Added unit tests for check_sample_and_mag_validity.py

tests/scripts/fixtures/check_sample_and_mag_validity/expected_result_no_remove_list.txt

@@ -0,0 +1 @@
+GCA_01526150500

tests/scripts/fixtures/check_sample_and_mag_validity/ftp/genomes-all_metadata.tsv

@@ -0,0 +1,10 @@
+Genome	Genome_type	Length	N_contigs	N50	GC_content	Completeness	Contamination	rRNA_5S	rRNA_16S	rRNA_23S	tRNAs	Genome_accession	Species_rep	Lineage	Sample_accession	Study_accession	Country	Continent	FTP_download
+MGYG000298000	MAG	1829829	675	3325	56.91	81.97	3.27	93.28	20.35	5.20	19	GCA_015265315	MGYG000298000	d__Bacteria;p__Firmicutes_C;c__Negativicutes;o__Selenomonadales;f__Selenomonadaceae;g__Centipeda;s__	SAMN14571041	SRP311368	USA	North America	ftp://ftp.ebi.ac.uk/pub/databases/metagenomics/mgnify_genomes/human-oral/v1.0/all_genomes/MGYG0002992/MGYG000299223/genomes1/MGYG000298000.gff.gz
+MGYG000298001	MAG	2546294	85	45191	37.09	96.30	0.00	0.00	0.00	0.00	15	GCA_01526150500	MGYG000298001	d__Bacteria;p__Bacteroidota;c__Bacteroidia;o__Bacteroidales;f__Bacteroidaceae;g__Prevotella;s__Prevotella pallens	SAMN14570949	SRP311368	USA	North America	ftp://ftp.ebi.ac.uk/pub/databases/metagenomics/mgnify_genomes/human-oral/v1.0/all_genomes/MGYG0002981/MGYG000298159/genomes1/MGYG000298001.gff.gz
+MGYG000298002	MAG	1481714	84	28959	38.14	95.74	0.71	0.00	0.00	0.00	18	CAJPUF01	MGYG000298013	d__Bacteria;p__Firmicutes_A;c__Clostridia;o__Peptostreptococcales;f__Anaerovoracaceae;g__Eubacterium_M;s__Eubacterium_M brachy	ERS6080781	ERP127228	not provided	not provided	ftp://ftp.ebi.ac.uk/pub/databases/metagenomics/mgnify_genomes/human-oral/v1.0/all_genomes/MGYG0002980/MGYG000298013/genomes1/MGYG000298002.gff.gz
+MGYG000298003	MAG	1494829	454	4402	40.13	80.42	0.75	0.00	0.00	0.00	7	GCA_015264635	MGYG000298003	d__Bacteria;p__Firmicutes_C;c__Negativicutes;o__Veillonellales;f__Veillonellaceae;g__Veillonella;s__Veillonella atypica	SAMN14571074	SRP311368	USA	North America	ftp://ftp.ebi.ac.uk/pub/databases/metagenomics/mgnify_genomes/human-oral/v1.0/all_genomes/MGYG0002980/MGYG000298071/genomes1/MGYG000298003.gff.gz
+MGYG000298004	MAG	1185879	236	5569	39.69	69.63	1.50	0.00	0.00	0.00	6	CAJPIW01	MGYG000299205	d__Bacteria;p__Firmicutes_C;c__Negativicutes;o__Veillonellales;f__Veillonellaceae;g__Veillonella;s__Veillonella parvula_A	ERS6076640	ERP124374	not provided	not provided	ftp://ftp.ebi.ac.uk/pub/databases/metagenomics/mgnify_genomes/human-oral/v1.0/all_genomes/MGYG0002992/MGYG000299205/genomes1/MGYG000298004.gff.gz
+MGYG000298005	MAG	1194299	554	2401	45.56	73.10	2.18	0.00	0.00	0.00	11	GCA_015263975	MGYG000299273	d__Bacteria;p__Firmicutes_C;c__Negativicutes;o__Veillonellales;f__Megasphaeraceae;g__Anaeroglobus;s__Anaeroglobus micronuciformis	SAMN14570822	SRP311368	USA	North America	ftp://ftp.ebi.ac.uk/pub/databases/metagenomics/mgnify_genomes/human-oral/v1.0/all_genomes/MGYG0002992/MGYG000299273/genomes1/MGYG000298005.gff.gz
+MGYG000298006	MAG	1160570	36	95371	55.48	68.42	0.00	94.12	21.59	30.05	11	GCA_015259755	MGYG000298453	d__Bacteria;p__Firmicutes_C;c__Negativicutes;o__Selenomonadales;f__Selenomonadaceae;g__Centipeda;s__Centipeda noxia	SAMN14571040	SRP311368	USA	North America	ftp://ftp.ebi.ac.uk/pub/databases/metagenomics/mgnify_genomes/human-oral/v1.0/all_genomes/MGYG0002984/MGYG000298453/genomes1/MGYG000298006.gff.gz
+MGYG000298007	MAG	2493662	476	5753	68.18	75.36	2.21	0.00	0.00	0.00	19	CAJZFF01	MGYG000298105	d__Bacteria;p__Actinobacteriota;c__Actinomycetia;o__Actinomycetales;f__Actinomycetaceae;g__Actinomyces;s__Actinomyces oris	ERS7599769	ERP108069	China	Asia	ftp://ftp.ebi.ac.uk/pub/databases/metagenomics/mgnify_genomes/human-oral/v1.0/all_genomes/MGYG0002981/MGYG000298105/genomes1/MGYG000298007.gff.gz
+MGYG000298008	MAG	1230350	184	7701	39.43	75.83	0	0.00	0.00	0.00	12	CAJPLM01	MGYG000298156	d__Bacteria;p__Firmicutes_C;c__Negativicutes;o__Veillonellales;f__Veillonellaceae;g__Veillonella;s__Veillonella rogosae	ERS6079605	ERP127558	not provided	not provided	ftp://ftp.ebi.ac.uk/pub/databases/metagenomics/mgnify_genomes/human-oral/v1.0/all_genomes/MGYG0002981/MGYG000298156/genomes1/MGYG000298008.gff.gz

