macroape 3.2.2 → 3.3.0
Sign up to get free protection for your applications and to get access to all the features.
- data/Rakefile +28 -7
- data/lib/macroape.rb +14 -26
- data/lib/macroape/aligned_pair_intersection.rb +24 -24
- data/lib/macroape/collection.rb +1 -2
- data/lib/macroape/count_by_threshold.rb +8 -26
- data/lib/macroape/exec/eval_alignment.rb +19 -19
- data/lib/macroape/exec/eval_similarity.rb +18 -16
- data/lib/macroape/exec/find_pvalue.rb +8 -6
- data/lib/macroape/exec/find_threshold.rb +7 -5
- data/lib/macroape/exec/preprocess_collection.rb +10 -7
- data/lib/macroape/exec/scan_collection.rb +13 -10
- data/lib/macroape/pwm_compare.rb +33 -2
- data/lib/macroape/pwm_compare_aligned.rb +38 -2
- data/lib/macroape/threshold_by_pvalue.rb +48 -43
- data/lib/macroape/version.rb +3 -3
- data/macroape.gemspec +2 -0
- data/test/data/test_collection.yaml +70 -4
- data/test/eval_alignment_similarity_test.rb +19 -0
- data/test/eval_similarity_test.rb +26 -0
- data/test/find_pvalue_test.rb +25 -0
- data/test/find_threshold_test.rb +29 -0
- data/test/preprocess_collection_test.rb +9 -0
- data/test/scan_collection_test.rb +17 -0
- data/test/test_helper.rb +10 -0
- metadata +33 -16
- data/lib/macroape/aligned_pair_metrics.rb +0 -24
- data/lib/macroape/aligned_pair_transformations.rb +0 -23
- data/lib/macroape/extract_pwm.rb +0 -32
- data/lib/macroape/gauss_estimation.rb +0 -30
- data/lib/macroape/matrix_information.rb +0 -29
- data/lib/macroape/matrix_on_background.rb +0 -16
- data/lib/macroape/matrix_transformations.rb +0 -29
- data/lib/macroape/pair_metrics.rb +0 -9
- data/lib/macroape/pair_transformations.rb +0 -28
- data/lib/macroape/single_matrix.rb +0 -45
- data/lib/macroape/support.rb +0 -34
- data/test/macroape_test.rb +0 -125
data/Rakefile
CHANGED
@@ -1,7 +1,28 @@
|
|
1
|
-
#!/usr/bin/env rake
|
2
|
-
require "bundler/gem_tasks"
|
3
|
-
|
4
|
-
|
5
|
-
task :
|
6
|
-
|
7
|
-
end
|
1
|
+
#!/usr/bin/env rake
|
2
|
+
require "bundler/gem_tasks"
|
3
|
+
|
4
|
+
namespace :spec do
|
5
|
+
task :find_threshold do
|
6
|
+
system("ruby -I ./test test/find_threshold_test.rb")
|
7
|
+
end
|
8
|
+
task :find_pvalue do
|
9
|
+
system("ruby -I ./test test/find_pvalue_test.rb")
|
10
|
+
end
|
11
|
+
task :eval_similarity do
|
12
|
+
system("ruby -I ./test test/eval_similarity_test.rb")
|
13
|
+
end
|
14
|
+
task :eval_alignment_similarity do
|
15
|
+
system("ruby -I ./test test/eval_alignment_similarity_test.rb")
|
16
|
+
end
|
17
|
+
task :preprocess_collection do
|
18
|
+
system("ruby -I ./test test/preprocess_collection_test.rb")
|
19
|
+
end
|
20
|
+
task :scan_collection do
|
21
|
+
system("ruby -I ./test test/scan_collection_test.rb")
|
22
|
+
end
|
23
|
+
task :all => [:find_threshold, :find_pvalue, :eval_similarity,
|
24
|
+
:eval_alignment_similarity, :scan_collection, :preprocess_collection]
|
25
|
+
end
|
26
|
+
|
27
|
+
desc 'Test all functionality of gem executables'
|
28
|
+
task :spec => ['spec:all']
|
data/lib/macroape.rb
CHANGED
@@ -1,26 +1,14 @@
|
|
1
|
-
require 'macroape/version'
|
2
|
-
|
3
|
-
|
4
|
-
require 'macroape/
|
5
|
-
require 'macroape/
|
6
|
-
|
7
|
-
require 'macroape/
|
8
|
-
require 'macroape/
|
9
|
-
require 'macroape/
|
10
|
-
require 'macroape/
|
11
|
-
|
12
|
-
|
13
|
-
|
14
|
-
|
15
|
-
require 'macroape/aligned_pair_metrics'
|
16
|
-
require 'macroape/aligned_pair_intersection'
|
17
|
-
require 'macroape/pwm_compare_aligned'
|
18
|
-
|
19
|
-
require 'macroape/pair_transformations'
|
20
|
-
require 'macroape/pair_metrics'
|
21
|
-
require 'macroape/pwm_compare'
|
22
|
-
require 'macroape/collection'
