ENH: add fetch-eggnog-proteins action

q2_moshpit/_utils.py

@@ -8,6 +8,27 @@
 import subprocess
 from typing import List
 
+# CONVERT shell colors to the same curses palette
+SHELL_COLORS = {
+    "wr": '\033[1;37;41m',  # white on red
+    "wo": '\033[1;37;43m',  # white on orange
+    "wm": '\033[1;37;45m',  # white on magenta
+    "wb": '\033[1;37;46m',  # white on blue
+    "bw": '\033[1;37;40m',  # black on white
+    "lblue": '\033[1;34m',  # light blue
+    "lred": '\033[1;31m',  # light red
+    "lgreen": '\033[1;32m',  # light green
+    "yellow": '\033[1;33m',  # yellow
+    "cyan": '\033[36m',  # cyan
+    "blue": '\033[34m',  # blue
+    "green": '\033[32m',  # green
+    "orange": '\033[33m',  # orange
+    "red": '\033[31m',  # red
+    "magenta": "\033[35m",  # magenta
+    "white": "\033[0m",  # white
+    None: "\033[0m",  # end
+}
+
 
 def run_command(cmd, env=None, verbose=True, pipe=False):
     if verbose:
@@ -70,3 +91,7 @@ def _process_common_input_params(processing_func, params: dict) -> List[str]:
             continue
 
     return processed_args
+
+
+def colorify(string, color):
+    return "%s%s%s" % (SHELL_COLORS[color], string, SHELL_COLORS[None])

q2_moshpit/eggnog/__init__.py

@@ -7,6 +7,10 @@
 # ----------------------------------------------------------------------------
 
 
-from ._method import (eggnog_diamond_search, eggnog_annotate, fetch_eggnog_db)
+from ._method import (eggnog_diamond_search, eggnog_annotate)
+from ._dbs import fetch_eggnog_db, fetch_eggnog_fasta
 
-__all__ = ['eggnog_diamond_search', 'eggnog_annotate', 'fetch_eggnog_db']
+__all__ = [
+    'eggnog_diamond_search', 'eggnog_annotate', 'fetch_eggnog_db',
+    'fetch_eggnog_fasta'
+]

q2_moshpit/eggnog/_dbs.py

@@ -0,0 +1,89 @@
+# ----------------------------------------------------------------------------
+# Copyright (c) 2022, QIIME 2 development team.
+#
+# Distributed under the terms of the Modified BSD License.
+#
+# The full license is in the file LICENSE, distributed with this software.
+# ----------------------------------------------------------------------------
+import os
+from q2_types_genomics.reference_db import (
+    EggnogRefDirFmt, EggnogSequenceTaxaDirFmt
+)
+from .._utils import run_command, colorify
+
+
+def fetch_eggnog_db() -> EggnogRefDirFmt:
+    """
+    Downloads eggnog reference database using the
+    `download_eggnog_data.py` script from eggNOG. Here, this
+    script downloads 3 files amounting to 47Gb in total.
+    """
+
+    # Initialize output objects
+    eggnog_db = EggnogRefDirFmt()
+
+    # Define command.
+    # Meaning of flags:
+    # y: Answer yest to all prompts thrown by download_eggnog_data.py
+    # D: Do not download the Diamond database
+    # data_dir: location where to save downloads
+    cmd = [
+        "download_eggnog_data.py", "-y", "-D",
+        "--data_dir", str(eggnog_db.path)
+    ]
+    run_command(cmd)
+
+    # Return objects
+    return eggnog_db
+
+
+def fetch_eggnog_fasta() -> EggnogSequenceTaxaDirFmt:
+    """
+    # TODO: Add description
+    """
+    # Initialize output objects
+    eggnog_fa = EggnogSequenceTaxaDirFmt()
+    fasta_file = os.path.join(str(eggnog_fa), "e5.proteomes.faa")
+    taxonomy_file = os.path.join(str(eggnog_fa), "e5.taxid_info.tsv")
+
+    # Download Diamond DB
+    print(
+        colorify(
+            "Downloading fasta file...", "lgreen"
+        )
+    )
+    run_command(
+        cmd=[
+            "wget", "-e", "robots=off", "-O", f"{fasta_file}",
+            "http://eggnog5.embl.de/download/eggnog_5.0/e5.proteomes.faa"
+        ]
+    )
+
+    # Decompressing file
+    print(
+        colorify(
+            "Download completed.\n"
+            "Downloading taxonomy file...",
+            "lgreen"
+        )
+    )
+    run_command(
+        cmd=[
+            "wget", "-e", "robots=off", "-O", f"{taxonomy_file}",
+            "http://eggnog5.embl.de/download/eggnog_5.0/e5.taxid_info.tsv"
+        ]
+    )
+
+    # Let user know that the process is done.
+    # The actual copying wil be taken care of by qiime behind the
+    # scenes.
+    print(
+        colorify(
+            "Download completed. \n"
+            "Copying files from temporary directory to final location "
+            "(this will take a few minutes)...",
+            "lgreen"
+        )
+    )
+
+    return eggnog_fa