tests/scripts/fixtures/check_sample_and_mag_validity/output1.txt

@@ -0,0 +1 @@
+GCA_01526150500

tests/scripts/fixtures/check_sample_and_mag_validity/remove_list.txt

@@ -0,0 +1,3 @@
+GCA_01526150500
+CAJZFF01
+GCA_015263975

tests/scripts/test_check_sample_and_mag_validity.py

@@ -0,0 +1,32 @@
+import filecmp
+import importlib
+import unittest
+import os
+
+from pathlib import Path
+
+script_dir = Path(__file__).resolve().parent
+
+check_validity = importlib.import_module("bin.check_sample_and_mag_validity")
+check_validity_main = check_validity.main
+
+
+class TestMAGValidityScript(unittest.TestCase):
+    def setUp(self):
+        self.test_data_dir = os.path.join(script_dir, "fixtures", "check_sample_and_mag_validity")
+        self.removal_file = os.path.join(self.test_data_dir, "remove_list.txt")
+        self.outfile1 = os.path.join(self.test_data_dir, "output1.txt")
+        self.outfile2 = os.path.join(self.test_data_dir, "output2.txt")
+        self.expected_file_no_remove_list = os.path.join(self.test_data_dir, "expected_result_no_remove_list.txt")
+
+    def test_empty_remove_list(self):
+        check_validity_main(self.test_data_dir, None, self.outfile1)
+        self.assertTrue(
+            filecmp.cmp(self.outfile1, self.expected_file_no_remove_list, shallow=False),
+            "The output file does not match the expected results"
+        )
+
+    def test_with_remove_list(self):
+        check_validity_main(self.test_data_dir, self.removal_file, self.outfile2)
+        self.assertFalse(os.path.exists(self.outfile2))
+