ENSG00000187634 chr1 [ 4807614, 4807614] + | ENSG00000187634

2025-06-07 03:44:16作者：羿妍玫Ivan

ENSG00000187634 chr1 [ 4807614, 4807614] + | ENSG00000187634

ENSG00000187961 chr1 [ 4855568, 4855568] + | ENSG00000187961

ENSG00000188290 chr1 [ 4899965, 4899965] + | ENSG00000188290

ENSG00000187608 chr1 [ 4955148, 4955148] + | ENSG00000187608

ENSG00000188157 chr1 [ 5016563, 5016563] + | ENSG00000188157

... ... ... ... . ...

ENSG00000131591 chr1 [24894508, 24894508] - | ENSG00000131591

ENSG00000177700 chr1 [24924056, 24924056] - | ENSG00000177700

ENSG00000131584 chr1 [24925042, 24925042] - | ENSG00000131584

ENSG00000177757 chr1 [24925345, 24925345] - | ENSG00000177757

ENSG00000131586 chr1 [24925405, 24925405] - | ENSG00000131586

-------

seqinfo: 1 sequence from an unspecified genome; no seqlengths


We normalize chromatin states in ESC and lung to the TSS of genes with
bivalent states in ESC.


```r
mat_states_esc = normalizeToMatrix(states, tss_biv, value_column = "states_simplified")
mat_states_lung = normalizeToMatrix(states_lung, tss_biv, value_column = "states_simplified")

We also normalize methylation in ESC and lung to the same TSS.

mat_meth_esc = normalizeToMatrix(meth, tss_biv, value_column = "E003", mean_mode = "absolute",
	smooth = TRUE)
mat_meth_lung = normalizeToMatrix(meth, tss_biv, value_column = "E096", mean_mode = "absolute",
	smooth = TRUE)

We apply k-means clustering on the chromatin states in ESC (1kb upstream and downstream of TSS) to separate genes with bivalent TSS into two groups.

split = kmeans(mat_states_esc[, 40:60], centers = 2)$cluster

Now we make the heatmap list. The order of heatmaps are: chromatin states in ESC, chromatin states in lung, methylation in ESC and methylation in lung. Expression in ESC and lung are also added to the right side of the heatmap list.

expr_esc = expr[names(tss_biv), "E003"]
expr_lung = expr[names(tss_biv), "E096"]
ht_list = EnrichedHeatmap(mat_states_esc, name = "states_esc", col = states_col, 
	row_split = split, cluster_rows = TRUE,
	top_annotation = HeatmapAnnotation(enrich = anno_enriched(gp = gpar(lty = 1:2)))) +
EnrichedHeatmap(mat_states_lung, name = "states_lung", col = states_col, 
	top_annotation = HeatmapAnnotation(enrich = anno_enriched(gp = gpar(lty = 1:2)))) +
EnrichedHeatmap(mat_meth_esc, name = "meth_esc", col = meth_col_fun,
	top_annotation = HeatmapAnnotation(enrich = anno_enriched(gp = gpar(lty = 1:2)))) +
EnrichedHeatmap(mat_meth_lung, name = "meth_lung", col = meth_col_fun,
	top_annotation = HeatmapAnnotation(enrich = anno_enriched(gp = gpar(lty = 1:2)))) +
Heatmap(log2(expr_esc + 1), name = "expr_esc", show_row_names = FALSE, width = unit(5, "mm"),
	col = colorRamp2(c(0, 5), c("white", "red"))) +
Heatmap(log2(expr_lung + 1), name = "expr_lung", show_row_names = FALSE, width = unit(5, "mm"),
	col = colorRamp2(c(0, 5), c("white", "red")))
draw(ht_list, ht_gap = unit(8, "mm"))

plot of chunk tssbiv_heatmap

From the heatmap, we can see in cluster 1, the bivalent TSS in ESC are transited to active states in lung while in cluster 2, the bivalent TSS in ESC are transited to repressive states in lung. Also in cluster 1, the methylation is low in both ESC and lung while in cluster 2, the methylation is high in lung. The expression in cluster 1 is higher than in cluster 2 in lung.