|
23
|
-
|
24
|
-
module Macroape
|
25
|
-
# Your code goes here...
|
26
|
-
end
|
1
|
+
require 'macroape/version'
|
2
|
+
|
3
|
+
require 'bioinform'
|
4
|
+
require 'macroape/threshold_by_pvalue'
|
5
|
+
require 'macroape/count_by_threshold'
|
6
|
+
|
7
|
+
require 'macroape/aligned_pair_intersection'
|
8
|
+
require 'macroape/pwm_compare_aligned'
|
9
|
+
require 'macroape/pwm_compare'
|
10
|
+
require 'macroape/collection'
|
11
|
+
|
12
|
+
module Macroape
|
13
|
+
# Your code goes here...
|
14
|
+
end
|
@@ -1,9 +1,9 @@
|
|
1
|
-
module
|
2
|
-
|
1
|
+
module Macroape
|
2
|
+
class PWMCompareAligned
|
3
3
|
|
4
4
|
def counts_for_two_matrices(threshold_first, threshold_second)
|
5
|
-
if first.
|
6
|
-
if first.
|
5
|
+
if first.background == second.background
|
6
|
+
if first.background == [1,1,1,1]
|
7
7
|
common_words_for_two_matrices(threshold_first, threshold_second)
|
8
8
|
else
|
9
9
|
counts_for_two_matrices_with_same_probabilities(threshold_first, threshold_second)
|
@@ -18,8 +18,8 @@ module PWMCompare
|
|
18
18
|
result_first = 0.0
|
19
19
|
result_second = 0.0
|
20
20
|
length.times do |column|
|
21
|
-
ending_weight_first = first.
|
22
|
-
ending_weight_second = second.
|
21
|
+
ending_weight_first = first.background_sum ** (length - column - 1)
|
22
|
+
ending_weight_second = second.background_sum ** (length - column - 1)
|
23
23
|
already_enough_first = threshold_first - first.worst_suffix[column + 1]
|
24
24
|
already_enough_second = threshold_second - second.worst_suffix[column + 1]
|
25
25
|
least_sufficient_first = threshold_first - first.best_suffix[column + 1]
|
@@ -33,23 +33,23 @@ module PWMCompare
|
|
33
33
|
if new_score_first >= already_enough_first
|
34
34
|
new_score_second = score_second + second.matrix[column][letter]
|
35
35
|
if new_score_second >= already_enough_second
|
36
|
-
result_first += count[0] * first.
|
37
|
-
result_second += count[1] * second.
|
36
|
+
result_first += count[0] * first.background[letter] * ending_weight_first
|
37
|
+
result_second += count[1] * second.background[letter] * ending_weight_second
|
38
38
|
elsif new_score_second >= least_sufficient_second
|
39
|
-
new_scores[new_score_first][new_score_second][0] += count[0] * first.
|
40
|
-
new_scores[new_score_first][new_score_second][1] += count[1] * second.
|
39
|
+
new_scores[new_score_first][new_score_second][0] += count[0] * first.background[letter]
|
40
|
+
new_scores[new_score_first][new_score_second][1] += count[1] * second.background[letter]
|
41
41
|
end
|
42
42
|
elsif new_score_first >= least_sufficient_first
|
43
43
|
new_score_second = score_second + second.matrix[column][letter]
|
44
44
|
if new_score_second >= least_sufficient_second
|
45
|
-
new_scores[new_score_first][new_score_second][0] += count[0] * first.
