Added tests for processing ecoli and saureus data

tests/__init__.py

@@ -0,0 +1 @@
+"""PRP unit tests."""

tests/conftest.py

@@ -0,0 +1,3 @@
+"""Test fixtures."""
+
+from .fixtures import *

tests/fixtures/__init__.py

@@ -0,0 +1,5 @@
+"""Fixtures"""
+
+from .ecoli import *
+from .mtuberculosis import *
+from .saureus import *

tests/fixtures/ecoli/__init__.py

@@ -0,0 +1,60 @@
+"""Ecoli input data fixutres."""
+
+import pytest
+
+from ..fixtures import data_path
+
+
+@pytest.fixture()
+def ecoli_analysis_meta_path(data_path):
+    return str(data_path.joinpath("ecoli", "analysis_meta.json"))
+
+
+@pytest.fixture()
+def ecoli_bwa_path(data_path):
+    return str(data_path.joinpath("ecoli", "bwa.qc"))
+
+
+@pytest.fixture()
+def ecoli_quast_path(data_path):
+    return str(data_path.joinpath("ecoli", "quast.tsv"))
+
+
+@pytest.fixture()
+def ecoli_amrfinder_path(data_path):
+    return str(data_path.joinpath("ecoli", "amrfinder.out"))
+
+
+@pytest.fixture()
+def ecoli_resfinder_path(data_path):
+    return str(data_path.joinpath("ecoli", "resfinder.json"))
+
+
+@pytest.fixture()
+def ecoli_resfinder_meta_path(data_path):
+    return str(data_path.joinpath("ecoli", "resfinder_meta.json"))
+
+
+@pytest.fixture()
+def ecoli_virulencefinder_path(data_path):
+    return str(data_path.joinpath("ecoli", "virulencefinder.json"))
+
+
+@pytest.fixture()
+def ecoli_virulencefinder_meta_path(data_path):
+    return str(data_path.joinpath("ecoli", "virulencefinder_meta.json"))
+
+
+@pytest.fixture()
+def ecoli_mlst_path(data_path):
+    return str(data_path.joinpath("ecoli", "mlst.json"))
+
+
+@pytest.fixture()
+def ecoli_chewbbaca_path(data_path):
+    return str(data_path.joinpath("ecoli", "chewbbaca.out"))
+
+
+@pytest.fixture()
+def ecoli_bracken_path(data_path):
+    return str(data_path.joinpath("ecoli", "bracken.out"))

