initial working version of refgenie integration

nf_core/refgenie.py

@@ -0,0 +1,160 @@
+#!/usr/bin/env python
+"""
+Update a nextflow.config file with refgenie genomes
+"""
+
+import logging
+import os
+import re
+from pathlib import Path
+from textwrap import dedent
+
+# import refgenconf
+from warnings import warn
+
+import rich
+from rich.logging import RichHandler
+
+import nf_core.utils
+
+# Set up logging
+log = logging.getLogger(__name__)
+log.setLevel(logging.INFO)
+
+# # Setup rich traceback
+stderr = rich.console.Console(stderr=True, force_terminal=nf_core.utils.rich_force_colors())
+rich.traceback.install(console=stderr, width=200, word_wrap=True, extra_lines=1)
+
+NF_CFG_TEMPLATE = """
+// This is a read-only config file managed by refgenie. Manual changes to this file will be overwritten
+// To make changes here, use refgenie to update the reference genome data
+params {{
+  genomes {{
+{content}
+  }}
+}}
+"""
+
+
+def _print_nf_config(rgc):
+    """
+    Generate a nextflow config file with the genomes
+    from the refgenie config file
+    Adapted from: https://github.com/refgenie/refgenie_nfcore
+
+    Takes a RefGenConf object as argument
+    """
+    abg = rgc.list_assets_by_genome()
+    genomes_str = ""
+    for genome, asset_list in abg.items():
+        genomes_str += f"    '{genome}' {{\n"
+        for asset in asset_list:
+            try:
+                pth = rgc.seek(genome, asset)
+            # Catch general exception instead of refgencof exception --> no refgenconf import needed
+            except Exception as e:
+                log.warn(f"{genome}/{asset} is incomplete, ignoring...")
+            else:
+                genomes_str += f'      {asset.ljust(20, " ")} = "{pth}"\n'
+        genomes_str += "    }\n"
+
+    return NF_CFG_TEMPLATE.format(content=genomes_str)
+
+
+def _update_nextflow_home_config(refgenie_genomes_config_file, nxf_home):
+    """
+    Update the $NXF_HOME/config file by adding a includeConfig statement to it
+    for the 'refgenie_genomes_config_file' if not already defined
+    """
+    # Check if NXF_HOME/config exists and has a
+    include_config_string = dedent(
+        f"""
+        ///// >>> nf-core + RefGenie >>> /////
+        // !! Contents within this block are managed by 'nf-core/tools' !!
+        // Includes auto-generated config file with RefGenie genome assets
+        includeConfig '{os.path.abspath(refgenie_genomes_config_file)}'
+        ///// <<< nf-core + RefGenie <<< /////
+        """
+    )
+    nxf_home_config = Path(nxf_home) / "config"
+    if os.path.exists(nxf_home_config):
+        # look for include statement in config
+        has_include_statement = False
+        with open(nxf_home_config, "r") as fh:
+            lines = fh.readlines()
+            for line in lines:
+                if re.match(f"\s*includeConfig\s*'{os.path.abspath(refgenie_genomes_config_file)}'", line):
+                    has_include_statement = True
+                    break
+
+        # if include statement is missing, add it to the last line
+        if not has_include_statement:
+            with open(nxf_home_config, "a") as fh:
+                fh.write(include_config_string)
+
+            log.info(f"Included refgenie_genomes.config to {nxf_home_config}")
+
+    else:
+        # create new config and add include statement
+        with open(nxf_home_config, "w") as fh:
+            fh.write(include_config_string)
+            log.info(f"Created new nextflow config file: {nxf_home_config}")
+
+
+def update_config(rgc):
+    """
+    Update the genomes.config file after a local refgenie database has been updated
+
+    This function is executed after running 'refgenie pull <genome>/<asset>'
+    The refgenie config file is transformed into a nextflow.config file, which is used to
+    overwrited the 'refgenie_genomes.config' file.
+    The path to the target  config file is inferred from the following options, in order:
+
+    - the 'nextflow_config' attribute in the refgenie config file
+    - the NXF_REFGENIE_PATH environment variable
+    - otherwise defaults to: $NXF_HOME/nf-core/refgenie_genomes.config
+
+    Additionaly, an 'includeConfig' statement is added to the file $NXF_HOME/config
+    """
+
+    # Compile nextflow refgenie_genomes.config from refgenie config
+    refgenie_genomes = _print_nf_config(rgc)
+
+    # Get the path to NXF_HOME
+    # If NXF_HOME is not set, create it at $HOME/.nextflow
+    # If $HOME is not set, set nxf_home to false
+    nxf_home = os.environ.get("NXF_HOME")
+    if not nxf_home:
+        try:
+            nxf_home = Path.home() / ".nextflow"
+            if not os.path.exists(nxf_home):
+                log.info(f"Creating NXF_HOME directory at {nxf_home}")
+                os.makedirs(nxf_home, exist_ok=True)
+        except RuntimeError:
+            nxf_home = False
+
+    # Get the path for storing the updated refgenie_genomes.config
+    if hasattr(rgc, "nextflow_config"):
+        refgenie_genomes_config_file = rgc.nextflow_config
+    elif "NXF_REFGENIE_PATH" in os.environ:
+        refgenie_genomes_config_file = os.environ.get("NXF_REFGENIE_PATH")
+    elif nxf_home:
+        refgenie_genomes_config_file = Path(nxf_home) / "nf-core/refgenie_genomes.config"
+    else:
+        log.info("Could not determine path to 'refgenie_genomes.config' file.")
+        return False
+
+    # Save the updated genome config
+    try:
+        with open(refgenie_genomes_config_file, "w") as fh:
+            fh.write(refgenie_genomes)
+        log.info(f"Updated nf-core genomes config: {refgenie_genomes_config_file}")
+    except FileNotFoundError as e:
+        log.warn(f"Could not write to {refgenie_genomes_config_file}")
+        return False
+
+    # Add include statement to NXF_HOME/config
+    if nxf_home:
+        _update_nextflow_home_config(refgenie_genomes_config_file, nxf_home)
+
+    return True