q2_moshpit/eggnog/_method.py

@@ -17,7 +17,6 @@
 from q2_types_genomics.reference_db import EggnogRefDirFmt
 from q2_types.feature_data import DNAFASTAFormat
 from q2_types_genomics.reference_db import DiamondDatabaseDirFmt
-from .._utils import run_command
 from q2_types_genomics.feature_data import (
     OrthologAnnotationDirFmt, MAGSequencesDirFmt
 )
@@ -132,28 +131,3 @@ def _annotate_seed_orthologs_runner(seed_ortholog, eggnog_db, sample_label,
         cmds.append('--dbmem')
 
     subprocess.run(cmds, check=True)
-
-
-def fetch_eggnog_db() -> EggnogRefDirFmt:
-    """
-    Downloads eggnog reference database using the
-    `download_eggnog_data.py` script from eggNOG. Here, this
-    script downloads 3 files amounting to 47Gb in total.
-    """
-
-    # Initialize output objects
-    eggnog_db = EggnogRefDirFmt()
-
-    # Define command.
-    # Meaning of flags:
-    # y: Answer yest to all prompts thrown by download_eggnog_data.py
-    # D: Do not download the Diamond database
-    # data_dir: location where to save downloads
-    cmd = [
-        "download_eggnog_data.py", "-y", "-D",
-        "--data_dir", str(eggnog_db.path)
-    ]
-    run_command(cmd)
-
-    # Return objects
-    return eggnog_db

q2_moshpit/eggnog/tests/test_dbs.py

@@ -0,0 +1,55 @@
+# ----------------------------------------------------------------------------
+# Copyright (c) 2022, QIIME 2 development team.
+#
+# Distributed under the terms of the Modified BSD License.
+#
+# The full license is in the file LICENSE, distributed with this software.
+# ----------------------------------------------------------------------------
+import os
+from unittest.mock import patch, call
+from qiime2.plugin.testing import TestPluginBase
+from .._dbs import fetch_eggnog_db, fetch_eggnog_fasta
+
+
+class TestFetchDB(TestPluginBase):
+    package = 'q2_moshpit.eggnog.tests'
+
+    @patch("subprocess.run")
+    def test_fetch_eggnog_db(self, subp_run):
+        # Call function. Patching will make sure nothing is
+        # actually ran
+        eggnog_db = fetch_eggnog_db()
+
+        # Check that command was called in the expected way
+        cmd = [
+            "download_eggnog_data.py", "-y", "-D",
+            "--data_dir", str(eggnog_db)
+        ]
+        subp_run.assert_called_once_with(cmd, check=True)
+
+    @patch("subprocess.run")
+    def test_fetch_eggnog_fasta(self, subp_run):
+        # Call function. Patching will make sure nothing is
+        # actually ran
+        eggnog_fa = fetch_eggnog_fasta()
+        fasta_file = os.path.join(str(eggnog_fa), "e5.proteomes.faa")
+        taxonomy_file = os.path.join(str(eggnog_fa), "e5.taxid_info.tsv")
+
+        # Check that command was called in the expected way
+        first_call = call(
+            [
+                "wget", "-e", "robots=off", "-O", f"{fasta_file}",
+                "http://eggnog5.embl.de/download/eggnog_5.0/e5.proteomes.faa"
+            ],
+            check=True
+        )
+        second_call = call(
+            [
+                "wget", "-e", "robots=off", "-O", f"{taxonomy_file}",
+                "http://eggnog5.embl.de/download/eggnog_5.0/e5.taxid_info.tsv"
+            ],
+            check=True,
+        )
+
+        # Check that commands are ran as expected
+        subp_run.assert_has_calls([first_call, second_call], any_order=False)

