Vb/bcftools

.github/workflows/main.yml

@@ -59,7 +59,7 @@ jobs:
         uses: actions/cache@v2
         with:
           path: testdata
-          key: hatchetcache04
+          key: hatchetcache05
 
       - name: Download Testing Data
         if: steps.cache-test-data.outputs.cache-hit != 'true'
@@ -75,17 +75,17 @@ jobs:
 
       - name: Install SAMtools
         run: |
-          wget https://sourceforge.net/projects/samtools/files/samtools/1.7/samtools-1.7.tar.bz2/download -O samtools-1.7.tar.bz2
-          tar xvjf samtools-1.7.tar.bz2
-          (cd samtools-1.7 && ./configure && make)
-          echo "HATCHET_PATHS_SAMTOOLS=$(realpath samtools-1.7)" >> $GITHUB_ENV
+          wget https://sourceforge.net/projects/samtools/files/samtools/1.11/samtools-1.11.tar.bz2/download -O samtools-1.11.tar.bz2
+          tar xvjf samtools-1.11.tar.bz2
+          (cd samtools-1.11 && ./configure && make)
+          echo "HATCHET_PATHS_SAMTOOLS=$(realpath samtools-1.11)" >> $GITHUB_ENV
 
       - name: Install BCFTools
         run: |
-          wget https://sourceforge.net/projects/samtools/files/samtools/1.7/bcftools-1.7.tar.bz2/download -O bcftools-1.7.tar.bz2
-          tar xvjf bcftools-1.7.tar.bz2
-          (cd bcftools-1.7 && ./configure && make)
-          echo "HATCHET_PATHS_BCFTOOLS=$(realpath bcftools-1.7)" >> $GITHUB_ENV
+          wget https://sourceforge.net/projects/samtools/files/samtools/1.11/bcftools-1.11.tar.bz2/download -O bcftools-1.11.tar.bz2
+          tar xvjf bcftools-1.11.tar.bz2
+          (cd bcftools-1.11 && ./configure && make)
+          echo "HATCHET_PATHS_BCFTOOLS=$(realpath bcftools-1.11)" >> $GITHUB_ENV
 
       - name: Install tabix
         run: |

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "hatchet"
-version = "1.0.3"
+version = "1.0.4"
 authors = [
   { name="Simone Zaccaria", email="[email protected]" },
   { name="Ben Raphael", email="[email protected]" },

src/hatchet/__init__.py

@@ -1,4 +1,4 @@
-__version__ = '1.0.3'
+__version__ = '1.0.4'
 
 import os.path
 from importlib.resources import path

src/hatchet/utils/check.py

@@ -7,6 +7,7 @@
 from contextlib import contextmanager
 import tempfile
 import shutil
+from packaging import version
 
 from hatchet import config
 import hatchet.data
@@ -29,7 +30,7 @@ def suppress_stdout():
             sys.stderr = old_stderr
 
 
-def _check_cmd(exe_path, exe_name, *args):
+def _check_cmd(exe_path, exe_name, cwd=None, *args):
     # This function should never raise Exceptions unless it's a genuine implementation bug
     # Only use exe and args that return a return code of 0
     # Use exe_path as '' if you expect <exe_name> to be on PATH
@@ -39,6 +40,7 @@ def _check_cmd(exe_path, exe_name, *args):
             cmd,
             stdout=subprocess.DEVNULL,
             stderr=subprocess.PIPE,
+            cwd=cwd,
             universal_newlines=True,
         )
         p.communicate()
@@ -49,22 +51,22 @@ def _check_cmd(exe_path, exe_name, *args):
         return True
 