|
46
|
-
new_scores[new_score_first][new_score_second][1] += count[1] * second.
|
45
|
+
new_scores[new_score_first][new_score_second][0] += count[0] * first.background[letter]
|
46
|
+
new_scores[new_score_first][new_score_second][1] += count[1] * second.background[letter]
|
47
47
|
end
|
48
48
|
end
|
49
49
|
end
|
50
50
|
end
|
51
51
|
end
|
52
|
-
raise 'Hash overflow in
|
52
|
+
raise 'Hash overflow in Macroape::AlignedPairIntersection#counts_for_two_matrices_with_different_probabilities' if new_scores.inject(0){|sum,hsh|sum+hsh.size} > MaxHashSizeDouble
|
53
53
|
scores = new_scores
|
54
54
|
end
|
55
55
|
[result_first, result_second]
|
@@ -58,9 +58,9 @@ module PWMCompare
|
|
58
58
|
def counts_for_two_matrices_with_same_probabilities(threshold_first, threshold_second)
|
59
59
|
scores = { 0 => {0 => 1} } # scores_on_first_pwm, scores_on_second_pwm --> count_on_first_probabilities, count_on_second_probabilities
|
60
60
|
result = 0.0
|
61
|
-
|
61
|
+
background = first.background
|
62
62
|
length.times do |column|
|
63
|
-
ending_weight = first.
|
63
|
+
ending_weight = first.background_sum ** (length - column - 1)
|
64
64
|
already_enough_first = threshold_first - first.worst_suffix[column + 1]
|
65
65
|
already_enough_second = threshold_second - second.worst_suffix[column + 1]
|
66
66
|
least_sufficient_first = threshold_first - first.best_suffix[column + 1]
|
@@ -74,20 +74,20 @@ module PWMCompare
|
|
74
74
|
if new_score_first >= already_enough_first
|
75
75
|
new_score_second = score_second + second.matrix[column][letter]
|
76
76
|
if new_score_second >= already_enough_second
|
77
|
-
result += count *
|
77
|
+
result += count * background[letter] * ending_weight
|
78
78
|
elsif new_score_second >= least_sufficient_second
|
79
|
-
new_scores[new_score_first][new_score_second] += count *
|
79
|
+
new_scores[new_score_first][new_score_second] += count * background[letter]
|
80
80
|
end
|
81
81
|
elsif new_score_first >= least_sufficient_first
|
82
82
|
new_score_second = score_second + second.matrix[column][letter]
|
83
83
|
if new_score_second >= least_sufficient_second
|
84
|
-
new_scores[new_score_first][new_score_second] += count *
|
84
|
+
new_scores[new_score_first][new_score_second] += count * background[letter]
|
85
85
|
end
|
86
86
|
end
|
87
87
|
end
|
88
88
|
end
|
89
89
|
end
|
90
|
-
raise 'Hash overflow in
|
90
|
+
raise 'Hash overflow in Macroape::AlignedPairIntersection#counts_for_two_matrices_with_same_probabilities' if new_scores.inject(0){|sum,hsh|sum+hsh.size} > MaxHashSizeDouble
|
91
91
|
scores = new_scores
|
92
92
|
end
|
93
93
|
[result, result]
|
@@ -97,7 +97,7 @@ module PWMCompare
|
|
97
97
|
def common_words_for_two_matrices(threshold_first, threshold_second)
|
98
98
|
scores = { 0 => {0 => 1} } # scores_on_first_pwm, scores_on_second_pwm --> count_on_first_probabilities, count_on_second_probabilities
|
99
99
|
result = 0
|
100
|
-
length.times do |column|
|
100
|
+
length.times do |column|
|
101
101
|
ending_weight = 4 ** (length - column - 1)
