Implementing new way to control the protein loop overlap for hicHyper…

…optDetectLoops and improve docs of hicValidateLocations. Feature request #723

hicexplorer/hicHyperoptDetectLoops.py

@@ -49,6 +49,11 @@ def parse_arguments(args=None):
                            default=10000,
                            help='Resolution of matrix'
                            ' (Default: %(default)s).')
+    parserOpt.add_argument('--chrPrefixLoops', '-cl',
+                           help='Adding / removing / do nothing a \'chr\'-prefix to chromosome name of the loops.',
+                           choices=[None, 'add', 'remove'],
+                           default=None
+                           )
     parserOpt.add_argument('--threads', '-t',
                            help='Number of threads (uses the python multiprocessing module)'
                            ' (Default: %(default)s).',
@@ -70,14 +75,13 @@ def parse_arguments(args=None):
     return parser
 
 
-def compute_score(pLoopFile, pProteinFile, pMaximumNumberOfLoops, pResolution):
+def compute_score(pLoopFile, pProteinFile, pMaximumNumberOfLoops, pResolution, pChrPrefixLoops):
     with open(pLoopFile, 'r') as file:
         lines = file.readlines()
         if len(lines) == 0:
             return 1
     outfile_statistics = NamedTemporaryFile()
-    args = "--data {} --protein {} -cl --resolution {} --outFileName {}".format(pLoopFile, pProteinFile, pResolution, outfile_statistics.name).split()
-    print(args)
+    args = "--data {} --protein {} -cl {} --resolution {} --outFileName {}".format(pLoopFile, pProteinFile, pChrPrefixLoops, pResolution, outfile_statistics.name).split()
     hicValidateLocations.main(args)
     data_dict = {}
 
@@ -108,15 +112,14 @@ def objective(pArgs):
             pArgs['maxLoopDistance'], pArgs['threads'], pArgs['threads']).split()
     hicDetectLoops.main(args)
 
-    error_score = compute_score(outfile_loop.name, pArgs['proteinFile'], pArgs['maximumNumberOfLoops'], pArgs['resolution'])
+    error_score = compute_score(outfile_loop.name, pArgs['proteinFile'], pArgs['maximumNumberOfLoops'], pArgs['resolution'], pArgs['chrPrefixLoops'])
     print('Error score: {}'.format(error_score))
     return error_score
 
 
 def main(args=None):
 
     args = parse_arguments().parse_args(args)
-
     space = {
 
         'pit': hp.uniform('pit', 0, 100),
@@ -130,7 +133,8 @@ def main(args=None):
         'proteinFile': args.proteinFile,
         'maximumNumberOfLoops': args.maximumNumberOfLoops,
         'resolution': args.resolution,
-        'threads': args.threads
+        'threads': args.threads,
+        'chrPrefixLoops': args.chrPrefixLoops
 
     }
 

hicexplorer/hicValidateLocations.py

@@ -61,12 +61,12 @@ def parse_arguments(args=None):
                            'First file contains all loop locations with protein location matches, second file contains statistics about this matching.'
                            )
     parserOpt.add_argument('--chrPrefixLoops', '-cl',
-                           help='Adding a \'chr\'-prefix to chromosome name of the loops.',
+                           help='Adding / removing / do nothing a \'chr\'-prefix to chromosome name of the loops.',
                            choices=[None, 'add', 'remove'],
                            default=None
                            )
     parserOpt.add_argument('--chrPrefixProtein', '-cp',
-                           help='Adding a \'chr\'-prefix to chromosome name of the protein.',
+                           help='Adding / removing / do nothing a \'chr\'-prefix to chromosome name of the protein.',
                            choices=[None, 'add', 'remove'],
                            default=None
                            )

hicexplorer/test/general/test_hicHyperoptDetectLoops.py

@@ -22,29 +22,44 @@
 
 def are_files_equal(file1, file2, delta=None):
     equal = True
-    if delta:
-        mismatches = 0
+    # if delta:
+    mismatches = 0
     with open(file1) as textfile1, open(file2) as textfile2:
         for x, y in zip(textfile1, textfile2):
             if x.startswith('File'):
                 continue
             if x != y:
-                if delta:
-                    mismatches += 1
-                    if mismatches > delta:
-                        equal = False
-                        break
-                else:
-                    equal = False
-                    break
+                mismatches += 1
+                if mismatches > delta:
+                    return False
     return equal
 
 
 def test_main():
     outfile = NamedTemporaryFile(suffix='.txt', delete=True)
-
+    outfile.close()
     args = "--matrix {} -p {} -ml {} -r {} --runs {} -o {}".format(
         ROOT + "hicDetectLoops/GSE63525_GM12878_insitu_primary_2_5mb.cool",
         ROOT + 'hicHyperoptDectedLoops/ctcf_sorted.bed', 3210, 10000, 2, outfile.name).split()
     hicHyperoptDetectLoops.main(args)
-    are_files_equal(outfile.name, ROOT + 'hicHyperoptDetectLoops/hyperopt_result.txt', delta=2)
+    assert are_files_equal(outfile.name, ROOT + 'hicHyperoptDetectLoops/hyperopt_result.txt', delta=2)
+
+
+def test_main_add():
+    outfile = NamedTemporaryFile(suffix='.txt', delete=True)
+    outfile.close()
+    args = "--matrix {} -p {} -ml {} -r {} --runs {} -o {} -cl {}".format(
+        ROOT + "hicDetectLoops/GSE63525_GM12878_insitu_primary_2_5mb.cool",
+        ROOT + 'hicHyperoptDectedLoops/ctcf_sorted.bed', 3210, 10000, 2, outfile.name, 'add').split()
+    hicHyperoptDetectLoops.main(args)
+    assert are_files_equal(outfile.name, ROOT + 'hicHyperoptDetectLoops/hyperopt_result.txt', delta=2)
+
+
+def test_main_remove():
+    outfile = NamedTemporaryFile(suffix='.txt', delete=True)
+    outfile.close()
+    args = "--matrix {} -p {} -ml {} -r {} --runs {} -o {} -cl {}".format(
+        ROOT + "hicDetectLoops/GSE63525_GM12878_insitu_primary_2_5mb.cool",
+        ROOT + 'hicHyperoptDectedLoops/ctcf_sorted.bed', 3210, 10000, 2, outfile.name, 'remove').split()
+    hicHyperoptDetectLoops.main(args)
+    assert are_files_equal(outfile.name, ROOT + 'hicHyperoptDetectLoops/hyperopt_result.txt', delta=2)