Fix EI-CoreBioinformatics#303 (EI-CoreBioinformatics#304)

Small bug in the compare utility, which led to wrong results in `pick`.

.travis.yml

@@ -1,6 +1,5 @@
 language: python
-sudo: required
-dist: xenial
+os: "linux"
 python:
   # - "3.4"
   # - "3.5"
@@ -21,12 +20,12 @@ install:
   - conda env create -n env_name --file environment.yml
   - source activate env_name
   - conda install -y -c conda-forge -- pytest-cov codecov;
+  - conda install -c bioconda -c conda-forge -y stringtie scallop gmap star hisat2 prodigal blast diamond transdecoder gnuplot kallisto samtools gffread
   - python setup.py develop;
 script:
-  - python setup.py test --addopts "-m slow Mikado/tests/test_system_calls.py::SerialiseChecker::test_subprocess_multi_empty_orfs"
-  - python setup.py test --addopts " --cov Mikado --cov-config=.coveragerc -m '(slow or not slow) and not triage'";
-  - cd sample_data; snakemake --use-conda --jobs 5 --cores 5
-  - rm -rf .snakemake/conda/;  # This causes weird bugs
+  - pytest -m slow Mikado/tests/test_system_calls.py::SerialiseChecker::test_subprocess_multi_empty_orfs
+  - pytest --cov Mikado --cov-config=.coveragerc -m '(slow or not slow) and not triage';
+  - cd sample_data; snakemake --jobs 5 --cores 5
   - cd ..;
   - python -c "import Mikado; Mikado.test(label='fast')";
   # Check that the seed is set properly

Mikado/scales/contrast.pyx

@@ -89,12 +89,16 @@ cdef str __assign_monoexonic_ccode(prediction, reference, long nucl_overlap, dou
                     ccode = "g"
                 else:
                     ccode = "e"
+        elif nucl_precision == 1:
+            ccode = "c"
         elif nucl_overlap > 0:
             ccode = "g"
         elif (nucl_recall == 0 and r_start < p_start < r_end):
             ccode = "i"  # Monoexonic fragment inside an intron
     elif p_exon_num > 1 and r_exon_num == 1:
-        if nucl_overlap > 0:
+        if nucl_recall == 1.0:  # The multiexonic is a complete extension of the monoexonic reference
+            ccode = "n"
+        elif nucl_overlap > 0:
             ccode = "G"  # Reverse generic overlap
         elif c_overlap(p_start, p_end, r_start, r_end, 0, 1) > 0:
             ccode = "ri"

Mikado/tests/assigner_test.py

@@ -1126,6 +1126,23 @@ def test_false_fusion(self):
         args.reference.close()
         args.prediction.close()
 
+    def test_monoexonic_contained(self):
+        t1 = loci.Transcript()
+        t1.chrom, t1.start, t1.end, t1.strand, t1.id = "Chr1", 9192977, 9193471, "+", "t1"
+        t1.add_exons([(t1.start, t1.end)])
+        t1.add_exons([(t1.start, t1.end)], features=["CDS"])
+        t1.finalize()
+        t2 = loci.Transcript()
+        t2.chrom, t2.start, t2.end, t2.strand, t2.id = "Chr1", 9182621, 9194526, "+", "t2"
+        t2.add_exons([(9192977, 9193480)], features=["CDS"])
+        t2.add_exons([(9186643, 9186780), (9182621, 9182770), (9192959, 9194526)])
+        t2.finalize()
+        mono_ref = scales.Assigner.compare(reference=t1, prediction=t2)[0]
+        mono_pred = scales.Assigner.compare(reference=t2, prediction=t1)[0]
+        self.assertEqual(mono_ref.n_recall, mono_pred.n_prec)
+        self.assertEqual(mono_pred.ccode[0], "c")
+        self.assertEqual(mono_ref.ccode[0], "n")
+
     def test_h_case(self):
 
         """