问题描述
我想复制以下paper中的图形。
它被卡在分隔X1列上。我想使用正则表达式，但不知道如何使用。 我有一个计划，用下划线分隔符分隔每个单词（我有一个列表），然后分别将[-tes＆-ria]和[-ceae]后缀单词提取到Phylum和Family中。在那之后，家庭之后的话应该被归为属。为了准确起见，可能会将“未分类”和少于5个字符的单词的条件分组到前面的单词。
还有，是否可以在每个家族中添加与相对丰度图相对应的小颜色标签？
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,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,"_"),value = T
)))
我是R的新手，上面的代码大部分是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)
APPEND
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("Sigmoi")))
) %>%
select(X1,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",perl = T))
family.naming2 <-
regmatches(james2$X1,regexpr("(?<=ceae_|les_)unclassified",perl = T))
family.naming3 <-
ifelse(
grepl("(?<=[(ceae_)|(les_)])\\d",perl = T),paste0(family.naming0," ",family.naming1),ifelse(
grepl("(?<=[(ceae_)|(les_)])unclassified",family.naming2),paste0(family.naming0)
))
james3 <- james2 %>%
gather("Cecum","Transverse","Sigmoid",key = "location",value = "abundance") %>%
mutate(relativeAbundance=abundance/sum(abundance) * 100) %>%
mutate(phylum=gsub("(_.*)","",X1)) %>% # extract phylum value with selecting first word
mutate(family=
ifelse(
grepl("(?<=[(ceae_)|(les_)])\\d",X1,ifelse(
grepl("(?<=[(ceae_)|(les_)])unclassified",paste0(family.naming0)
))) %>%
mutate(genus=gsub("_",sub("(.*ceae)+?_((unclassified|\\d)*(_)*)",X1)))
# change it into percentage
james4 <-
bind_cols(select(james2,as.data.frame(
apply(
select(
james2,function(x) x / sum(x) * 100)))
jamesReg <- james4 %>%
mutate(james4,meanAbundance=rowMeans(select(james4,Sigmoid))) %>%
arrange(desc(meanAbundance)) %>%
top_n(30,meanAbundance) %>%
pull(X1)
# collect top 30 from james4X reference
james5 <- james3 %>%
filter(X1 %in% jamesReg)
# change order
james5$location_f <-
factor(james5$location,labels = c("Cecum","Sigmoid"))
james6 <-
select(james5,location_f,relativeAbundance,genus)
# First plot
james.plot <-
ggplot(james6,aes(x = location_f,y = relativeAbundance,fill = genus)) +
geom_bar(position = "fill",stat = "identity",show.legend = F) +
scale_y_continuous(labels = scales::percent_format(accuracy = 1)) + # y axis percentage
#theme_minimal() +
theme(axis.title.x = element_blank(),panel.background = element_blank()) +
ylab("Relative abundances (%)") +
scale_fill_hue(l=60,c=80)
james.table <- data.frame("relativeAbundance"=james5$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,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
解决方法
这是一个示例，说明如何开始将图例制作为单独的图，以后可以将其拼凑到主图旁边。
基本上，您要为每个项目制作图块，然后按组对其进行分面。使刻面与刻面的比例完全为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","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,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