EnrichedHeatmap--对富集结果进行热图可视化！

作图丫

发布于 2022-03-29 13:50:12

1.6K0

发布于 2022-03-29 13:50:12

文章被收录于专栏：作图丫

导语

GUIDE ╲

富集热图是一种特殊类型的热图，可将特定目标区域上基因组信号的富集可视化。例如可以在转录起始位点富集组蛋白修饰。

背景介绍

今天小编给大家带来的就是一个专门用来绘制富集热图的R包--EnrichedHeatmap，作者是基于 ComplexHeatmap 包实现的热图绘制，通过使用EnrichedHeatmap包，我们可以对各种表观遗传学数据进行丰富的可视化！

R包安装

BiocManager::install("EnrichedHeatmap")
library(EnrichedHeatmap)

可视化展示

基本可视化

首先加载来自 Roadmap 数据集的人类肺组织的数据。

set.seed(123)
load(system.file("extdata", "chr21_test_data.RData", package = "EnrichedHeatmap")
##对 100000 个 CpG 位点进行了下采样以获得甲基化数据。
tss = promoters(genes, upstream = 0, downstream = 1)
tss[1:5]

与其他工具类似，可视化的任务分为两个步骤：

可视化的流程分为两部分，通过归一化为矩阵来获得基因组信号和目标之间的关联，和通过热图可视化矩阵。

mat1 = normalizeToMatrix(H3K4me3, tss, value_column = "coverage", 
    extend = 5000, mean_mode = "w0", w = 50)
mat1
## Normalize H3K4me3 to tss:
##   Upstream 5000 bp (100 windows)
##   Downstream 5000 bp (100 windows)
##   Include target regions (width = 1)
##   720 target regions
EnrichedHeatmap(mat1, name = "H3K4me3")

参数设置

与普通热图类似，设置颜色的最简单方法是提供颜色向量。

EnrichedHeatmap(mat1, col = c("white", "red"), name = "H3K4me3")

定义一个颜色映射函数，它只将颜色映射到小于 99% 的值，大于 99% 的值使用与 99% 相同的颜色（避免极值的影响）。

library(circlize)
col_fun = colorRamp2(quantile(mat1, c(0, 0.99)), c("white", "red"))
EnrichedHeatmap(mat1, col = col_fun, name = "H3K4me3")

按行拆分

通过指定 row_split 选项，按向量或数据框拆分行。

EnrichedHeatmap(mat1, col = col_fun, name = "H3K4me3", 
    row_split = sample(c("A", "B"), length(genes), replace = TRUE),
    column_title = "Enrichment of H3K4me3")

通过指定 row_km 选项，通过 k 均值聚类拆分行。

set.seed(123)
EnrichedHeatmap(mat1, col = col_fun, name = "H3K4me3", row_km = 3,
    column_title = "Enrichment of H3K4me3", row_title_rot = 0)

对行的聚类

EnrichedHeatmap(mat1, col = col_fun, name = "H3K4me3", 
    cluster_rows = TRUE, column_title = "Enrichment of H3K4me3")

目标区域的扩展

# 上游1kb, 下游2kb（上游或下游可以设置为0）
mat12 = normalizeToMatrix(H3K4me3, tss, value_column = "coverage", 
    extend = c(1000, 2000), mean_mode = "w0", w = 50)
EnrichedHeatmap(mat12, name = "H3K4me3", col = col_fun)

富集注释

富集注释的axis由 anno_enriched() 中的axis_param 设置。

EnrichedHeatmap(mat1, col = col_fun, name = "H3K4me3", 
    top_annotation = HeatmapAnnotation(
        enriched = anno_enriched(
            ylim = c(0, 10),
            axis_param = list(
                at = c(0, 5, 10),
                labels = c("zero", "five", "ten"),
                side = "left",
                facing = "outside"
    )))
)

