|
7
|
1 ---
|
|
|
2 title: 'DESeq2: Perform DESeq analysis'
|
|
|
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$set(
|
|
|
13 echo = ECHO
|
|
|
14 )
|
|
|
15
|
|
|
16 library(stringi)
|
|
|
17 library(DESeq2)
|
|
|
18 library(pheatmap)
|
|
|
19 library(PoiClaClu)
|
|
|
20 library(RColorBrewer)
|
|
|
21 ```
|
|
|
22
|
|
|
23 # `DESeqDataSet` object
|
|
|
24
|
|
|
25 ```{r}
|
|
|
26 count_files = strsplit(opt$count_files, ',')[[1]]
|
|
|
27 sample_table = read.table(opt$sample_table, header = TRUE)
|
|
|
28
|
|
|
29 ## copy count files into working directory
|
|
|
30 file_copy = file.copy(count_files, sample_table$fileName, overwrite = TRUE)
|
|
|
31
|
|
|
32 ## DESeqDataSet object
|
|
|
33 dds = DESeqDataSetFromHTSeqCount(sampleTable = sample_table,
|
|
|
34 directory = './',
|
|
|
35 design = DESIGN_FORMULA)
|
|
|
36 dds
|
|
|
37 ```
|
|
|
38
|
|
|
39 # Pre-filtering the dataset.
|
|
|
40
|
|
|
41 We can remove the rows that have 0 or 1 count to reduce object size and increase the calculation speed.
|
|
|
42
|
|
|
43 * Number of rows before pre-filtering
|
|
|
44 ```{r}
|
|
|
45 nrow(dds)
|
|
|
46 ```
|
|
|
47
|
|
|
48 * Number of rows after pre-filtering
|
|
|
49 ```{r}
|
|
|
50 dds = dds[rowSums(counts(dds)) > 1, ]
|
|
|
51 nrow(dds)
|
|
|
52 ```
|
|
|
53
|
|
|
54 # Peek at data {.tabset}
|
|
|
55
|
|
|
56 ## Count Data
|
|
|
57
|
|
|
58 ```{r}
|
|
|
59 datatable(head(counts(dds), 100), style="bootstrap",
|
|
|
60 class="table-condensed", options = list(dom = 'tp', scrollX = TRUE))
|
|
|
61 ```
|
|
|
62
|
|
|
63 ## Sample Table
|
|
|
64
|
|
|
65 ```{r}
|
|
|
66 datatable(sample_table, style="bootstrap",
|
|
|
67 class="table-condensed", options = list(dom = 'tp', scrollX = TRUE))
|
|
|
68 ```
|
|
|
69
|
|
|
70 # Sample distance on variance stabilized data {.tabset}
|
|
|
71
|
|
|
72 ## `rlog` Stabilizing transformation
|
|
|
73
|
|
|
74 ```{r}
|
|
|
75 rld = rlog(dds, blind = FALSE)
|
|
|
76 datatable(head(assay(rld), 100), style="bootstrap",
|
|
|
77 class="table-condensed", options = list(dom = 'tp', scrollX = TRUE))
|
|
|
78 ```
|
|
|
79
|
|
|
80 ## Sample distance
|
|
|
81
|
|
|
82 ```{r}
|
|
|
83 sampleDists <- dist(t(assay(rld)))
|
|
|
84 sampleDists
|
|
|
85 ```
|
|
|
86
|
|
|
87 # Differential expression analysis
|
|
|
88
|
|
|
89 ```{r}
|
|
|
90 dds <- DESeq(dds)
|
|
|
91 ```
|
|
|
92
|
|
|
93 ```{r}
|
|
|
94 rm("opt")
|
|
|
95 save(list=ls(all.names = TRUE), file='DESEQ_WORKSPACE')
|
|
|
96 ```
|
|
|
97
|
|
|
98
|
