create python function for clustalo

alignementseq_tests.py

@@ -19,6 +19,42 @@ import Bio.Entrez
 from Bio.SeqRecord import SeqRecord
 from Bio.SubsMat.MatrixInfo import blosum62
 
+#==============================================================================
+import importlib
+import os
+import sys
+import io
+
+spec, m = None, None
+
+def bali_score(ref_fasta_path, test_fasta_path):
+    """Portable wrapper for the bali_score.py script.
+    
+    Calculate BALI scores from a test and a reference multiple sequence alignment in FASTA format.
+    Reproduces the results of the bali_score C program.
+    """
+    path = os.path.join("balibase", "bali_score.py")
+    global spec, m
+    if spec is None:
+        spec = importlib.util.spec_from_file_location("baliscore", path)
+        m = importlib.util.module_from_spec(spec)
+    tmp_argv = sys.argv
+    tmp_stdout = sys.stdout
+    buf = sys.stdout = io.StringIO()
+    sys.argv = [path, ref_fasta_path, test_fasta_path]
+    try:
+        spec.loader.exec_module(m)
+    finally:
+        sys.argv = tmp_argv
+        sys.stdout = tmp_stdout
+    buf.seek(0)
+    score = {}
+    for line in buf:
+        name, var = line.partition("=")[::2]
+        score[name.strip()] = float(var)
+    return score
+#==============================================================================
+
 class BalibaseTestCase(unittest.TestCase):
     """Base class to inherit by all test cases that need access to the balibase.
     """
@@ -179,30 +215,22 @@ class AlignmentSeqTestCase(BalibaseTestCase):
                 print(ex_r2)
                 raise
     
-# =============================================================================
-#     def test_multiple_align(self):
-#         """Tests the multiple_align function (using blosum and gap extension)."""
-#         #from alignementseq import multiple_align
-#         from Bio.Align.Applications import ClustalwCommandline
-#         from Bio import AlignIO
-#         import pip
-#         
-#         if not os.path.isfile("clustalo.py"):
-#             print("Fetching clustalo.py from Github...")
-#             urllib.request.urlretrieve(
-#                 "https://raw.githubusercontent.com/ebi-wp/webservice-clients/master/python/clustalo.py",
-#                 "clustalo.py")
-#             pip.main(["install --upgrade", "xmltramp2"])
-#             #pip("install --upgrade xmltramp2")
-#             print("Done")
-#         import clustalo
-#         # How to use: python clustalo.py --email <your@email.com> --sequence sp:wap_rat,sp:wap_mouse,sp:wap_pig
-#         
-#         for records, filename in self.get_dataset_records():
-#             clustalo("ariane.delrocq@polytechnique.edu", *list(records))
-#             
-#             print("Fine", filename)
-# =============================================================================
+ 
+    def test_benchmark_multiple_align(self):
+        """Tests the multiple_align function (using blosum and gap extension)."""
+        #from alignementseq import multiple_align
+        from Bio.Align.Applications import ClustalwCommandline
+        from Bio import AlignIO
+        import pip
+        
+        print("Test-benchmark multiple_align")
+        import clustalo as cl
+     
+        for records, filename in list(self.get_dataset_records())[:3]:
+            cl.runClustalO("ariane-la-fusee@laposte.net", "sp:wap_rat,sp:wap_mouse,sp:wap_pig", fmt='fasta')
+            # TODO: use real seq instead of rat, pig...
+            print("Fine", filename)
+ 
    
     
     def save_alignments(self, NameFile):