From 854291cb450558b2be86b8cce47f0c390b393ed8 Mon Sep 17 00:00:00 2001
From: Amandine PERRIN <amandine.perrin@pasteur.fr>
Date: Wed, 19 Aug 2020 13:28:21 +0200
Subject: [PATCH] Adapt tests for prokka format
---
.../prokka_out_for_test.tbl | 1 -
.../prokka_out_tbl_changed-contnames.tbl | 78 +++++++++++++++++++
.../test_annotate/test_format_prokka.py | 47 ++++++++---
3 files changed, 116 insertions(+), 10 deletions(-)
create mode 100644 test/data/annotate/test_files/prokka_out_tbl_changed-contnames.tbl
diff --git a/test/data/annotate/test_files/original_name.fna-prokkaRes/prokka_out_for_test.tbl b/test/data/annotate/test_files/original_name.fna-prokkaRes/prokka_out_for_test.tbl
index 9b9cb50c..1bc2b52b 100755
--- a/test/data/annotate/test_files/original_name.fna-prokkaRes/prokka_out_for_test.tbl
+++ b/test/data/annotate/test_files/original_name.fna-prokkaRes/prokka_out_for_test.tbl
@@ -52,7 +52,6 @@
product hypothetical protein
db_xref 12345
>Feature test.0417.00002.0005
->Feature test.0417.00002.0006
>Feature test.0417.00002.0007
3039279 3039355 tRNA
inference COORDINATES:profile:Aragorn:1.2
diff --git a/test/data/annotate/test_files/prokka_out_tbl_changed-contnames.tbl b/test/data/annotate/test_files/prokka_out_tbl_changed-contnames.tbl
new file mode 100644
index 00000000..e8755cbc
--- /dev/null
+++ b/test/data/annotate/test_files/prokka_out_tbl_changed-contnames.tbl
@@ -0,0 +1,78 @@
+>Feature toto_1
+287 787 CDS
+ inference ab initio prediction:Prodigal:2.6
+ locus_tag JGIKIPIJ_00001
+ product hypothetical protein
+4416 6068 CDS
+ gene yiaD
+ inference ab initio prediction:Prodigal:2.6
+ inference similar to AA sequence:UniProtKB:P37665
+ locus_tag JGIKIPIJ_00005
+ product putative lipoprotein YiaD
+12000 9000 CDS
+ EC_number 6.3.2.-
+ gene vgrG1
+ inference ab initio prediction:Prodigal:2.6
+ inference similar to AA sequence:UniProtKB:Q9KS45
+ locus_tag JGIKIPIJ_00008
+ product Actin cross-linking toxin VgrG1
+>Feature toto_2
+77 1237 CDS
+ inference ab initio prediction:Prodigal:2.6
+ locus_tag JGIKIPIJ_00009
+ product hypothetical protein
+1279 2346 CDS
+ inference ab initio prediction:Prodigal:2.6
+ locus_tag JGIKIPIJ_00010
+ product hypothetical protein
+2419 3000 CDS
+ inference ab initio prediction:Prodigal:2.6
+ locus_tag JGIKIPIJ_00011
+ product hypothetical protein
+3500 4000 CDS
+ inference ab initio prediction:Prodigal:2.6
+ locus_tag JGIKIPIJ_00011
+ product hypothetical protein
+5000 4632 CDS
+ inference ab initio prediction:Prodigal:2.6
+ locus_tag JGIKIPIJ_00011
+ product hypothetical protein
+>Feature toto_3
+2968265 296902 repeat_region
+ rpt_family CRISPR
+ score 13
+3399 4538 CDS
+ inference ab initio prediction:Prodigal:2.6
+ locus_tag JGIKIPIJ_00013
+ product hypothetical protein
+>Feature toto_4
+4535 7888 CDS
+ inference ab initio prediction:Prodigal:2.6
+ locus_tag JGIKIPIJ_00014
+ product hypothetical protein
+ db_xref 12345
+>Feature toto_5
+>Feature toto_7
+3039279 3039355 tRNA
+ inference COORDINATES:profile:Aragorn:1.2
+ locus_tag PROKKA_02927
+ product tRNA-Met(cat)
+3039389 3039465 tRNA
+ inference COORDINATES:profile:Aragorn:1.2
+ locus_tag PROKKA_02928
+ product tRNA-Met(cat)
+7854 9491 CDS
+ inference ab initio prediction:Prodigal:2.6
+ locus_tag JGIKIPIJ_03015
+ product hypothetical protein
+9525 11285 CDS
+ inference ab initio prediction:Prodigal:2.6
+ locus_tag JGIKIPIJ_03016
+ product hypothetical protein
+2994578 2994971 repeat_region
+ rpt_family CRISPR
+ score 7
+11249 12328 CDS
+ inference ab initio prediction:Prodigal:2.6
+ locus_tag JGIKIPIJ_03017
+ product hypothetical protein