q2_moshpit/eggnog/tests/test_method.py

@@ -5,15 +5,12 @@
 #
 # The full license is in the file LICENSE, distributed with this software.
 # ----------------------------------------------------------------------------
-
 import pandas as pd
 import pandas.testing as pdt
-from unittest.mock import patch
 import qiime2
 from qiime2.plugin.testing import TestPluginBase
-
 from q2_types_genomics.feature_data import MAGSequencesDirFmt
-from .._method import eggnog_diamond_search, eggnog_annotate, fetch_eggnog_db
+from .._method import eggnog_diamond_search, eggnog_annotate
 from q2_types_genomics.reference_db import (
     DiamondDatabaseDirFmt, EggnogRefDirFmt)
 from q2_types_genomics.per_sample_data import ContigSequencesDirFmt
@@ -85,20 +82,3 @@ def test_small_good_hits(self):
         self.assertEqual(len(objs), 1)
         df = objs[0][1].view(pd.DataFrame)
         pdt.assert_frame_equal(df, exp)
-
-
-class TestFetchDB(TestPluginBase):
-    package = 'q2_moshpit.eggnog.tests'
-
-    @patch("subprocess.run")
-    def test_fetch_eggnog_db(self, subp_run):
-        # Call function. Patching will make sure nothing is
-        # actually ran
-        eggnog_db = fetch_eggnog_db()
-
-        # Check that command was called in the expected way
-        cmd = [
-            "download_eggnog_data.py", "-y", "-D",
-            "--data_dir", str(eggnog_db)
-        ]
-        subp_run.assert_called_once_with(cmd, check=True)

q2_moshpit/plugin_setup.py

@@ -31,7 +31,9 @@
 from q2_types_genomics.kraken2._type import BrackenDB
 from q2_types_genomics.per_sample_data import MAGs, Contigs
 from q2_types_genomics.per_sample_data._type import AlignmentMap
-from q2_types_genomics.reference_db import ReferenceDB, Diamond, Eggnog
+from q2_types_genomics.reference_db import (
+    ReferenceDB, Diamond, Eggnog, EggnogSequenceTaxa
+)
 
 citations = Citations.load('citations.bib', package='q2_moshpit')
 
@@ -505,6 +507,25 @@
                 "storage space is required to run this action. "
 )
 
+plugin.methods.register_function(
+    function=q2_moshpit.eggnog.fetch_eggnog_fasta,
+    inputs={},
+    parameters={},
+    outputs=[("eggnog_fasta", ReferenceDB[EggnogSequenceTaxa])],
+    output_descriptions={
+        "eggnog_fasta": "Artifact containing the eggNOG database "
+                        "of protein sequences and their corresponding"
+                        "taxonomy information."
+    },
+    name="Fetch the databases necessary to run to run the "
+         "build-eggnog-diamond-db action.",
+    description="Downloads eggnog proteome database  "
+                "This script downloads 2 files: "
+                "(e5.proteomes.faa and e5.taxid_info.tsv) "
+                "and creates and artifact with them. At least 18 Gb of "
+                "storage space is required to run this action. "
+)
+
 plugin.methods.register_function(
     function=q2_moshpit.eggnog.eggnog_diamond_search,
     inputs={