|
5
|
1 ---
|
|
|
2 title: 'DESeq2: Visualization'
|
|
|
3 output:
|
|
|
4 html_document:
|
|
|
5 number_sections: true
|
|
|
6 toc: true
|
|
|
7 theme: cosmo
|
|
|
8 highlight: tango
|
|
|
9 ---
|
|
|
10
|
|
|
11 ```{r setup, include=FALSE, warning=FALSE, message=FALSE}
|
|
|
12 knitr::opts_chunk$set(
|
|
|
13 echo = ECHO
|
|
|
14 )
|
|
|
15
|
|
|
16 library(stringi)
|
|
|
17 library(DESeq2)
|
|
|
18 library(pheatmap)
|
|
|
19 library(PoiClaClu)
|
|
|
20 library(RColorBrewer)
|
|
|
21 ```
|
|
|
22
|
|
|
23 # Import workspace
|
|
|
24
|
|
|
25 ```{r eval=TRUE}
|
|
|
26 fcp = file.copy("DESEQ_WORKSPACE", "deseq.RData")
|
|
|
27 load("deseq.RData")
|
|
|
28 ```
|
|
|
29
|
|
|
30 # Visualization
|
|
|
31
|
|
|
32 ## Heatmaps of sample-to-sample distances {.tabset}
|
|
|
33
|
|
|
34 ### rlog-transformed values
|
|
|
35
|
|
|
36 ```{r}
|
|
|
37 sampleDistMatrix <- as.matrix( sampleDists )
|
|
|
38 colors <- colorRampPalette( rev(brewer.pal(9, "Blues")) )(255)
|
|
|
39 pheatmap(sampleDistMatrix,
|
|
|
40 # clustering_distance_rows = sampleDists,
|
|
|
41 clustering_distance_cols = sampleDists,
|
|
|
42 col = colors)
|
|
|
43 ```
|
|
|
44
|
|
|
45 ### Poisson Distance
|
|
|
46
|
|
|
47 ```{r}
|
|
|
48 count_t = t(counts(dds))
|
|
|
49 rownames(count_t) = colnames(counts(dds))
|
|
|
50 poisd <- PoissonDistance(count_t)
|
|
|
51 samplePoisDistMatrix <- as.matrix( poisd$dd )
|
|
|
52 rownames(samplePoisDistMatrix) = rownames(count_t)
|
|
|
53 colnames(samplePoisDistMatrix) = rownames(count_t)
|
|
|
54 pheatmap(samplePoisDistMatrix,
|
|
|
55 # clustering_distance_rows = poisd$dd,
|
|
|
56 clustering_distance_cols = poisd$dd,
|
|
|
57 col = colors)
|
|
|
58 ```
|
|
|
59
|
|
|
60
|
|
|
61 ## PCA plots {.tabset}
|
|
|
62
|
|
|
63 ### Using `plotPCA()` function
|
|
|
64
|
|
|
65 ```{r}
|
|
|
66 # interest groups
|
|
|
67 col_index = as.numeric(strsplit("INTGROUPS_PCA", ',')[[1]])
|
|
|
68 intgroup_pca = colnames(sample_table)[col_index]
|
|
|
69 ```
|
|
|
70
|
|
|
71 ```{r}
|
|
|
72 plotPCA(rld, intgroup = intgroup_pca)
|
|
|
73 ```
|
|
|
74
|
|
|
75
|
|
|
76 ### Using *ggplot2*
|
|
|
77
|
|
|
78 ```{r}
|
|
|
79 pcaData <- plotPCA(rld, intgroup = intgroup_pca, returnData = TRUE)
|
|
|
80 percentVar <- round(100 * attr(pcaData, "percentVar"))
|
|
|
81 ggplot(pcaData, aes(x = PC1, y = PC2, color = time)) +
|
|
|
82 geom_point(size =3) +
|
|
|
83 xlab(paste0("PC1: ", percentVar[1], "% variance")) +
|
|
|
84 ylab(paste0("PC2: ", percentVar[2], "% variance")) +
|
|
|
85 coord_fixed()
|
|
|
86 ```
|
|
|
87
|
|
|
88 ### PCA data table
|
|
|
89
|
|
|
90 ```{r}
|
|
|
91 knitr::kable(pcaData)
|
|
|
92 ```
|
|
|
93
|
|
|
94
|
|
|
95 ## MDS plots {.tabset}
|
|
|
96
|
|
|
97 ### Using rlog-transformed values
|
|
|
98
|
|
|
99 ```{r}
|
|
|
100 mds <- as.data.frame(colData(rld)) %>%
|
|
|
101 cbind(cmdscale(sampleDistMatrix))
|
|
|
102 mds
|
|
|
103 ggplot(mds, aes(x = `1`, y = `2`, col = time)) +
|
|
|
104 geom_point(size = 3) + coord_fixed()
|
|
|
105 ```
|
|
|
106
|
|
|
107 ### Using the *Poisson Distance*
|
|
|
108
|
|
|
109 ```{r}
|
|
|
110 mdsPois <- as.data.frame(colData(dds)) %>%
|
|
|
111 cbind(cmdscale(samplePoisDistMatrix))
|
|
|
112 ggplot(mdsPois, aes(x = `1`, y = `2`, col = time)) +
|
|
|
113 geom_point(size = 3) + coord_fixed()
|
|
|
114 ```
|