|
102
102
|
already_enough_first = threshold_first - first.worst_suffix[column + 1]
|
103
103
|
already_enough_second = threshold_second - second.worst_suffix[column + 1]
|
@@ -107,13 +107,13 @@ module PWMCompare
|
|
107
107
|
new_scores = Hash.new{|h,k| h[k]=Hash.new{|h2,k2| h2[k2]=0} }
|
108
108
|
scores.each do |score_first, second_scores|
|
109
109
|
second_scores.each do |score_second, count|
|
110
|
-
4.times do |letter|
|
110
|
+
4.times do |letter|
|
111
111
|
new_score_first = score_first + first.matrix[column][letter]
|
112
|
-
if new_score_first >= already_enough_first
|
112
|
+
if new_score_first >= already_enough_first
|
113
113
|
new_score_second = score_second + second.matrix[column][letter]
|
114
114
|
if new_score_second >= already_enough_second
|
115
115
|
result += count * ending_weight
|
116
|
-
elsif new_score_second >= least_sufficient_second
|
116
|
+
elsif new_score_second >= least_sufficient_second
|
117
117
|
new_scores[new_score_first][new_score_second] += count
|
118
118
|
end
|
119
119
|
elsif new_score_first >= least_sufficient_first
|
@@ -126,7 +126,7 @@ module PWMCompare
|
|
126
126
|
end
|
127
127
|
end
|
128
128
|
|
129
|
-
raise 'Hash overflow in
|
129
|
+
raise 'Hash overflow in Macroape::AlignedPairIntersection#common_words_for_two_matrices' if defined? MaxHashSizeDouble and new_scores.inject(0){|sum,hsh|sum+hsh.size} > MaxHashSizeDouble
|
130
130
|
scores = new_scores
|
131
131
|
end
|
132
132
|
[result, result]
|
data/lib/macroape/collection.rb
CHANGED
@@ -1,4 +1,4 @@
|
|
1
|
-
module
|
1
|
+
module Macroape
|
2
2
|
class Collection
|
3
3
|
attr_reader :rough_discretization, :precise_discretization, :background, :pvalues, :pwms, :infos
|
4
4
|
def initialize(rough_discretization, precise_discretization, background, pvalues)
|
@@ -7,7 +7,6 @@ module PWM
|
|
7
7
|
@infos={}
|
8
8
|
end
|
9
9
|
def add_pwm(pwm,info)
|
10
|
-
#@pwms[pwm] = info
|
11
10
|
@pwms[pwm.name] = pwm
|
12
11
|
@infos[pwm.name] = info
|
13
12
|
end
|
@@ -1,34 +1,16 @@
|
|
1
|
-
|
2
|
-
|
1
|
+
require 'macroape/threshold_by_pvalue'
|
2
|
+
|
3
|
+
module Bioinform
|
4
|
+
class PWM
|
3
5
|
def counts_by_thresholds(*thresholds)
|
4
|
-
scores =
|
5
|
-
|
6
|
-
|
7
|
-
#scores.select{|score,count| score >= threshold}.map{|score,count| count}.inject(0){|sum,val|sum+val}
|
8
|
-
scores.inject(0){|sum,(score,count)| (score >= threshold) ? sum + count : sum}
|
6
|
+
scores = count_distribution_after_threshold(thresholds.min)
|
7
|
+
thresholds.map{ |threshold|
|
8
|
+
scores.inject(0.0){|sum,(score,count)| (score >= threshold) ? sum + count : sum}
|
9
9
|
}
|
10
|
-
=end
|
11
|
-
s_thr= thresholds.map.with_index{|threshold,index|[threshold,index]}.sort_by{|threshold,index| threshold}
|
12
|
-
|
13
|
-
cnt = 0
|
14
|
-
thr_cnts=[]
|
15
|
-
|
16
|
-
scores.sort.reverse.each do |score,count|
|
17
|
-
while !s_thr.empty? and score < s_thr.last[0]
|
18
|
-
thr_cnts.push(cnt)
|
19
|
-
s_thr.pop
|
20
|
-
end
|
21
|
-
cnt += count
|
22
|
-
end
|
23
|
-
