将type添加到scale_fill_discrete
type：以下选项之一：
·颜色代码的字符向量。
·颜色代码的字符向量列表。
·返回离散颜色/填充比例的函数

ggplot(data, aes(fill=condition, y=value, x=specie)) + 
  geom_bar(position="dodge", stat="identity") + 
  labs(y = "Number of genes", x = "Cytokines") + 
  theme_classic() + 
  theme(plot.title = element_text(hjust = 0.5)) + 
  scale_y_continuous(expand = c(0, 0), limits = c(0, NA)) + 
  scale_fill_discrete(labels=c('up', 'down'), type=c('#eb4034','#0a0a0a'))

这里有两个问题。首先，scale_fill_manual()本质上会“覆盖”scale_fill_discrete()。相反，只使用一个包含所有相关参数的scale_fill_*()调用：

library(ggplot2)

p + 
  labs(y = "Number of genes", x = "Cytokines")+
  theme_classic()+
  theme(plot.title = element_text(hjust = 0.5))+
  theme(axis.line=element_line(size=1))+
  scale_y_continuous(expand = c(0, 0), limits = c(0, NA))+
  scale_fill_manual(labels = c('up', 'down'), values=c('#eb4034', '#0a0a0a'))

然而，当前labels参数实际上是在重新编码数据，因此"up"值被标记为"down"，反之亦然，我认为这不是您想要的。我最好的猜测是，您实际上是在尝试更改标签在图例中显示的顺序。如果是这样，您可以将condition的因子水平更改为您想要的顺序：

data$condition <- factor(data$condition, c("up", "down"))

p <- ggplot(data, aes(fill=condition, y=value, x=specie)) +
  geom_bar(position="dodge", stat="identity")

p + 
  labs(y = "Number of genes", x = "Cytokines")+
  theme_classic()+
  theme(plot.title = element_text(hjust = 0.5))+
  theme(axis.line=element_line(size=1))+
  scale_y_continuous(expand = c(0, 0), limits = c(0, NA))+
  scale_fill_manual(values=c('#eb4034', '#0a0a0a'))