tests/fixtures/ecoli/amrfinder.out

@@ -0,0 +1,33 @@
+Protein identifier	Contig id	Start	Stop	Strand	Gene symbol	Sequence name	Scope	Element type	Element subtype	Class	Subclass	Method	Target length	Reference sequence length	% Coverage of reference sequence	% Identity to reference sequence	Alignment length	Accession of closest sequence	Name of closest sequence	HMM id	HMM description
+NA	Contig_109_177.84	272	1414	+	blaCMY-2	extended-spectrum class C beta-lactamase CMY-2	core	AMR	AMR	BETA-LACTAM	CEPHALOSPORIN	ALLELEX	381	381	100.00	100.00	381	WP_000976514.1	extended-spectrum class C beta-lactamase CMY-2	NA	NA
+NA	Contig_110_265.384	204	1076	+	blaCTX-M-15	extended-spectrum class A beta-lactamase CTX-M-15	core	AMR	AMR	BETA-LACTAM	CEPHALOSPORIN	ALLELEX	291	291	100.00	100.00	291	WP_000239590.1	extended-spectrum class A beta-lactamase CTX-M-15	NA	NA
+NA	Contig_11_258.988	1	246	-	arr	NAD(+)--rifampin ADP-ribosyltransferase	core	AMR	AMR	RIFAMYCIN	RIFAMYCIN	PARTIAL_CONTIG_ENDX	82	150	54.67	100.00	82	WP_000237816.1	NAD(+)--rifampin ADP-ribosyltransferase Arr-2	NA	NA
+NA	Contig_11_258.988	334	963	-	catB3	type B-3 chloramphenicol O-acetyltransferase CatB3	core	AMR	AMR	PHENICOL	CHLORAMPHENICOL	EXACTX	210	210	100.00	100.00	210	WP_000186237.1	type B-3 chloramphenicol O-acetyltransferase CatB3	NA	NA
+NA	Contig_11_258.988	1104	1931	-	blaOXA-1	oxacillin-hydrolyzing class D beta-lactamase OXA-1	core	AMR	AMR	BETA-LACTAM	CEPHALOSPORIN	ALLELEX	276	276	100.00	100.00	276	WP_001334766.1	oxacillin-hydrolyzing class D beta-lactamase OXA-1	NA	NA
+NA	Contig_11_258.988	2065	2616	-	aac(6')-Ib-cr5	fluoroquinolone-acetylating aminoglycoside 6'-N-acetyltransferase AAC(6')-Ib-cr5	core	AMR	AMR	AMINOGLYCOSIDE/QUINOLONE	AMIKACIN/KANAMYCIN/QUINOLONE/TOBRAMYCIN	ALLELEX	184	184	100.00	100.00	184	WP_063840321.1	fluoroquinolone-acetylating aminoglycoside 6'-N-acetyltransferase AAC(6')-Ib-cr5	NA	NA
+NA	Contig_13_186.862	271186	274287	+	acrF	multidrug efflux RND transporter permease subunit AcrF	plus	AMR	AMR	EFFLUX	EFFLUX	BLASTX	1034	1034	100.00	99.52	1034	AAC76298.1	multidrug efflux RND transporter permease subunit AcrF	NA	NA
+NA	Contig_14_245.792	292	1026	+	erm(B)	23S rRNA (adenine(2058)-N(6))-methyltransferase Erm(B)	core	AMR	AMR	LINCOSAMIDE/MACROLIDE/STREPTOGRAMIN	CLINDAMYCIN/ERYTHROMYCIN/STREPTOGRAMIN B	BLASTX	245	245	100.00	99.59	245	WP_002321849.1	23S rRNA (adenine(2058)-N(6))-methyltransferase Erm(B)	NA	NA
+NA	Contig_17_190.249	13948	14517	-	lpfA-O113	long polar fimbria major subunit LpfA-O113	plus	VIRULENCE	VIRULENCE	NA	NA	EXACTX	190	190	100.00	100.00	190	AAT76975.1	long polar fimbria major subunit LpfA-O113	NA	NA
+NA	Contig_18_257.093	801	2003	+	tet(B)	tetracycline efflux MFS transporter Tet(B)	core	AMR	AMR	TETRACYCLINE	TETRACYCLINE	EXACTX	401	401	100.00	100.00	401	WP_001089068.1	tetracycline efflux MFS transporter Tet(B)	NA	NA
+NA	Contig_1_265.424_Circ	13536	13898	-	ble	bleomycin binding protein Ble-MBL	core	AMR	AMR	BLEOMYCIN	BLEOMYCIN	EXACTX	121	121	100.00	100.00	121	WP_004201167.1	bleomycin binding protein Ble-MBL	NA	NA
+NA	Contig_1_265.424_Circ	13905	14714	-	blaNDM-1	subclass B1 metallo-beta-lactamase NDM-1	core	AMR	AMR	BETA-LACTAM	CARBAPENEM	ALLELEX	270	270	100.00	100.00	270	WP_004201164.1	subclass B1 metallo-beta-lactamase NDM-1	NA	NA
+NA	Contig_27_199.337	47060	48241	-	emrD	multidrug efflux MFS transporter EmrD	plus	AMR	AMR	EFFLUX	EFFLUX	EXACTX	394	394	100.00	100.00	394	BAE77620.1	multidrug efflux MFS transporter EmrD	NA	NA
+NA	Contig_32_189.336	75664	76893	+	mdtM	multidrug efflux MFS transporter MdtM	plus	AMR	AMR	EFFLUX	EFFLUX	BLASTX	410	410	100.00	97.80	410	AAC77293.1	multidrug efflux MFS transporter MdtM	NA	NA
+NA	Contig_35_511.471	839	1309	+	dfrA17	trimethoprim-resistant dihydrofolate reductase DfrA17	core	AMR	AMR	TRIMETHOPRIM	TRIMETHOPRIM	EXACTX	157	157	100.00	100.00	157	WP_001389366.1	trimethoprim-resistant dihydrofolate reductase DfrA17	NA	NA
