Add tests for resfinderv4.5 fastq output resolves #86

pha4ge · May 21, 2024 · 36557ba · 36557ba
1 parent 5852116
commit 36557ba
Show file tree

Hide file tree

Showing 6 changed files with 77 additions and 8 deletions.
diff --git a/test/data/dummy/resfinder/resfinder_inputfastq_inputfasta.txt b/test/data/dummy/resfinder/resfinder_inputfastq_inputfasta.txt
@@ -0,0 +1,2 @@
+Resistance gene	Identity	Alignment Length/Gene Length	Coverage	Position in reference	Contig	Position in contig	Phenotype	Accession no.
+aph(6)-Id	100.00	837/837	100.0	1..838	NA	NA..NA	Streptomycin	M28829
diff --git a/test/data/raw_outputs/resfinder/resfinder_inputfasta_tab.txt b/test/data/raw_outputs/resfinder/resfinder_inputfasta_tab.txt
@@ -0,0 +1,15 @@
+Resistance gene	Identity	Alignment Length/Gene Length	Coverage	Position in reference	Contig	Position in contig	Phenotype	Accession no.
+aph(6)-Id	100.00	837/837	100.0	1..837	CCI165_S85_contig_59 length 7808 coverage 252.0 normalized_cov 1.38	2438..3274	Streptomycin	M28829
+aph(3'')-Ib	100.00	803/804	99.87562189054727	2..804	CCI165_S85_contig_59 length 7808 coverage 252.0 normalized_cov 1.38	3274..4076	Streptomycin	AF024602
+aph(3'')-Ib	99.88	804/804	100.0	1..804	CCI165_S85_contig_59 length 7808 coverage 252.0 normalized_cov 1.38	3274..4077	Streptomycin	AF321551
+aph(3'')-Ib	99.88	804/804	100.0	1..804	CCI165_S85_contig_59 length 7808 coverage 252.0 normalized_cov 1.38	3274..4077	Streptomycin	AF313472
+aph(3'')-Ib	99.88	804/804	100.0	1..804	CCI165_S85_contig_59 length 7808 coverage 252.0 normalized_cov 1.38	3274..4077	Streptomycin	AF321550
+aadA1	100.00	792/792	100.0	1..792	CCI165_S85_contig_74 length 3262 coverage 232.9 normalized_cov 1.28	451..1242	Spectinomycin, Streptomycin	JX185132
+blaOKP-B-17	100.00	860/861	99.88385598141696	1..860	CCI165_S85_contig_2 length 348456 coverage 150.2 normalized_cov 0.82	263329..264188	Unknown Beta-lactam	AM850919
+fosA	93.81	420/420	100.0	1..420	CCI165_S85_contig_13 length 151149 coverage 198.6 normalized_cov 1.09	127705..128124	Fosfomycin	ACZD01000244
+OqxB	95.40	3153/3153	100.0	1..3153	CCI165_S85_contig_8 length 181163 coverage 173.9 normalized_cov 0.95	45285..48437	Chloramphenicol, Nalidixic acid, Ciprofloxacin, Trimethoprim	EU370913
+OqxA	93.62	1176/1176	100.0	1..1176	CCI165_S85_contig_8 length 181163 coverage 173.9 normalized_cov 0.95	44086..45261	Chloramphenicol, Nalidixic acid, Ciprofloxacin, Trimethoprim	EU370913
+sul2	100.00	816/816	100.0	1..816	CCI165_S85_contig_63 length 5843 coverage 215.4 normalized_cov 1.18	4351..5166	Sulfamethoxazole	AY034138
+dfrA1	99.79	474/474	100.0	1..474	CCI165_S85_contig_74 length 3262 coverage 232.9 normalized_cov 1.28	1335..1808	Trimethoprim	AF203818
+dfrA1	99.79	474/474	100.0	1..474	CCI165_S85_contig_74 length 3262 coverage 232.9 normalized_cov 1.28	1335..1808	Trimethoprim	AJ238350