 
-# Most command-line commands can be checked using _check_cmd(<folder_with_command>, <command>, '--version')
+# Most command-line commands can be checked using _check_cmd(<folder_with_command>, <command>, <cwd>, '--version')
 # Others, like below, need special handling because they have no simple invocations that return 0
 def _check_tabix():
     with tempfile.TemporaryDirectory() as tempdirname:
         with importlib.resources.path(hatchet.data, 'sample.sorted.gff.gz') as gz_path:
             _temp_gz_path = os.path.join(tempdirname, 'sample.sorted.gff.gz')
             shutil.copy(gz_path, _temp_gz_path)
-            return _check_cmd(config.paths.tabix, 'tabix', '-p', 'gff', _temp_gz_path, '-f')
+            return _check_cmd(config.paths.tabix, 'tabix', None, '-p', 'gff', _temp_gz_path, '-f')
 
 
 def _check_bgzip():
     with tempfile.TemporaryDirectory() as tempdirname:
         with importlib.resources.path(hatchet.data, 'sample.bbc') as file_path:
             _temp_file_path = os.path.join(tempdirname, 'sample.bbc')
             shutil.copy(file_path, _temp_file_path)
-            return _check_cmd(config.paths.bgzip, 'bgzip', _temp_file_path, '-f')
+            return _check_cmd(config.paths.bgzip, 'bgzip', None, _temp_file_path, '-f')
 
 
 def _check_picard():
@@ -92,6 +94,7 @@ def _check_picard():
             return _check_cmd(
                 exe_path,
                 exe_name,
+                None,
                 *args_pre,
                 'BuildBamIndex',
                 '--INPUT',
@@ -101,6 +104,17 @@ def _check_picard():
             )
 
 
+def _check_bcftools():
+    # The bcftools version we select should be capable of querying remote .vcf.gz files while also specifying
+    # a region; This is the use case in HATCHet's genotype_snps step; This seems to work with bcftools>=1.11
+    try:
+        cmd = os.path.join(config.paths.bcftools, 'bcftools')
+        bcftools_version = subprocess.check_output([cmd, '--version-only']).decode('utf-8')
+        return version.parse(bcftools_version) >= version.parse('1.11')
+    except:  # noqa: E722
+        return False
+
+
 def _check_python_import(which):
     try:
         importlib.import_module(which)
@@ -131,6 +145,7 @@ def _check_python_import(which):
             _check_cmd,
             config.paths.samtools,
             'samtools',
+            None,
             '--version',
         ),
         (
@@ -142,6 +157,7 @@ def _check_python_import(which):
             _check_cmd,
             config.paths.mosdepth,
             'mosdepth',
+            None,
             '--version',
         ),
     ],
@@ -155,18 +171,16 @@ def _check_python_import(which):
             _check_cmd,
             config.paths.samtools,
             'samtools',
+            None,
             '--version',
         ),
         (
             'bcftools',
             '',
-            'Please install bcftools executable and either ensure its on your PATH, or its location specified in '
+            'Please install bcftools>=1.11 executable and either ensure its on your PATH, or its location specified in '
             'hatchet.ini as config.paths.bcftools, or its location specified using the environment variable '
             'HATCHET_PATHS_BCFTOOLS',
-            _check_cmd,
-            config.paths.bcftools,
-            'bcftools',
-            '--version',
+            _check_bcftools,
         ),
     ],
     'count-alleles': [
@@ -179,6 +193,7 @@ def _check_python_import(which):
             _check_cmd,
             config.paths.samtools,
             'samtools',
+            None,
             '--version',
         ),
         (
@@ -187,10 +202,7 @@ def _check_python_import(which):
             'Please install bcftools executable and either ensure its on your PATH, or its location specified in '
             'hatchet.ini as config.paths.bcftools, or its location specified using the environment variable '
             'HATCHET_PATHS_BCFTOOLS',
-            _check_cmd,
-            config.paths.bcftools,
-            'bcftools',
-            '--version',
+            _check_bcftools,
         ),
     ],
     'phase-snps': [
@@ -200,10 +212,7 @@ def _check_python_import(which):
             'Please install bcftools executable and either ensure its on your PATH, or its location specified in '
             'hatchet.ini as config.paths.samtools, or its location specified using the environment variable '
             'HATCHET_PATHS_BCFTOOLS',
-            _check_cmd,
-            config.paths.bcftools,
-            'bcftools',
-            '--version',
+            _check_bcftools,
         ),
         (
             'picard',
@@ -223,6 +232,7 @@ def _check_python_import(which):
             _check_cmd,
             config.paths.shapeit,
             'shapeit',
+            None,
             '--version',
         ),
         (