Merge pull request #267 from bcgsc/feature/python3.9-10

Feature/python3.9 10

.github/workflows/build.yml

@@ -15,7 +15,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        python-version: ["3.7", "3.8"]
+        python-version: ["3.7", "3.8", "3.9", "3.10"]
     name: python-${{ matrix.python-version }}
     steps:
     - uses: actions/checkout@v2

.github/workflows/quick-tests.yml

@@ -10,7 +10,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        python-version: ["3.7", "3.8"]
+        python-version: ["3.7", "3.8", "3.9", "3.10"]
     name: python-${{ matrix.python-version }} quick
     steps:
     - uses: actions/checkout@v2

Dockerfile

@@ -28,6 +28,8 @@ RUN wget http://hgdownload.soe.ucsc.edu/admin/exe/linux.x86_64/blat/blat && \
 
 COPY setup.py setup.py
 COPY setup.cfg setup.cfg
+COPY MANIFEST.in MANIFEST.in
+COPY pyproject.toml pyproject.toml
 COPY src src
 COPY LICENSE LICENSE
 COPY README.md README.md

MANIFEST.in

@@ -1,5 +1,5 @@
 recursive-include src *.py *.json
 include README.md
 include LICENSE
-prune docs/build
-prune docs/source/auto
+prune docs
+prune tests

pyproject.toml

@@ -0,0 +1 @@
+build-backend = "setuptools.build_meta"

setup.cfg

@@ -1,23 +1,97 @@
-[nosetests]
-with-coverage=1
-cover-package=mavis,tab
-cover-html=1
-cover-html-dir=coverage
-cover-inclusive=1
-cover-erase=1
-processes=2
-process-timeout=600
-
 [metadata]
-description-file = README.md
+name = mavis
+version = 2.2.10
+url = https://github.com/bcgsc/mavis.git
+download_url = https://github.com/bcgsc/mavis/archive/v2.2.10.tar.gz
+description = A Structural Variant Post-Processing Package
+author_email = [email protected]
+author = Caralyn Reisle
+maintainer_email = [email protected]
+maintainer = mavis
+long_description = file: README.md
+long_description_content_type = text/markdown
 license_file = LICENSE
+project_urls = mavis = http://mavis.bcgsc.ca
 
 [bdist_wheel]
 universal = 1
 
 [pycodestyle]
-ignore = E501,W503,E203
+ignore = E501
+    W503
+    E203
 statistics = True
 
 [flake8]
-ignore = E501,W503,E203
+ignore = E501
+    W503
+    E203
+
+[options]
+packages = find:
+package_dir =
+     = src
+python_requires = >=3.7
+dependency_links = []
+include_package_data = True
+install_requires =
+    biopython>=1.70, <1.78
+    braceexpand==0.1.2
+    colour
+    Distance>=0.1.3
+    mavis_config>=1.1.0, <2.0.0
+    networkx>=2.5,<3
+    numpy>=1.13.1
+    pandas>=1.1, <2
+    pysam
+    Shapely>=1.6.4.post1
+    shortuuid>=0.5.0
+    svgwrite
+    typing_extensions>=4
+setup_requires =
+    pip>=9.0.0
+    setuptools>=36.0.0
+
+[options.packages.find]
+exclude = tests
+where = src
+
+[options.extras_require]
+doc =
+    mkdocs==1.1.2
+    markdown-refdocs
+    mkdocs-material==5.4.0
+    markdown-include
+    mkdocs-simple-hooks==0.1.2
+test =
+    timeout-decorator>=0.3.3
+    coverage>=4.2
+    pycodestyle>=2.3.1
+    pytest
+    pytest-cov
+dev =
+    black
+    flake8
+    twine
+    wheel
+    timeout-decorator>=0.3.3
+    coverage>=4.2
+    pycodestyle>=2.3.1
+    pytest
+    pytest-cov
+    mkdocs==1.1.2
+    markdown-refdocs
+    mkdocs-material==5.4.0
+    markdown-include
+    mkdocs-simple-hooks==0.1.2
+deploy =
+    twine
+    wheel
+tools =
+    pyensembl
+    simplejson
+
+[options.entry_points]
+console_scripts =
+    mavis = mavis.main:main
+    calculate_ref_alt_counts = tools.calculate_ref_alt_counts:main

