错误？

library(tidyverse)
library(patchwork)
library(ggtext)
library(glue)

james <- read_csv("tableS2a.csv")
james2 <- james %>% 
  mutate(
  Cecum = rowSums(select(james, contains("Caecum"))),
  Transverse = rowSums(select(james, contains("Transv"))),
  Sigmoid = rowSums(select(james, contains("Sigmoi")))
  ) %>% 
  select(X1, Cecum, Transverse, Sigmoid) %>% 
  filter(grepl("(ceae)", james$X1)) # Filter rows with -ceae suffix only

# extract family value with selecting -ceae/les suffix word
family.naming0 <-
  regmatches(james2$X1,
             regexpr("(?<=_)(.*?(ceae|les)(?=_))", james2$X1, perl = T))
#in between "_" to fail-safe double -ceae. E.g. Bacteria_Bacteriaceae_Aceae

family.naming1 <-
  regmatches(james2$X1, regexpr("(?<=ceae_|les_)\\d", james2$X1, perl = T))

family.naming2 <- 
  regmatches(james2$X1, regexpr("(?<=ceae_|les_)unclassified", james2$X1, perl = T))

family.naming3 <-
  ifelse(
    grepl("(?<=[(ceae_)|(les_)])\\d", james2$X1, perl = T),
    paste0(family.naming0, " ", family.naming1),
    ifelse(
      grepl("(?<=[(ceae_)|(les_)])unclassified", james2$X1, perl = T),
      paste0(family.naming0, " ", family.naming2),
      paste0(family.naming0)
    ))  

james3 <- james2 %>% 
  gather("Cecum", "Transverse", "Sigmoid", key = "location", value = "abundance") %>% 
  mutate(relativeAbundance=abundance/sum(abundance) * 100) %>%
  mutate(phylum=gsub("(_.*)","", X1)) %>% # extract phylum value with selecting first word
  mutate(family=
           ifelse(
             grepl("(?<=[(ceae_)|(les_)])\\d", X1, perl = T),
             paste0(family.naming0, " ", family.naming1),
             ifelse(
               grepl("(?<=[(ceae_)|(les_)])unclassified", X1, perl = T),
               paste0(family.naming0, " ", family.naming2),
               paste0(family.naming0)
             ))) %>% 
  mutate(genus=gsub("_", " ", sub("(.*ceae)+?_((unclassified|\\d)*(_)*)", "", X1)))

# change it into percentage
james4 <-
  bind_cols(select(james2, X1), as.data.frame(
    apply(
      select(
        james2, 
        Cecum, 
        Transverse, 
        Sigmoid), 2, 
      function(x) x / sum(x) * 100)))

jamesReg <- james4 %>% 
  mutate(james4, 
         meanAbundance=rowMeans(select(james4, Cecum, Transverse, Sigmoid))) %>% 
  arrange(desc(meanAbundance)) %>% 
  top_n(30, meanAbundance) %>% 
  pull(X1)

# collect top 30 from james4X reference
james5 <- james3 %>% 
  filter(X1 %in% jamesReg)

# change order
james5$location_f <- 
  factor(james5$location, labels = c("Cecum", "Transverse", "Sigmoid"))

james6 <- 
  select(james5, location_f, relativeAbundance, genus)

# First plot
james.plot <-
  ggplot(james6,
         aes(x = location_f, y = relativeAbundance, fill = genus)) +
  geom_bar(position = "fill", stat = "identity", show.legend = F) +
  scale_y_continuous(labels = scales::percent_format(accuracy = 1)) + # y axis percentage
  #theme_minimal() +
  theme(axis.title.x = element_blank(),
        panel.background = element_blank()) +
  ylab("Relative abundances (%)") +
  scale_fill_hue(l=60, c=80)


james.table <- data.frame("relativeAbundance"=james5$relativeAbundance[1:30]+
                            james5$relativeAbundance[31:60]+
                            james5$relativeAbundance[61:90],
                          "phylum"=james5$phylum[1:30],
                          "family"=james5$family[1:30],
                          "genus"=james5$genus[1:30])

# get colour pattern 
ggplotColours <- function(n = 6, h = c(0, 360) + 15) {
  if ((diff(h) %% 360) < 1)
    h[2] <- h[2] - 360 / n
  hcl(h = (seq(h[1], h[2], length = n)), c = 100, l = 65)
}

family <- pull(select(james.table, family))
genus <- pull(select(james.table, genus))
james.table2 <- james.table %>% 
  mutate(color=ggplotColours(nrow(james.table))) %>% # just in case 
  mutate(asv=glue("{family}: <i>{genus}</i>"))

# color for long vertical tile (phylum tile)
james.phyl.col <- c("#fddb47", "#58b9b2", "#6585c3", "#e25a4b")

# legend making or second plot
james.legend <- 
  ggplot(james.table2, aes(y = asv)) +
  geom_tile(aes(x = 1, fill = asv), width = 0.9, height = 0.9) +
  geom_tile(aes(x = 0.2), 
            fill = james.phyl.col[as.numeric(as.factor(james.table2$phylum))], 
            width = 0.4) +
  scale_y_discrete(position = "right", expand = c(0,0),
                   name = "") +
  scale_x_continuous(expand = c(0,0), breaks = NULL, name = "") +
  scale_fill_discrete(guide = "none") +
  facet_grid(phylum ~ ., scales = "free_y", space = "free_y",
             switch = "y") +
  theme(axis.ticks = element_blank(),
        strip.background = element_blank(),
        aspect.ratio = 1,
        axis.text.y = element_markdown())

# patchwork
james.plot + james.legend

最终图片final

Answer 1

这是一个示例，说明如何开始将图例制作为单独的情节，稍后您可以将其拼凑在主情节旁边。

基本上，您正在为每个项目制作图块，然后按组对它们进行刻面。将图块与刻面精确成 1:1 有点棘手，因此您必须使用width = ... 和height = ... 来使其看起来正确。

library(ggplot2)

# Example of item-group relations
df <- data.frame(
  group = c("Actinobacteria", "Actinobacteria", "Bacteroidetes", "Bacteroidetes",
            "Firmicutes", "Firmicutes", "Firmicutes"),
  item = c("Bifidobacteriaceae", "Coriobacteriaceae",
           "Bacteroidaceae", "Porphyromonadacea",
           "Acidaminococcacaea", "Clostridiacea", "Clostridiales")
)

group_colours <- c("blue", "green", "red")

ggplot(df, aes(y = item)) +
  geom_tile(aes(x = 1, fill = item), width = 0.9, height = 0.9) +
  geom_tile(aes(x = 0.2), 
            fill = group_colours[as.numeric(as.factor(df$group))], 
            width = 0.4) +
  scale_y_discrete(position = "right", expand = c(0,0),
                   name = "") +
  scale_x_continuous(expand = c(0,0), breaks = NULL, name = "") +
  scale_fill_discrete(guide = "none") +
  facet_grid(group ~ ., scales = "free_y", space = "free_y",
             switch = "y") +
  theme(axis.ticks = element_blank(),
        strip.background = element_blank(),
        aspect.ratio = 1)

^{由reprex package (v0.3.0) 于 2020-08-18 创建}