Fix for newer pybedtools versions (#98)

* Add test to catch pybedtools 0.9.1+ error

* Use mamba, add test with downgrading pybedtools

* Remove --name from mamba commands

* Add check for pybedtools version

* If 0.9.1+, use the newly introduced parameter, which is required for the run to complete successfully

* Typo fix

* Check pybedtools versions in CI

azure-pipelines.yml

@@ -21,8 +21,9 @@ jobs:
     displayName: Install clang-format and clang-tidy
   - script: |
       source activate ntjoin_CI
-      conda install --yes --quiet --name ntjoin_CI -c conda-forge -c bioconda python=3.9 pylint samtools=1.14
-      conda install --yes --quiet --name ntjoin_CI -c conda-forge -c bioconda --file requirements.txt
+      conda install --yes --quiet --name ntjoin_CI -c conda-forge -c bioconda python=3.9 mamba
+      mamba install --yes --quiet -c conda-forge -c bioconda pylint samtools
+      mamba install --yes --quiet -c conda-forge -c bioconda --file requirements.txt
     displayName: Install Anaconda packages
   - script: |
       source activate ntjoin_CI
@@ -32,10 +33,18 @@ jobs:
     displayName: Run pylint
   - script: |
       source activate ntjoin_CI
-      conda install --yes --quiet --name ntjoin_CI -c conda-forge -c bioconda pytest bedtools samtools
+      mamba install --yes --quiet -c conda-forge -c bioconda pytest
       cd tests
       pytest -vs ntjoin_test.py
     displayName: Run pytests
+  - script: |
+      source activate ntjoin_CI
+      python3 -c 'import pybedtools; print(pybedtools.__version__)'
+      mamba install --yes --quiet -c conda-forge -c bioconda pybedtools=0.9.0
+      python3 -c 'import pybedtools; print(pybedtools.__version__)'
+      cd tests
+      pytest -vs ntjoin_test.py
+    displayName: Run pytests with lower pybedtools version
 
 - job: macOS_default_clang
   pool:
@@ -49,12 +58,21 @@ jobs:
 
   - script: |
       source activate ntjoin_CI
-      conda install --yes --quiet --name ntjoin_CI -c conda-forge -c bioconda python=3.9 pylint samtools=1.14
-      conda install --yes --quiet --name ntjoin_CI -c conda-forge -c bioconda --file requirements.txt
+      conda install --yes --quiet --name ntjoin_CI -c conda-forge -c bioconda python=3.9 mamba
+      mamba install --yes --quiet -c conda-forge -c bioconda pylint samtools
+      mamba install --yes --quiet -c conda-forge -c bioconda --file requirements.txt
     displayName: Install Anaconda packages
   - script: |
       source activate ntjoin_CI
-      conda install --yes --quiet --name ntjoin_CI -c conda-forge -c bioconda pytest bedtools samtools
+      mamba install --yes --quiet -c conda-forge -c bioconda pytest
       cd tests
       pytest -vs ntjoin_test.py
     displayName: Run pytests
+  - script: |
+      source activate ntjoin_CI
+      python3 -c 'import pybedtools; print(pybedtools.__version__)'
+      mamba install --yes --quiet -c conda-forge -c bioconda pybedtools=0.9.0
+      python3 -c 'import pybedtools; print(pybedtools.__version__)'
+      cd tests
+      pytest -vs ntjoin_test.py
+    displayName: Run pytests with lower pybedtools version

bin/ntjoin_assemble.py

@@ -878,9 +878,14 @@ def print_unassigned(self, assembly, assembly_fa, incorporated_segments, params,
         "Also print out the sequences that were NOT scaffolded"
         incorporated_segments_str = "\n".join([f"{chrom}\t{s}\t{e}"
                                                for chrom, s, e in incorporated_segments])
-        incorporated_segments_bed = pybedtools.BedTool(incorporated_segments_str,
-                                                       from_string=True).sort()
         genome_bed, genome_dict = self.format_bedtools_genome(Ntjoin.scaffolds)
+        # Needed to deal with failure in complement step seen with pybedtools 0.9.1+
+        if pybedtools.__version__ < "0.9.1":
+            incorporated_segments_bed = pybedtools.BedTool(incorporated_segments_str,
+                                                           from_string=True).sort()
+        else:
+            incorporated_segments_bed = pybedtools.BedTool(incorporated_segments_str,
+                                                           from_string=True).sort(genome=genome_dict)
         missing_bed = genome_bed.complement(i=incorporated_segments_bed, g=genome_dict)
         missing_bed.saveas(self.args.p + "." + assembly + ".unassigned.bed")
 

tests/ntjoin_test.py

@@ -218,3 +218,8 @@ def test_mx_r_r_overlap():
     "Testing ntJoin with assembly + reference, rev-rev orientation, overlap code on"
     paths = run_ntjoin_overlap("ref.fa", "scaf.r-r.overlapping.fa", "f-r_test_overlap")
     assert paths.pop() == "ntJoin0\t1-:66-2099 20N 2-:0-2297"
+
+def test_more_sequences():
+    "Testing ntJoin with more sequences - to catch pybedtools error seen with 0.9.1+"
+    paths = run_ntjoin_overlap("ref.longer.fa", "scaf.more_seqs.fa", "more_seqs_test")
+    assert len(paths) == 1