降维
在组学分析中，一般通过降维算法得到低纬度如二维或三维的新坐标数据，再结合可视化技术去展示样本的在新坐标的空间分布，接着加上统计检验结果证实整体组学水平上组间的差异性。降维算法有基于线性模型的PCA，也有基于非线性的tSNE和UMAP等方法。
示例数据和代码领取
详见：R实战| PCA、tSNE、UMAP三种降维方法在R中的实现
PCA
主成分分析（Principal Component Analysis，PCA）是最常用的无监督学习方法。
rm(list = ls()) library(tidyverse) library(broom) library(palmerpenguins) # 示例数据 penguins <- penguins %>% drop_na() %>% select(-year) head(penguins) # 使用prcomp()进行PCA # PCA前对数值型数据进行标准化 pca_fit <- penguins %>% select(where(is.numeric)) %>% scale() %>% prcomp() # 查看成分重要性 summary(pca_fit) # 可视化PC1和PC2 pca_fit %>% augment(penguins) %>% rename_at(vars(starts_with(".fitted")), list(~str_replace(.,".fitted",""))) %>% ggplot(aes(x=PC1, y=PC2, color=species, shape=sex))+ geom_point()
PCA plot: PC1 vs PC2
UMAP
数据预处理
## UMAP rm(list = ls()) library(tidyverse) library(palmerpenguins) #install.packages("umap") library(umap) theme_set(theme_bw(18)) penguins <- penguins %>% drop_na() %>% select(-year)%>% mutate(ID=row_number()) penguins_meta <- penguins %>% select(ID, species, island, sex)
使用umap包进行umap分析
set.seed(142) umap_fit <- penguins %>% select(where(is.numeric)) %>% column_to_rownames("ID") %>% scale() %>% umap() umap_df <- umap_fit$layout %>% as.data.frame()%>% rename(UMAP1="V1", UMAP2="V2") %>% mutate(ID=row_number())%>% inner_join(penguins_meta, by="ID") umap_df %>% head()
可视化
# 可视化 umap_df %>% ggplot(aes(x = UMAP1, y = UMAP2, color = species, shape = sex))+ geom_point()+ labs(x = "UMAP1", y = "UMAP2", subtitle = "UMAP plot")
UMAP plot in R: Example 1
# 分面 umap_df %>% ggplot(aes(x = UMAP1, y = UMAP2, color = species)) + geom_point(size=3, alpha=0.5)+ facet_wrap(~island)+ labs(x = "UMAP1", y = "UMAP2", subtitle="UMAP plot")+ theme(legend.position="bottom")
UMAP plot in R: Example 2
# 圈出异常样本 library(ggforce) umap_df %>% ggplot(aes(x = UMAP1, y = UMAP2, color = species, shape = sex)) + geom_point() + labs(x = "UMAP1", y = "UMAP2", subtitle="UMAP plot") + geom_circle(aes(x0 = -5, y0 = -3.8, r = 0.65), color = "green", inherit.aes = FALSE)
UMAP Plot to Identify Potential sample mix-ups
tSNE
数据预处理
## tSNE rm(list = ls()) library(tidyverse) library(palmerpenguins) library(Rtsne) theme_set(theme_bw(18)) penguins <- penguins %>% drop_na() %>% select(-year)%>% mutate(ID=row_number()) penguins_meta <- penguins %>% select(ID,species,island,sex)
使用Rtsne 包进行tSNE 分析
set.seed(142) tSNE_fit <- penguins %>% select(where(is.numeric)) %>% column_to_rownames("ID") %>% scale() %>% Rtsne() tSNE_df <- tSNE_fit$Y %>% as.data.frame() %>% rename(tSNE1="V1", tSNE2="V2") %>% mutate(ID=row_number()) tSNE_df <- tSNE_df %>% inner_join(penguins_meta, by="ID") tSNE_df %>% head()
可视化
tSNE_df %>% ggplot(aes(x = tSNE1, y = tSNE2, color = species, shape = sex))+ geom_point()+ theme(legend.position="bottom")
tSNE Plot
参考
- How To Make tSNE plot in R - Data Viz with Python and R (datavizpyr.com)
- How to make UMAP plot in R - Data Viz with Python and R (datavizpyr.com)
- How To Make PCA Plot with R - Data Viz with Python and R (datavizpyr.com)