+dfrA1	99.79	474/474	100.0	1..474	CCI165_S85_contig_74 length 3262 coverage 232.9 normalized_cov 1.28	1335..1808	Trimethoprim	X00926
diff --git a/test/data/raw_outputs/resfinder/resfinder_inputfastq_inputfasta.txt b/test/data/raw_outputs/resfinder/resfinder_inputfastq_inputfasta.txt
@@ -0,0 +1,14 @@
+Resistance gene	Identity	Alignment Length/Gene Length	Coverage	Position in reference	Contig	Position in contig	Phenotype	Accession no.
+aph(6)-Id	100.00	837/837	100.0	1..838	NA	NA..NA	Streptomycin	M28829
+aph(3'')-Ib	93.78	804/804	93.78	1..805	NA	NA..NA	Streptomycin	AF321550
+aadA1	100.00	792/792	100.0	1..793	NA	NA..NA	Spectinomycin, Streptomycin	JX185132
+aph(3'')-Ib	99.88	803/804	99.88	2..805	NA	NA..NA	Streptomycin	AF024602
+aph(3'')-Ib	82.71	667/804	82.96	1..668	NA	NA..NA	Streptomycin	AF321551
+blaOKP-B-17	99.88	861/861	100.0	1..862	NA	NA..NA	Unknown Beta-lactam	AM850919
+fosA5	93.33	420/420	100.0	1..421	NA	NA..NA	Fosfomycin	EU195449
+fosA	93.57	420/420	100.0	1..421	NA	NA..NA	Fosfomycin	AFBO01000747
+fosA	93.81	420/420	100.0	1..421	NA	NA..NA	Fosfomycin	ACZD01000244
+OqxA	93.62	1176/1176	100.0	1..1177	NA	NA..NA	Chloramphenicol, Nalidixic acid, Ciprofloxacin, Trimethoprim	EU370913
+OqxB	95.40	3153/3153	100.0	1..3154	NA	NA..NA	Chloramphenicol, Nalidixic acid, Ciprofloxacin, Trimethoprim	EU370913
+sul2	100.00	816/816	100.0	1..817	NA	NA..NA	Sulfamethoxazole	AY034138
+dfrA1	96.41	458/474	96.62	1..459	NA	NA..NA	Trimethoprim	EU089668
diff --git a/test/data/raw_outputs/resfinder/resfinder_inputfastq_tab.txt b/test/data/raw_outputs/resfinder/resfinder_inputfastq_tab.txt
@@ -0,0 +1,14 @@
+Resistance gene	Identity	Alignment Length/Gene Length	Coverage	Position in reference	Contig	Position in contig	Phenotype	Accession no.
+aph(6)-Id	100.00	837/837	100.0	1..838	NA	NA..NA	Streptomycin	M28829
+aph(3'')-Ib	93.78	804/804	93.78	1..805	NA	NA..NA	Streptomycin	AF321550
+aadA1	100.00	792/792	100.0	1..793	NA	NA..NA	Spectinomycin, Streptomycin	JX185132
+aph(3'')-Ib	99.88	803/804	99.88	2..805	NA	NA..NA	Streptomycin	AF024602
+aph(3'')-Ib	82.71	667/804	82.96	1..668	NA	NA..NA	Streptomycin	AF321551
+blaOKP-B-17	99.88	861/861	100.0	1..862	NA	NA..NA	Unknown Beta-lactam	AM850919
+fosA5	93.33	420/420	100.0	1..421	NA	NA..NA	Fosfomycin	EU195449
+fosA	93.57	420/420	100.0	1..421	NA	NA..NA	Fosfomycin	AFBO01000747
+fosA	93.81	420/420	100.0	1..421	NA	NA..NA	Fosfomycin	ACZD01000244
+OqxA	93.62	1176/1176	100.0	1..1177	NA	NA..NA	Chloramphenicol, Nalidixic acid, Ciprofloxacin, Trimethoprim	EU370913
+OqxB	95.40	3153/3153	100.0	1..3154	NA	NA..NA	Chloramphenicol, Nalidixic acid, Ciprofloxacin, Trimethoprim	EU370913
+sul2	100.00	816/816	100.0	1..817	NA	NA..NA	Sulfamethoxazole	AY034138
+dfrA1	96.41	458/474	96.62	1..459	NA	NA..NA	Trimethoprim	EU089668
diff --git a/test/run_integration_test.sh b/test/run_integration_test.sh
@@ -13,8 +13,8 @@ hamronize ariba data/raw_outputs/ariba/report.tsv --reference_database_version d
 
