|
1
|
1 ---
|
|
|
2 output: html_document
|
|
|
3 ---
|
|
|
4
|
|
|
5 ```{r setup, include=FALSE, warning=FALSE, message=FALSE}
|
|
|
6 knitr::opts_chunk$set(
|
|
|
7 echo = as.logical(opt$X_e),
|
|
|
8 error = TRUE,
|
|
|
9 eval = TRUE
|
|
|
10 )
|
|
|
11 ```
|
|
|
12
|
|
|
13
|
|
|
14 ```{r 'function definition', echo=FALSE}
|
|
|
15 # Define a function to extract outputs for each module from fastqc output
|
|
|
16 extract_data_module = function(fastqc_data, module_name, header = TRUE, comment.char = "") {
|
|
|
17 f = readLines(fastqc_data)
|
|
|
18 start_line = grep(module_name, f)
|
|
|
19 end_module_lines = grep('END_MODULE', f)
|
|
|
20 end_line = end_module_lines[which(end_module_lines > start_line)[1]]
|
|
|
21 module_data = f[(start_line+1):(end_line-1)]
|
|
|
22 writeLines(module_data, '/tmp/temp.txt')
|
|
|
23 read.csv('/tmp/temp.txt', sep = '\t', header = header, comment.char = comment.char)
|
|
|
24 }
|
|
|
25 ```
|
|
|
26
|
|
|
27 # Per base sequence quality
|
|
|
28
|
|
|
29 ```{r 'per base sequence quality', fig.width=10}
|
|
|
30 ## reads 1
|
|
|
31 pbsq_1 = extract_data_module(paste0(opt$X_d, '/read_1_fastqc/fastqc_data.txt'), 'Per base sequence quality')
|
|
|
32 pbsq_1$id = 1:length(pbsq_1$X.Base)
|
|
|
33 pbsq_1$trim = 'before'
|
|
|
34
|
|
|
35 ## reads 2
|
|
|
36 pbsq_2 = extract_data_module(paste0(opt$X_d, '/read_2_fastqc/fastqc_data.txt'), 'Per base sequence quality')
|
|
|
37 pbsq_2$id = 1:length(pbsq_2$X.Base)
|
|
|
38 pbsq_2$trim = 'after'
|
|
|
39
|
|
|
40 comb_pbsq = rbind(pbsq_1, pbsq_2)
|
|
|
41 comb_pbsq$trim = factor(levels = c('before', 'after'), comb_pbsq$trim)
|
|
|
42
|
|
|
43 p = ggplot(data = comb_pbsq) +
|
|
|
44 geom_boxplot(mapping = aes(x = id,
|
|
|
45 lower = Lower.Quartile,
|
|
|
46 upper = Upper.Quartile,
|
|
|
47 middle = Median,
|
|
|
48 ymin = X10th.Percentile,
|
|
|
49 ymax = X90th.Percentile,
|
|
|
50 fill = "yellow"),
|
|
|
51 stat = 'identity') +
|
|
|
52 geom_line(mapping = aes(x = id, y = Mean, color = "red")) +
|
|
|
53 scale_x_continuous(breaks = pbsq_2$id, labels = pbsq_2$X.Base) +
|
|
|
54 scale_fill_identity() +
|
|
|
55 scale_color_identity() +
|
|
|
56 ylim(0, max(comb_pbsq$Upper.Quartile) + 5) +
|
|
|
57 xlab('Position in read (bp)') +
|
|
|
58 facet_grid(. ~ trim) +
|
|
|
59 theme(axis.text.x = element_text(angle=45))
|
|
|
60 p
|
|
|
61
|
|
|
62 ```
|