+NA	Contig_35_511.471	1443	2228	+	aadA5	ANT(3'')-Ia family aminoglycoside nucleotidyltransferase AadA5	core	AMR	AMR	AMINOGLYCOSIDE	STREPTOMYCIN	EXACTX	262	262	100.00	100.00	262	WP_000503573.1	ANT(3'')-Ia family aminoglycoside nucleotidyltransferase AadA5	NA	NA
+NA	Contig_35_511.471	2437	2781	+	qacEdelta1	quaternary ammonium compound efflux SMR transporter QacE delta 1	core	STRESS	BIOCIDE	QUATERNARY AMMONIUM	QUATERNARY AMMONIUM	ALLELEX	115	115	100.00	100.00	115	WP_000679427.1	quaternary ammonium compound efflux SMR transporter QacE delta 1	NA	NA
+NA	Contig_35_511.471	2778	3614	+	sul1	sulfonamide-resistant dihydropteroate synthase Sul1	core	AMR	AMR	SULFONAMIDE	SULFONAMIDE	EXACTX	279	279	100.00	100.00	279	WP_000259031.1	sulfonamide-resistant dihydropteroate synthase Sul1	NA	NA
+NA	Contig_35_511.471	8931	9833	-	mph(A)	Mph(A) family macrolide 2'-phosphotransferase	core	AMR	AMR	MACROLIDE	AZITHROMYCIN/ERYTHROMYCIN/SPIRAMYCIN/TELITHROMYCIN	EXACTX	301	301	100.00	100.00	301	WP_000219391.1	Mph(A) family macrolide 2'-phosphotransferase	NA	NA
+NA	Contig_37_258.787	522	1379	-	blaTEM-1	broad-spectrum class A beta-lactamase TEM-1	core	AMR	AMR	BETA-LACTAM	BETA-LACTAM	ALLELEX	286	286	100.00	100.00	286	WP_000027057.1	broad-spectrum class A beta-lactamase TEM-1	NA	NA
+NA	Contig_43_139.553	36759	37064	-	asr	acid resistance repetitive basic protein Asr	plus	STRESS	ACID	NA	NA	BLASTX	102	102	100.00	97.06	102	BAA15331.2	acid resistance repetitive basic protein Asr	NA	NA
+NA	Contig_5_179.498	80328	84569	+	fdeC	inverse autotransporter adhesin FdeC	plus	VIRULENCE	VIRULENCE	NA	NA	BLASTX	1414	1416	99.93	91.10	1415	ABE05822.1	inverse autotransporter adhesin FdeC	NA	NA
+NA	Contig_61_202.053	2951	5530	-	afaC	AfaC-I/III family usher protein	plus	VIRULENCE	VIRULENCE	NA	NA	BLASTX	860	860	100.00	98.95	860	CAW30799.1	AfaC-I/III family usher protein	NA	NA
+NA	Contig_61_202.053	5651	6367	-	nfaE	Dr family non-fimbrial adhesin I chaperone NfaE	plus	VIRULENCE	VIRULENCE	NA	NA	BLASTX	239	247	96.76	100.00	239	CAW30815.1	Dr family non-fimbrial adhesin I chaperone NfaE	NA	NA
+NA	Contig_82_185.683	118067	119485	+	espX1	type III secretion system effector EspX1	plus	VIRULENCE	VIRULENCE	NA	NA	BLASTX	473	473	100.00	91.75	473	ADD54679.1	type III secretion system effector EspX1	NA	NA
+NA	Contig_87_150.438	249055	249318	-	ariR	biofilm/acid-resistance regulator AriR	plus	STRESS	ACID	NA	NA	EXACTX	88	88	100.00	100.00	88	AAC74250.1	biofilm/acid-resistance regulator AriR	NA	NA
+NA	Contig_88_197.958	48509	49639	+	blaEC	BlaEC family class C beta-lactamase	plus	AMR	AMR	BETA-LACTAM	BETA-LACTAM	BLASTX	377	377	100.00	99.73	377	WP_063610930.1	extended-spectrum class C beta-lactamase EC-15	NA	NA
+NA	Contig_8_204.521	24476	24826	+	arsR	As(III)-sensing metalloregulatory transcriptional repressor ArsR	plus	STRESS	METAL	ARSENIC	ARSENIC	EXACTX	117	117	100.00	100.00	117	BAE77793.1	As(III)-sensing metalloregulatory transcriptional repressor ArsR	NA	NA
+NA	Contig_8_204.521	26185	26607	+	arsC	glutaredoxin-dependent arsenate reductase	plus	STRESS	METAL	ARSENIC	ARSENATE	BLASTX	141	141	100.00	93.62	141	AAA21096.1	glutaredoxin-dependent arsenate reductase	NA	NA
+NA	Contig_93_193.806	20014	20913	-	fieF	CDF family cation-efflux transporter FieF	plus	STRESS	METAL	NA	NA	EXACTX	300	300	100.00	100.00	300	BAE77395.1	CDF family cation-efflux transporter FieF	NA	NA
+NA	Contig_96_151.855	109126	110925	-	ybtQ	yersiniabactin ABC transporter ATP-binding/permease protein YbtQ	plus	VIRULENCE	VIRULENCE	NA	NA	BLASTX	600	600	100.00	99.83	600	AAC69584.1	yersiniabactin ABC transporter ATP-binding/permease protein YbtQ	NA	NA
+NA	Contig_96_151.855	110915	112714	-	ybtP	yersiniabactin ABC transporter ATP-binding/permease protein YbtP	plus	VIRULENCE	VIRULENCE	NA	NA	BLASTX	600	600	100.00	99.83	600	CAA21388.1	yersiniabactin ABC transporter ATP-binding/permease protein YbtP	NA	NA