s_thr = thresholds.map.with_index{|threshold,index|[threshold,index]}.sort_by{|threshold,index| threshold}
|
24
|
-
while thr_cnts.size < s_thr.size
|
25
|
-
thr_cnts.push(cnt)
|
26
|
-
end
|
27
|
-
s_thr.reverse.zip(thr_cnts).sort_by{|(threshold,index), count| index}.map{|(threshold,index), count| count.to_f}
|
28
10
|
end
|
29
11
|
|
30
12
|
def pvalue_by_threshold(threshold)
|
31
|
-
counts_by_thresholds(threshold).first /
|
13
|
+
counts_by_thresholds(threshold).first / vocabulary_volume
|
32
14
|
end
|
33
15
|
end
|
34
16
|
end
|
@@ -24,7 +24,7 @@ Examples:
|
|
24
24
|
or in linux
|
25
25
|
cat motifs/KLF4.pat motifs/SP1.pat | ruby eval_alignment.rb .stdin .stdin 3 direct -p 0.0005 -d 100 -b 0.4 0.3 0.2 0.1
|
26
26
|
}
|
27
|
-
|
27
|
+
$:.unshift File.join(File.dirname(__FILE__),'./../../')
|
28
28
|
require 'macroape'
|
29
29
|
|
30
30
|
if ARGV.empty? or ARGV.include? '-h' or ARGV.include? '-help' or ARGV.include? '--help' or ARGV.include? '--h'
|
@@ -69,9 +69,9 @@ begin
|
|
69
69
|
when '-d'
|
70
70
|
discretization = ARGV.shift.to_f
|
71
71
|
when '-m'
|
72
|
-
|
72
|
+
Macroape::MaxHashSizeSingle = ARGV.shift.to_f
|
73
73
|
when '-md'
|
74
|
-
|
74
|
+
Macroape::MaxHashSizeDouble = ARGV.shift.to_f
|
75
75
|
when '-b'
|
76
76
|
second_background = first_background = ARGV.shift(4).map(&:to_f)
|
77
77
|
when '-b1'
|
@@ -84,35 +84,35 @@ begin
|
|
84
84
|
raise 'background should be symmetric: p(A)=p(T) and p(G) = p(C)' unless second_background == second_background.reverse
|
85
85
|
|
86
86
|
|
87
|
-
|
88
|
-
|
87
|
+
Macroape::MaxHashSizeSingle = 1000000 unless defined? Macroape::MaxHashSizeSingle
|
88
|
+
Macroape::MaxHashSizeDouble = 1000 unless defined? Macroape::MaxHashSizeDouble
|
89
89
|
|
90
|
-
if first_file == '.stdin' || second_file == '.stdin'
|
91
|
-
r_stream, w_stream = IO.pipe
|
92
|
-
STDIN.readlines.each{|line| w_stream.write(line)}
|
93
|
-
w_stream.close
|
94
|
-
end
|
90
|
+
# if first_file == '.stdin' || second_file == '.stdin'
|
91
|
+
# r_stream, w_stream = IO.pipe
|
92
|
+
# STDIN.readlines.each{|line| w_stream.write(line)}
|
93
|
+
# w_stream.close
|
94
|
+
# end
|
95
95
|
|
96
96
|
if first_file == '.stdin'
|
97
|
-
r_stream, w_stream, extracted_pwm = extract_pwm(r_stream, w_stream)
|
98
|
-
pwm_first =
|
97
|
+
# r_stream, w_stream, extracted_pwm = extract_pwm(r_stream, w_stream)
|
98
|
+
# pwm_first = Macroape::SingleMatrix.load_from_line_array(extracted_pwm).with_background(first_background).discrete(discretization)
|
99
99
|
else
|
100
100
|
raise "Error! File #{first_file} don't exist" unless File.exist?(first_file)
|
101
|
-
pwm_first = PWM
|
101
|
+
pwm_first = Bioinform::PWM.new(File.read(first_file)).background(first_background).discrete(discretization)
|
102
102
|
end
|
103
103
|
|
104
104
|
if second_file == '.stdin'
|
105
|
-
r_stream, w_stream, extracted_pwm = extract_pwm(r_stream, w_stream)
|
106
|
-
pwm_second =
|
105
|
+
# r_stream, w_stream, extracted_pwm = extract_pwm(r_stream, w_stream)
|
106
|
+