requirements-dev.txt

@@ -2,6 +2,7 @@ pytest-datafiles
 pytest-cov
 mock
 black
+refgenie
 Sphinx
 sphinx_rtd_theme
 isort

setup.py

@@ -31,7 +31,10 @@
     author_email="phil.ewels@scilifelab.se",
     url="https://github.com/nf-core/tools",
     license="MIT",
-    entry_points={"console_scripts": ["nf-core=nf_core.__main__:run_nf_core"]},
+    entry_points={
+        "console_scripts": ["nf-core=nf_core.__main__:run_nf_core"],
+        "refgenie.hooks.post_update": ["nf-core-refgenie=nf_core.refgenie:update_config"],
+    },
     install_requires=required,
     packages=find_packages(exclude=("docs")),
     include_package_data=True,

tests/test_refgenie.py

@@ -0,0 +1,58 @@
+#!/usr/bin/env python
+""" Tests covering the refgenie integration code
+"""
+
+import os
+import shlex
+import subprocess
+import tempfile
+import unittest
+
+import nf_core.refgenie
+
+
+class TestRefgenie(unittest.TestCase):
+    """Class for refgenie tests"""
+
+    def setUp(self):
+        """
+        Prepare a refgenie config file
+        """
+        self.tmp_dir = tempfile.mkdtemp()
+        self.NXF_HOME = os.path.join(self.tmp_dir, ".nextflow")
+        self.NXF_REFGENIE_PATH = os.path.join(self.NXF_HOME, "nf-core", "refgenie_genomes.config")
+        self.REFGENIE = os.path.join(self.tmp_dir, "genomes_config.yaml")
+        # Set NXF_HOME environment variable
+        # avoids adding includeConfig statement to config file outside the current tmpdir
+        try:
+            self.NXF_HOME_ORIGINAL = os.environ["NXF_HOME"]
+        except:
+            self.NXF_HOME_ORIGINAL = None
+        os.environ["NXF_HOME"] = self.NXF_HOME
+
+        # create NXF_HOME and nf-core directories
+        os.makedirs(os.path.join(self.NXF_HOME, "nf-core"), exist_ok=True)
+
+        # Initialize a refgenie config
+        os.system(f"refgenie init -c {self.REFGENIE}")
+
+        # Add NXF_REFGENIE_PATH to refgenie config
+        with open(self.REFGENIE, "a") as fh:
+            fh.write(f"nextflow_config: {os.path.join(self.NXF_REFGENIE_PATH)}\n")
+
+    def tearDown(self) -> None:
+        # Remove the tempdir again
+        os.system(f"rm -rf {self.tmp_dir}")
+        # Reset NXF_HOME environment variable
+        if self.NXF_HOME_ORIGINAL is None:
+            del os.environ["NXF_HOME"]
+        else:
+            os.environ["NXF_HOME"] = self.NXF_HOME_ORIGINAL
+
+    def test_update_refgenie_genomes_config(self):
+        """Test that listing pipelines works"""
+        # Populate the config with a genome
+        cmd = f"refgenie pull t7/fasta -c {self.REFGENIE}"
+        out = subprocess.check_output(shlex.split(cmd), stderr=subprocess.STDOUT)
+
+        assert "Updated nf-core genomes config" in str(out)