生成矩阵时，可以通过将 smooth 设置为 TRUE 来进行平滑处理。

mat1_smoothed = normalizeToMatrix(H3K4me3, tss, value_column = "coverage", 
    extend = 5000, mean_mode = "w0", w = 50, background = 0, smooth = TRUE)
EnrichedHeatmap(mat1_smoothed, col = col_fun, name = "H3K4me3_smoothed",
    column_title = "smoothed") +
EnrichedHeatmap(mat1, col = col_fun, name = "H3K4me3", column_title = "unsmoothed")

在上面的图中，大家可能会觉得左侧的热图比右侧的未平滑热图更好。下面，我们将证明平滑可以显着改善甲基化数据集的富集模式。

##未平滑
mat2 = normalizeToMatrix(meth, tss, value_column = "meth", mean_mode = "absolute",
                         extend = 5000, w = 50, background = NA)
meth_col_fun = colorRamp2(c(0, 0.5, 1), c("blue", "white", "red"))
EnrichedHeatmap(mat2, col = meth_col_fun, name = "methylation", column_title = "methylation near TSS")
##平滑
mat2 = normalizeToMatrix(meth, tss, value_column = "meth", mean_mode = "absolute",
                         extend = 5000, w = 50, background = NA, smooth = TRUE)
EnrichedHeatmap(mat2, col = meth_col_fun, name = "methylation", column_title = "methylation near TSS")

多个热图

EnrichedHeatmap 包的强大之处在于可以串联并行热图，可以用于丰富热图、普通热图以及行注释。这提供了一种非常有效的方式来可视化多个信息源。

EnrichedHeatmap(mat1, col = col_fun, name = "H3K4me3",
                top_annotation = HeatmapAnnotation(enrich = anno_enriched(axis_param = list(side = "left")))) + 
  EnrichedHeatmap(mat2, col = meth_col_fun, name = "methylation") +
  Heatmap(log2(rpkm+1), col = c("white", "orange"), name = "log2(rpkm+1)", 
          show_row_names = FALSE, width = unit(5, "mm"))

partition = paste0("cluster", kmeans(mat1, centers = 3)$cluster)
lgd = Legend(at = c("cluster1", "cluster2", "cluster3"), title = "Clusters", 
    type = "lines", legend_gp = gpar(col = 2:4))
ht_list = Heatmap(partition, col = structure(2:4, names = paste0("cluster", 1:3)), name = "partition",
              show_row_names = FALSE, width = unit(3, "mm")) +
          EnrichedHeatmap(mat1, col = col_fun, name = "H3K4me3",
              top_annotation = HeatmapAnnotation(lines = anno_enriched(gp = gpar(col = 2:4))), 
              column_title = "H3K4me3") + 
          EnrichedHeatmap(mat2, col = meth_col_fun, name = "methylation",
              top_annotation = HeatmapAnnotation(lines = anno_enriched(gp = gpar(col = 2:4))), 
              column_title = "Methylation") +
          Heatmap(log2(rpkm+1), col = c("white", "orange"), name = "log2(rpkm+1)", 
              show_row_names = FALSE, width = unit(15, "mm"),
              top_annotation = HeatmapAnnotation(summary = anno_summary(gp = gpar(fill = 2:4),
                outline = FALSE, axis_param = list(side = "right"))))
draw(ht_list, split = partition, annotation_legend_list = list(lgd), 
    ht_gap = unit(c(2, 8, 8), "mm"))

分别可视化正信号和负信号

load(system.file("extdata", "H3K4me1_corr_normalize_to_tss.RData", package = "EnrichedHeatmap"))
mat_H3K4me1
corr_col_fun = colorRamp2(c(-1, 0, 1), c("darkgreen", "white", "red"))
EnrichedHeatmap(mat_H3K4me1, col = corr_col_fun, name = "corr_H3K4me1",
                top_annotation = HeatmapAnnotation(
                  enrich = anno_enriched(gp = gpar(neg_col = "darkgreen", pos_col = "red"),
                                         axis_param = list(side = "left"))
                ), column_title = "separate neg and pos") +
  EnrichedHeatmap(mat_H3K4me1, col = corr_col_fun, show_heatmap_legend = FALSE,
                  top_annotation = HeatmapAnnotation(enrich = anno_enriched(value = "abs_mean")),
                  column_title = "pool neg and pos")