使用ggplot2绘制非数值分类数据的可视化方案咨询

方案1：分组丰度热图 #

适用于同时展示6个组别、20个分类项的丰度等级全局分布，是科研场景中多分类组间对比的常用方案，可直观呈现不同组的分类丰度规律。
先对丰度等级做有序因子处理，匹配丰度从高到低的逻辑：

library(tidyverse)

# 数据预处理
df_long <- df %>% 
  pivot_longer(starts_with("cat_"), names_to = "category") %>%
  mutate(value = ifelse(value == "-", NA, value),
         # 定义丰度等级顺序
         value = factor(value, levels = c("D", "Mj", "Mn", "Tr"))) %>%
  drop_na() %>%
  # 统计每组+每个分类+每个丰度等级的出现频次
  count(group, category, value, name = "freq")

# 绘制热图
ggplot(df_long, aes(x = category, y = group, fill = value)) +
  geom_tile(color = "white", size = 0.3) +
  # 可叠加频次文本
  geom_text(aes(label = freq), color = "black", size = 3) +
  scale_fill_brewer(palette = "RdYlBu", direction = -1, na.value = "grey90",
                    labels = c("D: >50%", "Mj: 5%-50%", "Mn: 1%-5%", "Tr: <1%")) +
  labs(x = "分类项", y = "组别", fill = "丰度等级") +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))

方案2：堆叠百分比条形图 #

适用于对比不同组别下，同一分类项的丰度等级占比差异，可直接呈现组间分布的异质性。

ggplot(df_long, aes(x = category, y = freq, fill = value)) +
  geom_col(position = "fill") +
  facet_wrap(~ group, ncol = 2) +
  scale_y_continuous(labels = scales::percent_format()) +
  scale_fill_brewer(palette = "RdYlBu", direction = -1,
                    labels = c("D: >50%", "Mj: 5%-50%", "Mn: 1%-5%", "Tr: <1%")) +
  labs(x = "分类项", y = "占比", fill = "丰度等级") +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))

方案3：分面抖动散点图 #

适用于需要保留单样本层面的原始观测信息的场景，适配组内样本量不固定的情况，不会因汇总丢失原始数据特征。

# 保留样本维度的预处理
df_sample_long <- df %>% 
  pivot_longer(starts_with("cat_"), names_to = "category") %>%
  mutate(value = ifelse(value == "-", NA, value),
         value = factor(value, levels = c("D", "Mj", "Mn", "Tr"))) %>%
  drop_na()

ggplot(df_sample_long, aes(x = value, y = category, color = value)) +
  geom_jitter(width = 0.2, height = 0.2, alpha = 0.7, size = 2) +
  facet_wrap(~ group, ncol = 2) +
  scale_color_brewer(palette = "RdYlBu", direction = -1,
                    labels = c("D: >50%", "Mj: 5%-50%", "Mn: 1%-5%", "Tr: <1%")) +
  labs(x = "丰度等级", y = "分类项", color = "丰度等级") +
  theme_minimal()

内容的提问来源于stack exchange，提问作者aterhorst