diff --git a/test/test_unit/test_annotate/test_format_prokka.py b/test/test_unit/test_annotate/test_format_prokka.py
index fa606818..0310f1c1 100644
--- a/test/test_unit/test_annotate/test_format_prokka.py
+++ b/test/test_unit/test_annotate/test_format_prokka.py
@@ -15,14 +15,13 @@ import PanACoTA.annotate_module.format_prokka as prokkafunc
import PanACoTA.utils as utils
import test.test_unit.utilities_for_tests as tutil
-logger = logging.getLogger("test_prokka")
-
-
ANNOTEDIR = os.path.join("test", "data", "annotate")
EXP_ANNOTE = os.path.join(ANNOTEDIR, "exp_files")
TEST_ANNOTE = os.path.join(ANNOTEDIR, "test_files")
GENEPATH = os.path.join(ANNOTEDIR, "generated_by_unit-tests")
+LOGFILE_BASE = os.path.join(GENEPATH, "logfile")
+LOGFILES = [LOGFILE_BASE + ext for ext in [".log", ".log.debug", ".log.details", ".log.err"]]
@pytest.fixture(autouse=True)
def setup_teardown_module():
@@ -37,19 +36,48 @@ def setup_teardown_module():
- remove all log files
- remove directory with generated results
"""
- # utils.init_logger(LOGFILE_BASE, 0, 'test_fastme', verbose=1)
os.mkdir(GENEPATH)
+ utils.init_logger(LOGFILE_BASE, 0, 'test_fastme', verbose=1)
print("setup")
yield
- # for f in LOGFILES:
- # if os.path.exists(f):
- # os.remove(f)
+ for f in LOGFILES:
+ if os.path.exists(f):
+ os.remove(f)
# shutil.rmtree(GENEPATH)
print("teardown")
-def test_tbl_to_lst(caplog):
+def test_tbl_to_lst_changed_names(caplog):
+ """
+ Check that generated lstinfo file is as expected, when the genome name is the same as
+ it already was in the genome given to prokka.
+ The test tblfile contains the following aspects:
+ - gene in D strand (start < end)
+ - gene in C strand (start > end)
+ - CDS features (some with all info = ECnumber, gene name, product etc. ;
+ some with missing info)
+ - tRNA type
+ - repeat_region type (*2)
+ - contigs with more than 2 genes
+ - contig with only 2 genes (both 'b' loc)
+ - contig with 1 gene ('b' loc)
+ - contig without gene (should be skipped)
+ """
+ caplog.set_level(logging.DEBUG)
+ logger = logging.getLogger("test_prokka")
+ tblfile = os.path.join(TEST_ANNOTE, "prokka_out_tbl_changed-contnames.tbl")
+ lstfile = os.path.join(GENEPATH, "res_test_tbl2lst.lst")
+ contigs = ["test.0417.00002.0001\t50", "test.0417.00002.0002\t50", "test.0417.00002.0003\t50",
+ "test.0417.00002.0004\t50", "test.0417.00002.0005\t50", "test.0417.00002.0006\t50",
+ "test.0417.00002.0007\t50"]
+ name = "test.0417.00002"
+ assert prokkafunc.tbl2lst(tblfile, lstfile, contigs, name, logger, changed_name=True)
+ exp_lst = os.path.join(EXP_ANNOTE, "res_tbl2lst.lst")
+ assert tutil.compare_order_content(exp_lst, lstfile)
+
+
+def test_tbl_to_lst_not_changed_names(caplog):
"""
Check that generated lstinfo file is as expected, when the genome name is the same as
it already was in the genome given to prokka.
@@ -66,13 +94,14 @@ def test_tbl_to_lst(caplog):
- contig without gene (should be skipped)
"""
caplog.set_level(logging.DEBUG)
+ logger = logging.getLogger("test_prokka")
tblfile = os.path.join(TEST_ANNOTE, "original_name.fna-prokkaRes", "prokka_out_for_test.tbl")
lstfile = os.path.join(GENEPATH, "res_test_tbl2lst.lst")
contigs = ["test.0417.00002.0001\t50", "test.0417.00002.0002\t50", "test.0417.00002.0003\t50",
"test.0417.00002.0004\t50", "test.0417.00002.0005\t50", "test.0417.00002.0006\t50",
"test.0417.00002.0007\t50"]
name = "test.0417.00002"
- assert prokkafunc.tbl2lst(tblfile, lstfile, contigs, name)
+ assert prokkafunc.tbl2lst(tblfile, lstfile, contigs, name, logger, changed_name=False)
exp_lst = os.path.join(EXP_ANNOTE, "res_tbl2lst.lst")
assert tutil.compare_order_content(exp_lst, lstfile)
--
GitLab