|
0
|
1 #!/usr/bin/perl -w
|
|
|
2
|
|
|
3 # This program draws, in a pdf file, a stacked bars plot for different categories of data and for
|
|
|
4 # different criteria. For each criterion a stacked bar is drawn, such that the height of each stacked
|
|
|
5 # sub-bar represents the number of elements in each category satisfying that criterion.
|
|
|
6 # The input consists of a TABULAR format file, where the left column represents the names of categories
|
|
|
7 # and the other columns are headed by the names of criteria, such that each data value in the file
|
|
|
8 # represents the number of elements in a certain category satisfying a certain criterion.
|
|
|
9 # The output is a PDF file containing a stacked bars plot representing the number of elements in each
|
|
|
10 # category satisfying each criterion. The drawing is done using R code.
|
|
|
11
|
|
|
12
|
|
|
13 use strict;
|
|
|
14 use warnings;
|
|
|
15
|
|
|
16 my $criterion;
|
|
|
17 my @criteriaArray = ();
|
|
|
18 my $criteriaNumber = 0;
|
|
|
19 my $lineCounter = 0;
|
|
|
20
|
|
|
21 #variable to store the names of R script file
|
|
|
22 my $r_script;
|
|
|
23
|
|
|
24 # check to make sure having correct files
|
|
|
25 my $usage = "usage: draw_stacked_bar_plot.pl [TABULAR.in] [PDF.out] \n";
|
|
|
26 die $usage unless @ARGV == 2;
|
|
|
27
|
|
|
28 my $categoriesInputFile = $ARGV[0];
|
|
|
29
|
|
|
30 my $categories_criteria_bars_plot_outputFile = $ARGV[1];
|
|
|
31
|
|
|
32 #open the input file
|
|
|
33 open (INPUT, "<", $categoriesInputFile) || die("Could not open file $categoriesInputFile \n");
|
|
|
34 open (OUTPUT, ">", $categories_criteria_bars_plot_outputFile) || die("Could not open file $categories_criteria_bars_plot_outputFile \n");
|
|
|
35
|
|
|
36 # R script to implement the drawing of a stacked bar plot representing thes significant motifs in each category of motifs
|
|
|
37 #construct an R script file
|
|
|
38 $r_script = "motif_significance_bar_plot.r";
|
|
|
39 open(Rcmd,">", $r_script) or die "Cannot open $r_script \n\n";
|
|
|
40 print Rcmd "
|
|
|
41 #store the table content of the first file into a matrix
|
|
|
42 categoriesTable <- read.table(\"$categoriesInputFile\", header = TRUE);
|
|
|
43 categoriesMatrix <- as.matrix(categoriesTable);
|
|
|
44
|
|
|
45
|
|
|
46 #compute the sum of elements in the column with the maximum sum in each matrix
|
|
|
47 columnSumsVector <- colSums(categoriesMatrix);
|
|
|
48 maxColumn <- max (columnSumsVector);
|
|
|
49
|
|
|
50 if (maxColumn %% 10 != 0){
|
|
|
51 maxColumn <- maxColumn + 10;
|
|
|
52 }
|
|
|
53
|
|
|
54 plotHeight = maxColumn/8;
|
|
|
55 criteriaVector <- names(categoriesTable);
|
|
|
56
|
|
|
57 pdf(file = \"$categories_criteria_bars_plot_outputFile\", width = length(criteriaVector), height = plotHeight, family = \"Times\", pointsize = 12, onefile = TRUE);
|
|
|
58
|
|
|
59
|
|
|
60
|
|
|
61 #draw the first barplot
|
|
|
62 barplot(categoriesMatrix, ylab = \"No. of elements in each category\", xlab = \"Criteria\", ylim = range(0, maxColumn), col = \"black\", density = c(10, 20, 30, 40, 50, 60, 70, 80), angle = c(45, 90, 135), names.arg = criteriaVector);
|
|
|
63
|
|
|
64 #draw the legend
|
|
|
65 legendX = 0.2;
|
|
|
66 legendY = maxColumn;
|
|
|
67
|
|
|
68 legend (legendX, legendY, legend = rownames(categoriesMatrix), density = c(10, 20, 30, 40, 50, 60, 70, 80), angle = c(45, 90, 135));
|
|
|
69
|
|
|
70 dev.off();
|
|
|
71
|
|
|
72 #eof\n";
|
|
|
73 close Rcmd;
|
|
|
74 system("R --no-restore --no-save --no-readline < $r_script > $r_script.out");
|
|
|
75
|
|
|
76 #close the input files
|
|
|
77 close(OUTPUT);
|
|
|
78 close(INPUT);
|