# pwm_second = Macroape::SingleMatrix.load_from_line_array(extracted_pwm).with_background(second_background).discrete(discretization)
|
107
107
|
else
|
108
108
|
raise "Error! File #{second_file} don't exist" unless File.exist?(second_file)
|
109
|
-
pwm_second = PWM
|
109
|
+
pwm_second = Bioinform::PWM.new(File.read(second_file)).background(second_background).discrete(discretization)
|
110
110
|
end
|
111
111
|
|
112
|
-
r_stream.close if first_file == '.stdin' || second_file == '.stdin'
|
112
|
+
# r_stream.close if first_file == '.stdin' || second_file == '.stdin'
|
113
113
|
|
114
114
|
|
115
|
-
pwm_second
|
115
|
+
pwm_second.reverse_complement! if reverse
|
116
116
|
|
117
117
|
first_pwm_alignment = '.' * [-shift, 0].max + '>' * pwm_first.length
|
118
118
|
second_pwm_alignment = '.' * [shift, 0].max + (orientation == :direct ? '>' : '<') * pwm_second.length
|
@@ -121,7 +121,7 @@ begin
|
|
121
121
|
(first_pwm_alignment.length...alignment_length).each{|i| first_pwm_alignment[i] = '.'}
|
122
122
|
(second_pwm_alignment.length...alignment_length).each{|i| second_pwm_alignment[i] = '.'}
|
123
123
|
|
124
|
-
cmp =
|
124
|
+
cmp = Macroape::PWMCompareAligned.new(pwm_first.left_augment([-shift,0].max),
|
125
125
|
pwm_second.left_augment([shift,0].max))
|
126
126
|
|
127
127
|
first_threshold = pwm_first.threshold(pvalue)
|
@@ -26,7 +26,9 @@ Examples:
|
|
26
26
|
cat motifs/KLF4.pat motifs/SP1.pat | ruby eval_similarity.rb .stdin .stdin -p 0.0005 -d 100 -b 0.4 0.3 0.2 0.1
|
27
27
|
}
|
28
28
|
|
29
|
+
$:.unshift File.join(File.dirname(__FILE__),'./../../')
|
29
30
|
require 'macroape'
|
31
|
+
require 'bioinform'
|
30
32
|
|
31
33
|
if ARGV.empty? or ARGV.include? '-h' or ARGV.include? '-help' or ARGV.include? '--help' or ARGV.include? '--h'
|
32
34
|
STDERR.puts help_string
|
@@ -51,9 +53,9 @@ begin
|
|
51
53
|
when '-d'
|
52
54
|
discretization = ARGV.shift.to_f
|
53
55
|
when '-m'
|
54
|
-
|
56
|
+
Macroape::MaxHashSizeSingle = ARGV.shift.to_f
|
55
57
|
when '-md'
|
56
|
-
|
58
|
+
Macroape::MaxHashSizeDouble = ARGV.shift.to_f
|
57
59
|
when '-b'
|
58
60
|
second_background = first_background = ARGV.shift(4).map(&:to_f)
|
59
61
|
when '-b1'
|
@@ -65,35 +67,35 @@ begin
|
|
65
67
|
raise 'background should be symmetric: p(A)=p(T) and p(G) = p(C)' unless first_background == first_background.reverse
|
66
68
|
raise 'background should be symmetric: p(A)=p(T) and p(G) = p(C)' unless second_background == second_background.reverse
|
67
69
|
|
68
|
-
|
69
|
-
|
70
|
+
Macroape::MaxHashSizeSingle = 1000000 unless defined? Macroape::MaxHashSizeSingle
|
71
|
+
Macroape::MaxHashSizeDouble = 1000 unless defined? Macroape::MaxHashSizeDouble
|
70
72
|
|
71
73
|
|
72
|
-
if first_file == '.stdin' || second_file == '.stdin'
|
73
|
-
r_stream, w_stream = IO.pipe
|
74
|
-
STDIN.readlines.each{|line| w_stream.write(line)}
|
75
|
-
w_stream.close
|
76
|
-
end
|
74
|
+
# if first_file == '.stdin' || second_file == '.stdin'
|
75
|
+
# r_stream, w_stream = IO.pipe
|
76
|
+
# STDIN.readlines.each{|line| w_stream.write(line)}
|
