annotate DESeq.Rmd @ 6:2f8ddef8d545 draft

update deseq2

---
title: 'DESeq2: Perform DESeq analysis'
output:
    html_document:
      number_sections: true
      toc: true
      theme: cosmo
      highlight: tango
---

```{r setup, include=FALSE, warning=FALSE, message=FALSE}
knitr::opts_chunk$set(
  echo = ECHO,
  error = TRUE
)
```

# `DESeqDataSet` object

```{r 'DESeqDataSet object'}
count_file_paths = strsplit(opt$count_file_paths, ',')[[1]]
count_file_names = strsplit(opt$count_file_names, ',')[[1]]
sample_table = read.table(opt$sample_table, header = TRUE)
row.names(sample_table) = sample_table[,2]
sample_table = sample_table[count_file_names, ]

## copy count files into working directory
file_copy = file.copy(count_file_paths, count_file_names, overwrite = TRUE)

## DESeqDataSet object
dds = DESeqDataSetFromHTSeqCount(sampleTable = sample_table,
                                 directory = './',
                                 design = DESIGN_FORMULA)
dds
```

# Pre-filtering the dataset.

We can remove the rows that have 0 or 1 count to reduce object size and increase the calculation speed.

* Number of rows before pre-filtering
```{r}
nrow(dds)
```

* Number of rows after pre-filtering
```{r}
dds = dds[rowSums(counts(dds)) > 1, ]
nrow(dds)
```

# Peek at data {.tabset}

## Count Data

```{r 'count data'}
datatable(head(counts(dds), 100), style="bootstrap", 
          class="table-condensed", options = list(dom = 'tp', scrollX = TRUE))
```

## Sample Table 

```{r 'sample table'}
datatable(sample_table, style="bootstrap",
          class="table-condensed", options = list(dom = 'tp', scrollX = TRUE))
```

# Sample distance on variance stabilized data {.tabset}

## `rlog` Stabilizing transformation

```{r}
rld = rlog(dds, blind = FALSE)
datatable(head(assay(rld), 100), style="bootstrap", 
          class="table-condensed", options = list(dom = 'tp', scrollX = TRUE))
```

## Sample distance

```{r}
sampleDists <- dist(t(assay(rld)))
sampleDists
```

# Differential expression analysis

```{r}
dds <- DESeq(dds)
```

```{r}
rm("opt")
save(list=ls(all.names = TRUE), file='DESEQ_WORKSPACE')
```