[2] | 1 | #!/usr/bin/perl -w |
---|
| 2 | |
---|
| 3 | # The program takes as input a set of categories, such that each category contains many elements. |
---|
| 4 | # It also takes a table relating elements with criteria, such that each element is assigned a number |
---|
| 5 | # representing the number of times the element satisfies a certain criterion. |
---|
| 6 | # The first input is a TABULAR format file, such that the left column represents the name of categories and, |
---|
| 7 | # all other columns represent the names of elements. |
---|
| 8 | # The second input is a TABULAR format file relating elements with criteria, such that the first line |
---|
| 9 | # represents the names of criteria and the left column represents the names of elements. |
---|
| 10 | # The output is a TABULAR format file relating catergories with criteria, such that each categoy is |
---|
| 11 | # assigned a number representing the total number of times its elements satisfies a certain criterion. |
---|
| 12 | # Each category is assigned as many numbers as criteria. |
---|
| 13 | |
---|
| 14 | use strict; |
---|
| 15 | use warnings; |
---|
| 16 | |
---|
| 17 | #variables to handle information of the categories input file |
---|
| 18 | my @categoryElementsArray = (); |
---|
| 19 | my @categoriesArray = (); |
---|
| 20 | my $categoryMemberNames; |
---|
| 21 | my $categoryName; |
---|
| 22 | my %categoryMembersHash = (); |
---|
| 23 | my $memberNumber = 0; |
---|
| 24 | my $totalMembersNumber = 0; |
---|
| 25 | my $totalCategoriesNumber = 0; |
---|
| 26 | my @categoryCountersTwoDimArray = (); |
---|
| 27 | my $lineCounter1 = 0; |
---|
| 28 | |
---|
| 29 | #variables to handle information of the criteria and elements data input file |
---|
| 30 | my $elementLine; |
---|
| 31 | my @elementDataArray = (); |
---|
| 32 | my $elementName; |
---|
| 33 | my @criteriaArray = (); |
---|
| 34 | my $criteriaNumber = 0; |
---|
| 35 | my $totalCriteriaNumber = 0; |
---|
| 36 | my $lineCounter2 = 0; |
---|
| 37 | |
---|
| 38 | #variable representing the row and column indices used to store results into a two-dimensional array |
---|
| 39 | my $row = 0; |
---|
| 40 | my $column = 0; |
---|
| 41 | |
---|
| 42 | # check to make sure having correct files |
---|
| 43 | my $usage = "usage: categorize_motifs_significance.pl [TABULAR.in] [TABULAR.in] [TABULAR.out] \n"; |
---|
| 44 | die $usage unless @ARGV == 3; |
---|
| 45 | |
---|
| 46 | #get the categories input file |
---|
| 47 | my $categories_inputFile = $ARGV[0]; |
---|
| 48 | |
---|
| 49 | #get the criteria and data input file |
---|
| 50 | my $elements_data_inputFile = $ARGV[1]; |
---|
| 51 | |
---|
| 52 | #get the output file |
---|
| 53 | my $categorized_data_outputFile = $ARGV[2]; |
---|
| 54 | |
---|
| 55 | #open the input and output files |
---|
| 56 | open (INPUT1, "<", $categories_inputFile) || die("Could not open file $categories_inputFile \n"); |
---|
| 57 | open (INPUT2, "<", $elements_data_inputFile ) || die("Could not open file $elements_data_inputFile \n"); |
---|
| 58 | open (OUTPUT, ">", $categorized_data_outputFile) || die("Could not open file $categorized_data_outputFile \n"); |
---|
| 59 | |
---|
| 60 | #store the first input file into an array |
---|
| 61 | my @categoriesData = <INPUT1>; |
---|
| 62 | |
---|
| 63 | #reset the value of $lineCounter1 to 0 |
---|
| 64 | $lineCounter1 = 0; |
---|
| 65 | |
---|
| 66 | #iterate through the first input file to get the names of categories and their corresponding elements |
---|
| 67 | foreach $categoryMemberNames (@categoriesData){ |
---|
| 68 | chomp ($categoryMemberNames); |
---|
| 69 | |
---|
| 70 | @categoryElementsArray = split(/\t/, $categoryMemberNames); |
---|
| 71 | |
---|
| 72 | #store the name of the current category into an array |
---|
| 73 | $categoriesArray [$lineCounter1] = $categoryElementsArray[0]; |
---|
| 74 | |
---|
| 75 | #store the name of the current category into a two-dimensional array |
---|
| 76 | $categoryCountersTwoDimArray [$lineCounter1] [0] = $categoryElementsArray[0]; |