tests/fixtures/ecoli/analysis_meta.json

@@ -0,0 +1,15 @@
+{
+  "workflow_name": "focused_snyder",
+  "sample_name": "ecoli_test_1",
+  "sequencing_platform": "illumina",
+  "sequencing_type": "PE",
+  "date": "2023-12-06T15:53:45.482390+01:00",
+  "pipeline": "main.nf",
+  "version": "1.0.0",
+  "commit": "null",
+  "configuration_files": [
+    "/fs1/pipelines/JASEN/nextflow.config"
+  ],
+  "analysis_profile": "escherichia_coli",
+  "command": "nextflow run /fs1/pipelines/JASEN/main.nf -profile escherichia_coli --csv /fs1/ryan/pipelines/jasen/test-csvs/prp_test_samples_ecoli.csv --queue high -with-singularity /fs1/pipelines/JASEN/container/jasen_2023_02-28.sif -with-report /fs1/nextflow//reports/ecoli_test_1.jasen-ecoli-dev.report.html -with-trace /fs1/nextflow//reports/ecoli_test_1.jasen-ecoli-dev.trace.txt -with-timeline /fs1/nextflow//reports/ecoli_test_1.jasen-ecoli-dev.timeline.html -work-dir /fs1/nextflow//ecoli_test_1.jasen-ecoli-dev"
+}