setup.py

@@ -1,35 +1,7 @@
 import os
 import re
 
-from setuptools import find_packages, setup
-
-VERSION = '2.2.8'
-
-
-def parse_md_readme():
-    """
-    pypi won't render markdown. After conversion to rst it will still not render unless raw directives are removed
-    """
-    try:
-        from m2r import parse_from_file
-
-        rst_lines = parse_from_file('README.md').split('\n')
-        long_description = [
-            '.. image:: http://mavis.bcgsc.ca/docs/latest/_static/acronym.svg\n\n|\n'
-        ]  # backup since pip can't handle raw directives
-        i = 0
-        while i < len(rst_lines):
-            if re.match(r'^..\s+raw::.*', rst_lines[i]):
-                i += 1
-                while re.match(r'^(\s\s+|\t|$).*', rst_lines[i]):
-                    i += 1
-            else:
-                long_description.append(re.sub('>`_ ', '>`__ ', rst_lines[i]))  # anonymous links
-                i += 1
-        long_description = '\n'.join(long_description)
-    except (ImportError, OSError):
-        long_description = ''
-    return long_description
+from setuptools import setup
 
 
 def check_nonpython_dependencies():
@@ -54,78 +26,5 @@ def check_nonpython_dependencies():
         print('Found: aligner at', pth)
 
 
-# HSTLIB is a dependency for pysam.
-# The cram file libraries fail for some OS versions and mavis does not use cram files so we disable these options
-os.environ['HTSLIB_CONFIGURE_OPTIONS'] = '--disable-lzma --disable-bz2 --disable-libcurl'
-
-
-TEST_REQS = [
-    'timeout-decorator>=0.3.3',
-    'coverage>=4.2',
-    'pycodestyle>=2.3.1',
-    'pytest',
-    'pytest-cov',
-]
-
-
-DOC_REQS = [
-    'mkdocs==1.1.2',
-    'markdown_refdocs',
-    'mkdocs-material==5.4.0',
-    'markdown-include',
-    'mkdocs-simple-hooks==0.1.2',
-]
-
-
-INSTALL_REQS = [
-    'Distance>=0.1.3',
-    'Shapely>=1.6.4.post1',
-    'biopython>=1.70, <1.78',
-    'braceexpand==0.1.2',
-    'colour',
-    'networkx==1.11.0',
-    'numpy>=1.13.1',
-    'pandas>=1.1, <2',
-    'pysam',
-    'shortuuid>=0.5.0',
-    'svgwrite',
-    'mavis_config>=1.1.0, <2.0.0',
-]
-
-DEPLOY_REQS = ['twine', 'm2r', 'wheel']
-
-
-setup(
-    name='mavis',
-    version='{}'.format(VERSION),
-    url='https://github.com/bcgsc/mavis.git',
-    download_url='https://github.com/bcgsc/mavis/archive/v{}.tar.gz'.format(VERSION),
-    package_dir={'': 'src'},
-    packages=find_packages(where='src'),
-    description='A Structural Variant Post-Processing Package',
-    long_description=parse_md_readme(),
-    install_requires=INSTALL_REQS,
-    extras_require={
-        'docs': DOC_REQS,
-        'test': TEST_REQS,
-        'dev': ['black==20.8b1', 'flake8'] + DOC_REQS + TEST_REQS + DEPLOY_REQS,
-        'deploy': DEPLOY_REQS,
-        'tools': ['pyensembl', 'simplejson'],
-    },
-    tests_require=TEST_REQS,
-    setup_requires=['pip>=9.0.0', 'setuptools>=36.0.0'],
-    python_requires='>=3.7',
-    author='Caralyn Reisle',
-    author_email='[email protected]',
-    test_suite='tests',
-    entry_points={
-        'console_scripts': [
-            'mavis = mavis.main:main',
-            'calculate_ref_alt_counts = tools.calculate_ref_alt_counts:main',
-        ]
-    },
-    include_package_data=True,
-    data_files=[('mavis', ['src/mavis/schemas/config.json', 'src/mavis/schemas/overlay.json'])],
-    project_urls={'mavis': 'http://mavis.bcgsc.ca'},
-)
+setup()
 check_nonpython_dependencies()

src/mavis/assemble.py

@@ -66,6 +66,15 @@ class DeBruijnGraph(nx.DiGraph):
     enforces edge weights
     """
 
+    def get_out_edges(self, *args, **kwargs):
+        return list(self.out_edges(*args, **kwargs))
+
+    def get_in_edges(self, *args, **kwargs):
+        return list(self.in_edges(*args, **kwargs))
+
+    def get_nodes(self, *args, **kwargs):
+        return list(self.nodes(*args, **kwargs))
+
     def get_edge_freq(self, n1, n2):
         """
         returns the freq from the data attribute for a specified edge
@@ -85,7 +94,7 @@ def add_edge(self, n1, n2, freq=1):
         nx.DiGraph.add_edge(self, n1, n2, freq=freq)
 
     def all_edges(self, *nodes, data=False):
-        return self.in_edges(*nodes, data=data) + self.out_edges(*nodes, data=data)
+        return self.get_in_edges(*nodes, data=data) + self.get_out_edges(*nodes, data=data)
 
     def trim_tails_by_freq(self, min_weight):
         """
@@ -95,7 +104,7 @@ def trim_tails_by_freq(self, min_weight):
             min_weight (int): the minimum weight for an edge to be retained
         """
         ends = sorted(
-            [n for n in self.nodes() if self.out_degree(n) == 0 or self.in_degree(n) == 0]
+            [n for n in self.get_nodes() if self.out_degree(n) == 0 or self.in_degree(n) == 0]
         )
         visited = set()
 