---|
| 77 | |
---|
| 78 | #get the total number of elements in the current category |
---|
| 79 | $totalMembersNumber = @categoryElementsArray; |
---|
| 80 | |
---|
| 81 | #store the names of categories and their corresponding elements into a hash |
---|
| 82 | for ($memberNumber = 1; $memberNumber < $totalMembersNumber; $memberNumber++) { |
---|
| 83 | |
---|
| 84 | $categoryMembersHash{$categoryElementsArray[$memberNumber]} = $categoriesArray[$lineCounter1]; |
---|
| 85 | } |
---|
| 86 | |
---|
| 87 | $lineCounter1++; |
---|
| 88 | } |
---|
| 89 | |
---|
| 90 | #store the second input file into an array |
---|
| 91 | my @elementsData = <INPUT2>; |
---|
| 92 | |
---|
| 93 | #reset the value of $lineCounter2 to 0 |
---|
| 94 | $lineCounter2 = 0; |
---|
| 95 | |
---|
| 96 | #iterate through the second input file in order to count the number of elements |
---|
| 97 | #in each category that satisfy each criterion |
---|
| 98 | foreach $elementLine (@elementsData){ |
---|
| 99 | chomp ($elementLine); |
---|
| 100 | |
---|
| 101 | $lineCounter2++; |
---|
| 102 | |
---|
| 103 | @elementDataArray = split(/\t/, $elementLine); |
---|
| 104 | |
---|
| 105 | #if at the first line, get the total number of criteria and the total |
---|
| 106 | #number of catergories and initialize the two-dimensional array |
---|
| 107 | if ($lineCounter2 == 1){ |
---|
| 108 | @criteriaArray = @elementDataArray; |
---|
| 109 | $totalCriteriaNumber = @elementDataArray; |
---|
| 110 | |
---|
| 111 | $totalCategoriesNumber = @categoriesArray; |
---|
| 112 | |
---|
| 113 | #initialize the two-dimensional array |
---|
| 114 | for ($row = 0; $row < $totalCategoriesNumber; $row++) { |
---|
| 115 | |
---|
| 116 | for ($column = 1; $column <= $totalCriteriaNumber; $column++) { |
---|
| 117 | |
---|
| 118 | $categoryCountersTwoDimArray [$row][$column] = 0; |
---|
| 119 | } |
---|
| 120 | } |
---|
| 121 | } |
---|
| 122 | else{ |
---|
| 123 | #get the element data |
---|
| 124 | $elementName = $elementDataArray[0]; |
---|
| 125 | |
---|
| 126 | #do the counting and store the result in the two-dimensional array |
---|
| 127 | for ($criteriaNumber = 0; $criteriaNumber < $totalCriteriaNumber; $criteriaNumber++) { |
---|
| 128 | |
---|
| 129 | if ($elementDataArray[$criteriaNumber + 1] > 0){ |
---|
| 130 | |
---|
| 131 | $categoryName = $categoryMembersHash{$elementName}; |
---|
| 132 | |
---|
| 133 | my ($categoryIndex) = grep $categoriesArray[$_] eq $categoryName, 0 .. $#categoriesArray; |
---|
| 134 | |
---|
| 135 | $categoryCountersTwoDimArray [$categoryIndex] [$criteriaNumber + 1] = $categoryCountersTwoDimArray [$categoryIndex] [$criteriaNumber + 1] + $elementDataArray[$criteriaNumber + 1]; |
---|
| 136 | } |
---|
| 137 | } |
---|
| 138 | } |
---|
| 139 | } |
---|
| 140 | |
---|
| 141 | print OUTPUT "\t"; |
---|
| 142 | |
---|
| 143 | #store the criteria names into the output file |
---|
| 144 | for ($column = 1; $column <= $totalCriteriaNumber; $column++) { |
---|
| 145 | |
---|
| 146 | if ($column < $totalCriteriaNumber){ |
---|
| 147 | print OUTPUT $criteriaArray[$column - 1] . "\t"; |
---|
| 148 | } |
---|
| 149 | else{ |
---|
| 150 | print OUTPUT $criteriaArray[$column - 1] . "\n"; |
---|
| 151 | } |
---|
| 152 | } |
---|
| 153 | |
---|
| 154 | #store the category names and their corresponding number of elements satisfying criteria into the output file |
---|
| 155 | for ($row = 0; $row < $totalCategoriesNumber; $row++) { |
---|
| 156 | |
---|
| 157 | for ($column = 0; $column <= $totalCriteriaNumber; $column++) { |
---|
| 158 | |
---|
| 159 | if ($column < $totalCriteriaNumber){ |
---|
| 160 | print OUTPUT $categoryCountersTwoDimArray [$row][$column] . "\t"; |
---|
| 161 | } |
---|
| 162 | else{ |
---|
| 163 | print OUTPUT $categoryCountersTwoDimArray [$row][$column] . "\n"; |
---|
| 164 | } |
---|
| 165 | } |
---|
| 166 | } |
---|
| 167 | |
---|
| 168 | #close the input and output file |
---|
| 169 | close(OUTPUT); |
---|
| 170 | close(INPUT2); |
---|
| 171 | close(INPUT1); |
---|
| 172 | |
---|