Session info

sessionInfo()

## R version 3.4.0 (2017-04-21)
## Platform: x86_64-pc-linux-gnu (64-bit)
## Running under: Ubuntu 16.04.2 LTS
## 
## Matrix products: default
## BLAS: /usr/lib/openblas-base/libblas.so.3
## LAPACK: /usr/lib/libopenblasp-r0.2.18.so
## 
## locale:
##  [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C               LC_TIME=en_US.UTF-8       
##  [4] LC_COLLATE=en_US.UTF-8     LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8   
##  [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                  LC_ADDRESS=C              
## [10] LC_TELEPHONE=C             LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       
## 
## attached base packages:
## [1] grid      stats4    parallel  stats     graphics  grDevices utils     datasets  methods  
## [10] base     
## 
## other attached packages:
##  [1] circlize_0.4.1            EnrichedHeatmap_1.7.3     data.table_1.10.4        
##  [4] GenomicFeatures_1.28.3    AnnotationDbi_1.38.1      Biobase_2.36.2           
##  [7] GenomicRanges_1.28.3      GenomeInfoDb_1.12.2       IRanges_2.10.2           
## [10] S4Vectors_0.14.3          BiocGenerics_0.22.0       knitr_1.16               
## [13] markdown_0.8              evaluate_0.10             stringr_1.2.0            
## [16] rtracklayer_1.36.4        GenomicAlignments_1.12.1  Rsamtools_1.28.0         
## [19] Biostrings_2.44.1         XVector_0.16.0            SummarizedExperiment_1.6.3
## [22] DelayedArray_0.2.7        matrixStats_0.52.2        BiocParallel_1.10.1      
## [25] BSgenome_1.44.0           rmarkdown_1.6             TxDb.Hsapiens.UCSC.hg19.knownGene_3.2.2
## [28] GenomicInfoDb_1.12.1     
## 
## loaded via a namespace (and not attached):
##  [1] colorspace_1.3-2            rprojroot_1.2               htmlTable_1.9              
##  [4] base64enc_0.1-3             dichromat_2.0-0             rstudioapi_0.6             
##  [7] bit64_0.9-7                 splines_3.4.0               R.methodsS3_1.7.1          
## [10] doParallel_1.0.10           geneplotter_1.54.0          annotate_1.54.0            
## [13] cluster_2.0.6               R.oo_1.21.0                 shiny_1.0.3                
## [16] compiler_3.4.0              httr_1.2.1                  backports_1.1.0            
## [19] assertthat_0.2.0            Matrix_1.2-10               lazyeval_0.2.0             
## [22] htmltools_0.3.6             tools_3.4.0                 gtable_0.2.0               
## [25] glue_1.1.1                  GenomeInfoDbData_0.99.0     reshape2_1.4.2             
## [28] dplyr_0.5.0                 Rcpp_0.12.11                BiocInstaller_1.26.0       
## [31] iterators_1.0.8             xfun_0.1                    XML_3.98-1.7               
## [34] zlibbioc_1.22.0             scales_0.4.1                VariantAnnotation_1.22.1   
## [37] hms_0.3                     yaml_2.1.14                 memoise_1.1.0              
## [40] gridExtra_2.2.1             ggplot2_2.2.1               rpart_4.1-11               
## [43] latticeExtra_0.6-28         stringi_1.1.5               RSQLite_1.1-2              
## [46] highr_0.6                   foreach_1.4.3               checkmate_1.8.2            
## [49] caTools_1.17.1              BiocStyle_2.4.0             rlang_0.1.1                