 hamronize amrfinderplus --input_file_name amrfinderplus_nucleotide_report --analysis_software_version AFP_nt_v1 --reference_database_version db_v_1 data/raw_outputs/amrfinderplus/report_nucleotide.tsv --format json --output hamronized_amrfinderplus_nt.json
 hamronize amrfinderplus --input_file_name amrfinderplus_nucleotide_report --analysis_software_version AFP_nt_v1 --reference_database_version db_v_1 data/raw_outputs/amrfinderplus/report_nucleotide.tsv --format tsv --output hamronized_amrfinderplus_nt.tsv
-hamronize amrfinderplus --input_file_name amrfinderplus_portein_report --analysis_software_version AFP_aa_v1 --reference_database_version db_v_1 data/raw_outputs/amrfinderplus/report_protein.tsv --format json --output hamronized_amrfinderplus_aa.json
-hamronize amrfinderplus --input_file_name amrfinderplus_portein_report --analysis_software_version AFP_aa_v1 --reference_database_version db_v_1 data/raw_outputs/amrfinderplus/report_protein.tsv --format tsv --output hamronized_amrfinderplus_aa.tsv
+hamronize amrfinderplus --input_file_name amrfinderplus_protein_report --analysis_software_version AFP_aa_v1 --reference_database_version db_v_1 data/raw_outputs/amrfinderplus/report_protein.tsv --format json --output hamronized_amrfinderplus_aa.json
+hamronize amrfinderplus --input_file_name amrfinderplus_protein_report --analysis_software_version AFP_aa_v1 --reference_database_version db_v_1 data/raw_outputs/amrfinderplus/report_protein.tsv --format tsv --output hamronized_amrfinderplus_aa.tsv
 
 hamronize rgi --input_file_name rgi_report --analysis_software_version rgi_v1 --reference_database_version card_v1 data/raw_outputs/rgi/rgi.txt --format json --output hamronized_rgi.json
 hamronize rgi --input_file_name rgi_report --analysis_software_version rgi_v1 --reference_database_version card_v1 data/raw_outputs/rgi/rgi.txt --format tsv --output hamronized_rgi.tsv
@@ -27,6 +27,15 @@ hamronize rgi --input_file_name rgi_report --analysis_software_version rgi_v1 --
 hamronize resfinder --input_file_name resfinder_report --analysis_software_version resfinder_v4 --reference_database_version resfinder_db_v1 data/raw_outputs/resfinder/ResFinder_results_tab.txt --format json --output hamronized_resfinder.json
 hamronize resfinder --input_file_name resfinder_report --analysis_software_version resfinder_v4 --reference_database_version resfinder_db_v1 data/raw_outputs/resfinder/ResFinder_results_tab.txt --format tsv --output hamronized_resfinder.tsv
 