@@ -126,16 +135,16 @@ def trim_forks_by_freq(self, min_weight):
         for all nodes in the graph, if the node has an out-degree > 1 and one of the outgoing
         edges has freq < min_weight. then that outgoing edge is deleted
         """
-        nodes = [n for n in self.nodes() if self.degree(n) > 2]
+        nodes = [n for n in self.get_nodes() if self.degree(n) > 2]
         for node in sorted(nodes):
             if self.out_degree(node) > 1:
-                outgoing_edges = self.out_edges(node, data=True)
+                outgoing_edges = self.get_out_edges(node, data=True)
                 best = max([e[2]['freq'] for e in outgoing_edges])
                 for src, tgt, data in outgoing_edges:
                     if data['freq'] < min_weight and data['freq'] != best:
                         self.remove_edge(src, tgt)
             if self.in_degree(node) > 1:
-                ingoing_edges = self.in_edges(node, data=True)
+                ingoing_edges = self.get_in_edges(node, data=True)
                 best = max([e[2]['freq'] for e in ingoing_edges])
                 for src, tgt, data in ingoing_edges:
                     if data['freq'] < min_weight and data['freq'] != best:
@@ -157,15 +166,15 @@ def trim_noncutting_paths_by_freq(self, min_weight):
             else:
                 path = []
                 while self.in_degree(src) == 1 and self.out_degree(src) == 1:
-                    s, t, data = self.in_edges(src, data=True)[0]
+                    s, t, data = self.get_in_edges(src, data=True)[0]
                     if data['freq'] >= min_weight or s in path:
                         break
                     path.insert(0, src)
                     src = s
                 path.insert(0, src)
 
                 while self.in_degree(tgt) == 1 and self.out_degree(tgt) == 1:
-                    s, t, data = self.out_edges(tgt, data=True)[0]
+                    s, t, data = self.get_out_edges(tgt, data=True)[0]
                     if data['freq'] >= min_weight or t in path:
                         break
                     path.append(tgt)
@@ -193,7 +202,7 @@ def get_sinks(self, subgraph=None):
         """
         nodeset = set()
         if subgraph is None:
-            subgraph = self.nodes()
+            subgraph = self.get_nodes()
         for node in subgraph:
             if self.out_degree(node) == 0:
                 nodeset.add(node)
@@ -205,7 +214,7 @@ def get_sources(self, subgraph=None):
         """
         nodeset = set()
         if subgraph is None:
-            subgraph = self.nodes()
+            subgraph = self.get_nodes()
         for node in subgraph:
             if self.in_degree(node) == 0:
                 nodeset.add(node)
@@ -227,7 +236,7 @@ def digraph_connected_components(graph, subgraph=None):
         List[List]: returns a list of compnents which are lists of node names
     """
     if subgraph is None:
-        subgraph = set(graph.nodes())
+        subgraph = set(graph.get_nodes())
     g = nx.Graph()
     for src, tgt in graph.edges():
         if src in subgraph and tgt in subgraph:
@@ -387,7 +396,7 @@ def assemble(
         for kmer in kmers_list:
             assembly.add_edge(kmer[:-1], kmer[1:])
     # use the ab min edge weight to remove all low weight edges first
-    nodes = list(assembly.nodes())
+    nodes = assembly.get_nodes()
     for n in nodes:
         if assembly.in_degree(n) == 0 and assembly.out_degree(n) == 0:
             assembly.remove_node(n)
@@ -396,7 +405,7 @@ def assemble(
         subgraph = assembly.subgraph(component)
         if not nx.is_directed_acyclic_graph(subgraph):
             log('dropping cyclic component', time_stamp=False)
-            for node in subgraph.nodes():
+            for node in subgraph.get_nodes():
                 assembly.remove_node(node)
     # initial data cleaning
     assembly.trim_forks_by_freq(min_edge_trim_weight)
@@ -409,7 +418,7 @@ def assemble(
         # pull the path scores
         path_scores.update(
             pull_contigs_from_component(
-                assembly.subgraph(component),
+                assembly.subgraph(component).copy(),
                 component,
                 min_edge_trim_weight=min_edge_trim_weight,
                 assembly_max_paths=assembly_max_paths,