Benchmark for ClustalO

.gitignore

@@ -6,3 +6,4 @@ timings.png
 balibase.zip
 *.fasta
 data/
+balibase_results/

.gitlab-ci.yml

@@ -8,6 +8,9 @@ cache:
 test:
         script:
                 - python3 -m unittest alignementseq_tests test_alignementseq_multiple
+        artifacts:
+                paths:
+                        - balibase_results/
 
 test_performance:
         script:

alignementseq_tests.py

@@ -30,7 +30,7 @@ import os
 import sys
 import io
 
-spec, m = None, None
+bali_spec, bali_m = None, None
 
 def bali_score(ref_fasta_path, test_fasta_path):
     """Portable wrapper for the bali_score.py script.
@@ -39,16 +39,16 @@ def bali_score(ref_fasta_path, test_fasta_path):
     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)
+    global bali_spec, bali_m
+    if bali_spec is None:
+        bali_spec = importlib.util.spec_from_file_location("baliscore", path)
+        bali_m = importlib.util.module_from_spec(bali_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)
+        bali_spec.loader.exec_module(bali_m)
     finally:
         sys.argv = tmp_argv
         sys.stdout = tmp_stdout
@@ -162,8 +162,6 @@ class BalibaseTestCase(unittest.TestCase):
             item.append(name)
             yield item
 
-class AlignmentSeqTestCase(BalibaseTestCase):
-    
     def get_dataset_records(self):
         """Generator function to iterate over the record generator  of each
         unaligned fasta file.
@@ -175,6 +173,8 @@ class AlignmentSeqTestCase(BalibaseTestCase):
                     "fasta")
             yield records, filename
 
+class AlignmentSeqTestCase(BalibaseTestCase):
+
     def assertSameResidues(self, str1, str2):
         """Strip strings of their '-' before comparing them
         """
@@ -288,9 +288,13 @@ class AlignmentSeqTestCase(BalibaseTestCase):
                 print(ex_r2)
                 raise
  
-    def test_benchmark_multiple_align(self):
-        """Tests the multiple_align function (using blosum and gap extension)."""
-        from Bio import AlignIO
+    def create_clustalo_alignments(self):
+        """
+        Submit alignment jobs to the EMBL pipeline to get reference ClustalO
+        alignments.
+
+        A rate limiting is applied, typically 5 jobs/second.
+        """
         import clustalo as cl
      
         dataset_dir = os.path.join(self.balibase_path, "RV11.unaligned")
@@ -315,6 +319,29 @@ class AlignmentSeqTestCase(BalibaseTestCase):
             jobs.append(j)
         for j in jobs:
             j.join()
+
+    def test_benchmark_multiple_align(self):
+        """Tests the multiple_align function (using blosum and gap extension)."""
+
+        res_dir = "balibase_results"
+        if not os.path.isdir(res_dir):
+            os.mkdir(res_dir)
+
+        # Cached
+        self.create_clustalo_alignments()
+         
+        ref_dataset_dir = os.path.join(self.balibase_path, "RV11.aligned")
+        clustalo_dir = os.path.join(self.balibase_path, "clustalo")
+
+        scores = []
+        for filename in os.listdir(ref_dataset_dir):
+            scores.append(bali_score(
+                    os.path.join(ref_dataset_dir, filename),
+                    os.path.join(clustalo_dir, filename)
+                    ))
+        scores = pd.DataFrame(scores)
+        scores.to_csv(os.path.join(res_dir, "clustalo.csv"))
+
  
     def test_align_dihedrals(self):
         """Iterates over the balibase dataset, match the chains from the mmcif