+hamronize resfinder --input_file_name resfinder_report --analysis_software_version resfinder_v4.5 --reference_database_version resfinder_db_v1 data/raw_outputs/resfinder/resfinder_inputfasta_tab.txt --format json --output hamronized_resfinder_inputfasta.json
+hamronize resfinder --input_file_name resfinder_report --analysis_software_version resfinder_v4.5 --reference_database_version resfinder_db_v1 data/raw_outputs/resfinder/resfinder_inputfasta_tab.txt --format tsv --output hamronized_resfinder_inputfasta.tsv
+
+hamronize resfinder --input_file_name resfinder_report --analysis_software_version resfinder_v4.5 --reference_database_version resfinder_db_v1 data/raw_outputs/resfinder/resfinder_inputfastq_tab.txt --format json --output hamronized_resfinder_inputfastq.json
+hamronize resfinder --input_file_name resfinder_report --analysis_software_version resfinder_v4.5 --reference_database_version resfinder_db_v1 data/raw_outputs/resfinder/resfinder_inputfastq_tab.txt --format tsv --output hamronized_resfinder_inputfastq.tsv
+
+hamronize resfinder --input_file_name resfinder_report --analysis_software_version resfinder_v4.5 --reference_database_version resfinder_db_v1 data/raw_outputs/resfinder/resfinder_inputfastq_inputfasta.txt --format json --output hamronized_resfinder_inputfastq_inputfastq.json
+hamronize resfinder --input_file_name resfinder_report --analysis_software_version resfinder_v4.5 --reference_database_version resfinder_db_v1 data/raw_outputs/resfinder/resfinder_inputfastq_inputfasta.txt --format tsv --output hamronized_resfinder_inputfastq_inputfasta.tsv
+
 hamronize pointfinder --input_file_name pointfinder_report --analysis_software_version resfinder_v4 --reference_database_version pointfinder_db_v1 data/raw_outputs/pointfinder/PointFinder_results.txt --format json --output hamronized_pointfinder.json
 hamronize pointfinder --input_file_name pointfinder_report --analysis_software_version resfinder_v4 --reference_database_version pointfinder_db_v1 data/raw_outputs/pointfinder/PointFinder_results.txt --format tsv --output hamronized_pointfinder.tsv
 

diff --git a/test/test_parsing_validity.py b/test/test_parsing_validity.py
@@ -292,7 +292,7 @@ def test_kmerresistance():
         assert result.reference_gene_stop is None
 
 
-def test_resfinder():
+def test_pointfinder():
     metadata = {
         "analysis_software_version": "4.1.0",
         "reference_database_version": "2021-02-01",
@@ -318,17 +318,32 @@ def test_resfinder():
         assert result.nucleotide_mutation == "GGT -> GAT"
         assert result.amino_acid_mutation == "p.G81D"
 
-
-def test_pointfinder():
+def test_resfinder_fastq():
     metadata = {
-        "analysis_software_version": "4.1.0",
-        "reference_database_version": "2019-Jul-28",
+        "analysis_software_version": "4.5.0",
+        "reference_database_version": "2021-02-01",
         "input_file_name": "Dummy",
     }
     parsed_report = hAMRonization.parse(
-        "data/dummy/resfinder/ResFinder_results_tab.txt", metadata, "resfinder"
+        "data/dummy/resfinder/resfinder_inputfastq_inputfasta.txt", metadata, "resfinder"
     )
 
+    for result in parsed_report:
+        # assert mandatory fields
+        assert result.input_file_name == "Dummy"
+        assert result.gene_symbol == "aph(6)-Id"
+        assert result.gene_name == "aph(6)-Id"
+        assert result.reference_database_name == "resfinder"
+        assert result.reference_database_version == "2021-02-01"
+        assert result.reference_accession == "M28829"
+        assert result.analysis_software_name == "resfinder"
+        assert result.analysis_software_version == "4.5.0"
+        assert result.genetic_variation_type == "gene_presence_detected"
+        assert result.coverage_percentage == 100
+        assert result.drug_class == "Streptomycin"
+        assert result.reference_gene_length == 837
+        assert result.sequence_identity == 100
+
 
 def test_rgi_variants():
     metadata = {