fixed tests

hicexplorer/hicBuildMatrix.py

@@ -445,12 +445,14 @@ def check_dangling_end(read, dangling_sequences):
     ds = dangling_sequences
     # skip forward read that stars with the restriction sequence
     if not read.is_reverse and \
-            read.seq.upper()[0:len(ds['pat_forw'])] == ds['pat_forw']:
+            read.seq.upper().starswidth(ds['pat_forw']):
+            # read.seq.upper()[0:len(ds['pat_forw'])] == ds['pat_forw']:
         return True
 
     # skip reverse read that ends with the restriction sequence
     if read.is_reverse and \
-            read.seq.upper()[-len(ds['pat_rev']):] == ds['pat_rev']:
+            read.seq.upper().endswidth(ds['pat_rev']):
+            # read.seq.upper()[-len(ds['pat_rev']):] == ds['pat_rev']:
         return True
 
     return False
@@ -1059,7 +1061,7 @@ def main(args=None):
         args.restrictionSequence = args.restrictionSequence.upper()
         args.danglingSequence = args.danglingSequence.upper()
         dangling_sequences['pat_forw'] = args.danglingSequence
-        dangling_sequences['pat_rev'] = str(Seq(args.danglingSequence, generic_dna).complement())
+        dangling_sequences['pat_rev'] = str(Seq(args.danglingSequence, generic_dna).reverse_complement())
 
         sys.stderr.write("dangling sequences to check "
                          "are {}\n".format(dangling_sequences))

hicexplorer/test/test_buildMatrix.py

@@ -2,7 +2,6 @@
 from hicexplorer import HiCMatrix as hm
 from tempfile import NamedTemporaryFile, mkdtemp
 import shutil
-import filecmp
 import os
 import numpy.testing as nt
 
@@ -13,11 +12,23 @@
 dpnii_file = ROOT + "DpnII.bed"
 
 
+def are_files_equal(file1, file2):
+    equal = True
+    with open(file1) as textfile1, open(file2) as textfile2:
+        for x, y in zip(textfile1, textfile2):
+            if x.startswith('File'):
+                continue
+            if x != y:
+                equal = False
+                break
+    return equal
+
+
 def test_build_matrix():
     outfile = NamedTemporaryFile(suffix='.h5', delete=False)
     outfile.close()
     qc_folder = mkdtemp(prefix="testQC_")
-    args = "-s {} {} -o {} -bs 5000 -b /tmp/test.bam --QCfolder {}".format(sam_R1, sam_R2,
+    args = "-s {} {} -o {} -bs 5000 -b /tmp/test.bam --QCfolder {} --threads 4".format(sam_R1, sam_R2,
                                                                            outfile.name,
                                                                            qc_folder).split()
     hicBuildMatrix.main(args)
@@ -28,7 +39,7 @@ def test_build_matrix():
     nt.assert_equal(test.cut_intervals, new.cut_intervals)
 
     print set(os.listdir(ROOT + "QC/"))
-    filecmp(ROOT + "QC/QC.log", qc_folder + "/QC.log")
+    assert are_files_equal(ROOT + "QC/QC.log", qc_folder + "/QC.log")
     assert set(os.listdir(ROOT + "QC/")) == set(os.listdir(qc_folder))
 
     os.unlink(outfile.name)
@@ -54,6 +65,7 @@ def test_build_matrix_rf():
     nt.assert_equal(test.cut_intervals, new.cut_intervals)
 
     print set(os.listdir(ROOT + "QC_rc/"))
+    assert are_files_equal(ROOT + "QC_rc/QC.log", qc_folder + "/QC.log")
     assert set(os.listdir(ROOT + "QC_rc/")) == set(os.listdir(qc_folder))
 
     os.unlink(outfile.name)

hicexplorer/test/test_data/QC_rc/QC.log

@@ -4,21 +4,21 @@ Pairs considered	99983
 Min rest. site distance	150		
 Max rest. site distance	1500		
 
-Pairs used	36576	(36.58)	(40.10)
+Pairs used	36552	(36.56)	(40.08)
 One mate unmapped	8777	(8.78)	(9.62)
 One mate not unique	3603	(3.60)	(3.95)
 One mate low quality	34877	(34.88)	(38.24)
-dangling end	190	(0.19)	(0.21)
-self ligation (removed)	5061	(5.06)	(5.55)
+dangling end	209	(0.21)	(0.23)
+self ligation (removed)	5056	(5.06)	(5.54)
 One mate not close to rest site	751	(0.75)	(0.82)
-same fragment (800 bp)	10136	(10.14)	(11.11)
+same fragment (800 bp)	10146	(10.15)	(11.12)
 self circle	4274	(4.27)	(4.69)
 duplicated pairs	12	(0.01)	(0.01)
 Of pairs used:
-inter chromosomal	5849	(15.99)
-short range < 20kb	8694	(23.77)
-long range	22033	(60.24)
-inward pairs	6993	(19.12)
-outward pairs	9525	(26.04)
-left pairs	7020	(19.19)
-right pairs	7189	(19.65)
+inter chromosomal	5849	(16.00)
+short range < 20kb	8689	(23.77)
+long range	22014	(60.23)
+inward pairs	6977	(19.09)
+outward pairs	9525	(26.06)
+left pairs	7012	(19.18)
+right pairs	7189	(19.67)

hicexplorer/test/test_data/QC_rc/hicQC.html

@@ -131,11 +131,11 @@ <h2 id="table">Number of reads table</h2>
     </tr>
     <tr>
       <th>Pairs used</th>
-      <td>36576</td>
+      <td>36552</td>
     </tr>
     <tr>
       <th>dangling end</th>
-      <td>190</td>
+      <td>209</td>
     </tr>
     <tr>
       <th>duplicated pairs</th>
@@ -147,15 +147,15 @@ <h2 id="table">Number of reads table</h2>
     </tr>
     <tr>
       <th>inward pairs</th>
-      <td>6993</td>
+      <td>6977</td>
     </tr>
     <tr>
       <th>left pairs</th>
-      <td>7020</td>
+      <td>7012</td>
     </tr>
     <tr>
       <th>long range</th>
-      <td>22033</td>
+      <td>22014</td>
     </tr>
     <tr>
       <th>outward pairs</th>
@@ -167,19 +167,19 @@ <h2 id="table">Number of reads table</h2>
     </tr>
     <tr>
       <th>same fragment (800 bp)</th>
-      <td>10136</td>
+      <td>10146</td>
     </tr>
     <tr>
       <th>self circle</th>
       <td>4274</td>
     </tr>
     <tr>
       <th>self ligation (removed)</th>
-      <td>5061</td>
+      <td>5056</td>
     </tr>
     <tr>
       <th>short range &lt; 20kb</th>
-      <td>8694</td>
+      <td>8689</td>
     </tr>
   </tbody>
 </table>

hicexplorer/test/test_data/QC_rc/table.txt

@@ -1,2 +1,2 @@
 File	Max rest. site distance	Min rest. site distance	One mate low quality	One mate not close to rest site	One mate not unique	One mate unmapped	Pairs considered	Pairs used	dangling end	duplicated pairs	inter chromosomal	inward pairs	left pairs	long range	outward pairs	right pairs	same fragment (800 bp)	self circle	self ligation (removed)	short range < 20kb
-small_test_rf_matrix.h5	1500	150	34877	751	3603	8777	99983	36576	190	12	5849	6993	7020	22033	9525	7189	10136	4274	5061	8694
+small_test_rf_matrix.h5	1500	150	34877	751	3603	8777	99983	36552	209	12	5849	6977	7012	22014	9525	7189	10146	4274	5056	8689