## [52] pkgconfig_2.0.1             bitops_1.0-6                matrixcalc_1.0-3           
## [55] lattice_0.20-35             purrr_0.2.2.2               htmlwidgets_0.8            
## [58] bit_1.1-12                  tidyselect_0.2.0            plyr_1.8.4                 
## [61] magrittr_1.5                R6_2.2.1                    DBI_0.6-1                  
## [64] pillar_1.0.1                foreign_0.8-68              survival_2.41-3            
## [67] RCurl_1.95-4.8              tibble_1.3.3                rjson_0.2.15               
## [70] GetoptLong_0.1.6            digest_0.6.12               xtable_1.8-2               
## [73] tidyr_0.6.3                 R.utils_2.5.0               munsell_0.4.3              
## [76] viridisLite_0.2.0

引言

在基因组学研究中，我们经常需要分析不同类型的基因组信号在特定基因组特征（如转录起始位点TSS、基因体等）周围的富集模式。EnrichedHeatmap包提供了一种强大的方法来可视化这些富集模式。本文将重点介绍如何使用EnrichedHeatmap处理和分析分类（categorical）基因组信号，特别是染色质状态数据。

染色质状态简介

染色质状态是通过整合多种表观遗传标记（如组蛋白修饰）来定义的基因组区域分类。ChromHMM等工具可以将基因组划分为不同的功能状态，如：

活跃转录起始位点（TssActive）
转录区域（Transcript）
增强子区域（Enhancer）
异染色质（Heterochromatin）
双价状态（TssBivalent）
抑制状态（Repressive）
静息状态（Quiescent）

这些分类数据为我们理解基因组功能提供了重要线索。

数据准备

首先我们需要加载必要的R包并准备数据：

library(GenomicRanges)
library(data.table)
library(EnrichedHeatmap)
library(circlize)

从Roadmap项目中获取染色质状态数据后，我们可以将其转换为GRanges对象：

states_bed = fread("染色质状态数据文件路径")
states = GRanges(seqnames = states_bed[[1]], 
                ranges = IRanges(states_bed[[2]] + 1, states_bed[[3]]), 
                states = states_bed[[4]])

为了简化分析，我们可以将相似的染色质状态进行合并：

state_mapping = c(
    "1_TssA" = "TssActive",
    "2_TssAFlnk" = "TssActive",
    # 其他状态映射...
)
states$simplified_states = state_mapping[states$states]

基本可视化

转录起始位点分析

首先我们提取基因的TSS区域：

library(GenomicFeatures)
txdb = loadDb("转录组数据库路径")
genes = genes(txdb)
tss = promoters(genes, upstream = 0, downstream = 1)

然后我们将染色质状态信号标准化到TSS周围：

mat_states = normalizeToMatrix(states, tss, value_column = "simplified_states")

使用EnrichedHeatmap进行可视化：

state_colors = c(
    TssActive = "red",
    Transcript = "green",
    # 其他状态颜色...
)
EnrichedHeatmap(mat_states, name = "states", col = state_colors, cluster_rows = TRUE)

基因体分析

我们也可以分析染色质状态在基因体上的分布：

mat_gene_body = normalizeToMatrix(states, genes, value_column = "simplified_states")
EnrichedHeatmap(mat_gene_body, name = "states", col = state_colors) +
rowAnnotation(gene_len = anno_points(log10(width(genes) + 1))

高级分析：整合多组学数据

结合DNA甲基化和基因表达

为了更全面地理解染色质状态的功能意义，我们可以整合DNA甲基化和基因表达数据：

# 标准化甲基化数据
mat_meth = normalizeToMatrix(meth_data, tss, value_column = "sample1")

# 创建热图列表
ht_list = EnrichedHeatmap(mat_states, name = "states") +
          EnrichedHeatmap(mat_meth, name = "methylation") +
          Heatmap(log2(expr_data + 1), name = "expression")
draw(ht_list)

双价TSS状态分析

在胚胎干细胞中，双价TSS状态（同时具有活跃和抑制标记）是一个重要特征。我们可以分析这些状态在分化过程中的变化：

# 识别具有双价状态的TSS
mat_bivalent = normalizeToMatrix(states[states$simplified_states == "TssBivalent"], tss)
bivalent_tss = tss[rowSums(mat_bivalent[, 40:60]) > 0]  # 1kb窗口内的双价状态

# 比较不同细胞类型
mat_states_esc = normalizeToMatrix(esc_states, bivalent_tss)
mat_states_diff = normalizeToMatrix(diff_states, bivalent_tss)

# 可视化比较
ht_list = EnrichedHeatmap(mat_states_esc, name = "ESC states") +
          EnrichedHeatmap(mat_states_diff, name = "Differentiated states")
draw(ht_list)