## 基本散點圖展示 ### 應用場景 heatmap運用的地方比較多，能夠展示的信息也非常豐富，常用的就是展示基因在不同組別樣本中的表達情況。 ### 利用limma包做一個差異表達分析 ```R load("./test_exp.Rdata") # heatmap 在差異表達基因可以顯示的更有層次，那這里我們簡單的做一個差異分析，用heatmap展示差異表達基因 library(limma) # 構建一個phenotype的文件標示出各個樣本tumor和normal屬性 rt_sam_m <- data.frame(cbind(colnames(rt), c(rep('Tumor', 57), rep('Normal', 57))), stringsAsFactors = FALSE) colnames(rt_sam_m) <- c('samples_id', 'group') m_group = factor(rt_sam_m$group, levels=c('Tumor', 'Normal')) design_m = model.matrix(~0 + m_group) row.names(design_m) <- rt_sam_m$samples_id colnames(design_m) <- c('Tumor','Normal') # 差異分析 m_fit <- lmFit(rt, design_m) cont.matrix <- makeContrasts(TumorvsNormal = Tumor-Normal, levels = design_m) m_fit2 <- contrasts.fit(m_fit, cont.matrix) m_fit3 <- eBayes(m_fit2) rt_diff <- topTreat(m_fit3, number = length(row.names(rt))) rt_diff2 <- rt_diff[which(abs(rt_diff$logFC) > 1 & rt_diff$P.Value < 0.05), ] dim(rt_diff2)#798 6, 有798個差異基因 ``` >rt_diff2數據形式如下圖所示  ### 利用heatmap.2做heatmap ```R # 作為實例數據，對上述前100個基因做heatmap library(gplots) rt_ht <- rt[row.names(rt_diff2)[order(abs(rt_diff2$logFC), decreasing = TRUE)[1:100]], ]#取出前100的DEGs mat_ht <- as.matrix(apply(rt_ht, 2, function(x){as.numeric(x)})) mat_ht_z <- apply(mat_ht, 2, FUN = function(x){(x-median(x))/sd(x)}) row.names(mat_ht_z) <- row.names(rt_ht) heatmap.2(mat_ht_z, col = colorpanel(99, "blue", "black", "red"), dendrogram = "both", keysize = 1, hclustfun = function(x){hclust(x, method = 'ward.D2')}, trace = "none", density.info = "none") # 顯示tumor和normal標示 heatmap.cols <- c(rep('red', 57), rep('blue', 57)) # ColSideColors設置分組的ColSide heatmap.2(mat_ht_z, col = colorpanel(99, "blue", "black", "red"), dendrogram = "both", keysize = 1, hclustfun = function(x){hclust(x, method = 'ward.D2')}, trace = "none", density.info = "none", ColSideColors = heatmap.cols) ```   ```R # 選擇row或者col是否聚類 # Colv = T, Rowv = F heatmap.2(mat_ht_z, col = colorpanel(99, "blue", "black", "red"), dendrogram = "both", keysize = 1, hclustfun = function(x){hclust(x, method = 'ward.D2')}, trace = "none", density.info = "none", ColSideColors = heatmap.cols, Colv = F, Rowv = T) # 可以labs都去掉 # labRow = NA, labCol = NA heatmap.2(mat_ht_z, col = colorpanel(99, "blue", "black", "red"), dendrogram = "both", keysize = 1, hclustfun = function(x){hclust(x, method = 'ward.D2')}, trace = "none", density.info = "none", ColSideColors = heatmap.cols, Colv = F, Rowv = T, labRow = NA, labCol = NA) # 加上legend來標示tumor 和Normal heatmap.2(mat_ht_z, col = colorpanel(99, "blue", "black", "red"), dendrogram = "both", keysize = 1, hclustfun = function(x){hclust(x, method = 'ward.D2')}, trace = "none", density.info = "none", ColSideColors = heatmap.cols, Colv = F, Rowv = T, labRow = NA, labCol = NA) legend('topleft', legend = c('tumor', 'normal'), pch = 15, bty = 'n', col = c('red', 'blue'), text.col = "black") ```