77
|
+
# w_stream.close
|
78
|
+
# end
|
77
79
|
|
78
80
|
if first_file == '.stdin'
|
79
|
-
r_stream, w_stream, extracted_pwm = extract_pwm(r_stream, w_stream)
|
80
|
-
pwm_first =
|
81
|
+
# r_stream, w_stream, extracted_pwm = extract_pwm(r_stream, w_stream)
|
82
|
+
# pwm_first = Macroape::SingleMatrix.load_from_line_array(extracted_pwm).with_background(first_background).discrete(discretization)
|
81
83
|
else
|
82
84
|
raise "Error! File #{first_file} don't exist" unless File.exist?(first_file)
|
83
|
-
pwm_first = PWM
|
85
|
+
pwm_first = Bioinform::PWM.new(File.read(first_file)).background(first_background).discrete(discretization)
|
84
86
|
end
|
85
87
|
|
86
88
|
if second_file == '.stdin'
|
87
|
-
r_stream, w_stream, extracted_pwm = extract_pwm(r_stream, w_stream)
|
88
|
-
pwm_second =
|
89
|
+
# r_stream, w_stream, extracted_pwm = extract_pwm(r_stream, w_stream)
|
90
|
+
# pwm_second = Macroape::SingleMatrix.load_from_line_array(extracted_pwm).with_background(second_background).discrete(discretization)
|
89
91
|
else
|
90
92
|
raise "Error! File #{second_file} don't exist" unless File.exist?(second_file)
|
91
|
-
pwm_second = PWM
|
93
|
+
pwm_second = Bioinform::PWM.new(File.read(second_file)).background(second_background).discrete(discretization)
|
92
94
|
end
|
93
95
|
|
94
96
|
r_stream.close if first_file == '.stdin' || second_file == '.stdin'
|
95
97
|
|
96
|
-
cmp =
|
98
|
+
cmp = Macroape::PWMCompare.new(pwm_first, pwm_second)
|
97
99
|
|
98
100
|
first_threshold = pwm_first.threshold(pvalue)
|
99
101
|
second_threshold = pwm_second.threshold(pvalue)
|
@@ -24,7 +24,9 @@ Examples:
|
|
24
24
|
cat motifs/KLF4.pat | ruby find_pvalue.rb .stdin 7.32 4.31 5.42
|
25
25
|
}
|
26
26
|
|
27
|
+
$:.unshift File.join(File.dirname(__FILE__),'./../../')
|
27
28
|
require 'macroape'
|
29
|
+
require 'bioinform'
|
28
30
|
|
29
31
|
if ARGV.empty? or ARGV.include? '-h' or ARGV.include? '-help' or ARGV.include? '--help' or ARGV.include? '--h'
|
30
32
|
STDERR.puts help_string
|
@@ -56,22 +58,22 @@ begin
|
|
56
58
|
when '-d'
|
57
59
|
discretization = ARGV.shift.to_f
|
58
60
|
when '-m'
|
59
|
-
|
61
|
+
Macroape::MaxHashSizeSingle = ARGV.shift.to_f
|
60
62
|
end
|
61
63
|
end
|
62
|
-
|
64
|
+
Macroape::MaxHashSizeSingle = 1000000 unless defined? Macroape::MaxHashSizeSingle
|
63
65
|
|
64
66
|
|
65
67
|
if filename == '.stdin'
|
66
|
-
|
68
|
+
# TODO
|
67
69
|
else
|
68
70
|
raise "Error! File #{filename} doesn't exist" unless File.exist?(filename)
|
69
|
-
pwm = PWM
|
71
|
+
pwm = Bioinform::PWM.new( File.read(filename) )
|
70
72
|
end
|
71
|
-
pwm
|
73
|
+
pwm.background(background)
|
72
74
|
|
73
75
|
counts = pwm.discrete(discretization).counts_by_thresholds(* thresholds.map{|count| count * discretization})
|
74
|
-
pvalues = counts.map{|count| count.to_f / pwm.
|
76
|
+
pvalues = counts.map{|count| count.to_f / pwm.vocabulary_volume}
|
75
77
|
pvalues.zip(thresholds,counts).each{|pvalue,threshold,count|
|
76
78
|
puts "#{threshold}\t#{count}\t#{pvalue}"
|
77
79
|
}
|