From 7f863a0d6c145ec442ca4dcfb8a55f4bebfb80e4 Mon Sep 17 00:00:00 2001
From: Kenzo-Hugo Hillion <kenzo-hugo.hillion1@pasteur.fr>
Date: Mon, 9 Dec 2019 17:37:00 +0100
Subject: [PATCH] add eggnog line parser
---
.../common/utils/parsers/__init__.py | 1 +
.../metagenedb/common/utils/parsers/eggnog.py | 22 +++++++++++++++++++
.../common/utils/parsers/test_eggnog.py | 21 ++++++++++++++++++
3 files changed, 44 insertions(+)
create mode 100644 backend/metagenedb/common/utils/parsers/eggnog.py
create mode 100644 backend/metagenedb/common/utils/parsers/test_eggnog.py
diff --git a/backend/metagenedb/common/utils/parsers/__init__.py b/backend/metagenedb/common/utils/parsers/__init__.py
index 7c8b8f5..92f59fa 100644
--- a/backend/metagenedb/common/utils/parsers/__init__.py
+++ b/backend/metagenedb/common/utils/parsers/__init__.py
@@ -1,3 +1,4 @@
+from .eggnog import EggNogAnnotationLineParser # noqa
from .igc import IGCLineParser # noqa
from .kegg import KEGGLineParser # noqa
from .ncbi_taxonomy import NCBITaxonomyLineParser # noqa
diff --git a/backend/metagenedb/common/utils/parsers/eggnog.py b/backend/metagenedb/common/utils/parsers/eggnog.py
new file mode 100644
index 0000000..4397493
--- /dev/null
+++ b/backend/metagenedb/common/utils/parsers/eggnog.py
@@ -0,0 +1,22 @@
+import logging
+
+_LOGGER = logging.getLogger(__name__)
+
+
+class EggNogAnnotationLineParser(object):
+
+ @staticmethod
+ def ko_list(line):
+ """
+ Parse line from Eggnog annotations.tsv file to return organized dict
+ """
+ try:
+ elements = line.split('\t')
+ return {
+ 'functional_category': elements[2],
+ 'function_id': elements[1],
+ 'name': elements[3],
+ }
+ except Exception:
+ _LOGGER.error(f"Could not parse: {line.rstrip()}. Are you sure it comes from eggnog annotations.tsv?")
+ raise
diff --git a/backend/metagenedb/common/utils/parsers/test_eggnog.py b/backend/metagenedb/common/utils/parsers/test_eggnog.py
new file mode 100644
index 0000000..0f48ef7
--- /dev/null
+++ b/backend/metagenedb/common/utils/parsers/test_eggnog.py
@@ -0,0 +1,21 @@
+from unittest import TestCase
+
+from metagenedb.common.utils.parsers import EggNogAnnotationLineParser
+
+
+class TestEggNogAnnotationLineParser(TestCase):
+
+ def test_ko_list(self):
+ ko_line = "1\t28H54\tK\ttranslational termination"
+ expected_dict = {
+ 'function_id': "28H54",
+ 'name': "translational termination",
+ 'functional_category': "K"
+ }
+ test_dict = EggNogAnnotationLineParser.ko_list(ko_line)
+ self.assertDictEqual(test_dict, expected_dict)
+
+ def test_ko_list_wrong_format(self):
+ ko_line = "This is a wrong line format, with; information and tab"
+ with self.assertRaises(Exception) as context: # noqa
+ EggNogAnnotationLineParser.ko_list(ko_line)
--
GitLab