Merge pull request #35 from UW-GAC/test_data

Add pheno test data

test_data/cancer_breast.tsv

@@ -0,0 +1,21 @@
+subject_id	age_at_obs	visit	breast_cancer_status_emerge_1	breast_cancer_status_registry_1	breast_cancer_status_survey_1	age_at_diagnosis_1	year_at_diagnosis_1	breast_cancer_type_1	cancer_behavior_1	her2_1	pr_1	er_1	T_stage_clinical_1	T_stage_pathological_1	T_stage_uknown_1	T_stage_clinical_2	T_stage_pathological_2	T_stage_unknown_2	nodal_involvement_1	distant_metastasis_1	stage_system_1	grade_clinical_1	grade_pathological_1	grade_unknown_1	screening_history_1	recurrence_1	surgery_1	radiotherapy_1	chemotherapy_1	hormone_therapy_1	NSAID_1	age_at_natural_menopause_1	post_menopausal_hormone_use_1	parity_1	age_at_first_birth_1	age_at_menarche_1	deceased_1	cause_of_death_breast_cancer_1	age_at_death_1
+subject1	59	visit_1	0	0	0	63	2017	unilateral	benign	negative	unknown	positive	unstaged	stage 4	stage 3	unknown	unknown	unstaged	N2	M1	NA	grade 1	grade 1	grade 3	0	recurrence_second_primary	0	0	1	pharmaceutical	1	64	1	0	26	15	0	0	64
+subject2	46	visit_1	0	1	0	46	2013	bilateral	borderline	unknown	unknown	negative	unstaged	stage 4	stage 3	localized	distant	in_situ	NX	MX	NA	grade 1	grade 3	grade 1	1	none	0	0	0	surgical	0	58	0	2	30	9	1	1	48
+subject3	55	visit_1	0	0	0	62	2011	unilateral	invasive	positive	negative	positive	stage 3	unstaged	stage 1	unstaged	unstaged	regional	N2	MX	NA	grade 2	grade 2	grade 1	0	unknown	1	0	0	surgical	1	62	1	2	32	18	0	0	72
+subject4	55	visit_1	0	0	0	64	2017	bilateral	borderline	unknown	negative	unknown	stage 3	stage 2	stage 4	localized	localized	regional	N0	MX	NA	grade 2	grade 3	grade 3	0	unknown	1	1	0	pharmaceutical	1	59	0	2	30	14	0	0	78
+subject5	62	visit_1	0	0	0	63	2016	unilateral	benign	positive	positive	negative	stage 3	unstaged	unknown	regional	regional	unknown	NX	MX	NA	grade 2	grade 3	grade 2	0	recurrence_second_primary	0	0	0	unknown	0	63	1	2	30	13	0	0	77
+subject6	55	visit_1	0	0	0	59	2002	bilateral	in_situ	negative	positive	positive	stage 1	stage 4	stage 1	localized	localized	unknown	N3	MX	NA	grade 2	grade 3	grade 2	1	none	1	1	1	both	0	57	0	0	23	20	0	0	73
+subject7	61	visit_1	0	0	0	61	2002	bilateral	invasive	negative	positive	unknown	stage 3	stage 4	stage 3	localized	unstaged	unknown	NX	MX	NA	grade 1	grade 1	grade 3	1	unknown	1	1	0	none	0	61	0	2	17	12	0	0	73
+subject8	65	visit_1	0	0	0	68	2021	unilateral	benign	unknown	positive	positive	stage 3	stage 3	unstaged	in_situ	localized	unknown	NX	M0	NA	grade 1	grade 3	grade 3	0	unknown	1	1	0	none	1	81	1	0	31	13	0	0	86
+subject9	66	visit_1	0	0	0	72	2006	bilateral	in_situ	unknown	positive	unknown	stage 4	stage 3	unknown	regional	regional	regional	N3	M1	NA	grade 3	grade 2	grade 1	1	none	0	0	1	none	0	68	1	0	25	14	0	0	81
+subject10	51	visit_1	0	0	0	62	2017	bilateral	borderline	negative	unknown	unknown	stage 1	unknown	stage 2	unknown	distant	unknown	N2	MX	NA	grade 1	grade 1	grade 1	1	recurrence_primary	1	0	0	surgical	0	59	1	0	27	16	0	0	74
+subject11	61	visit_1	0	0	0	63	2006	unilateral	borderline	negative	positive	negative	stage 1	stage 4	unknown	unknown	regional	unstaged	NX	M0	NA	grade 1	grade 1	grade 3	0	recurrence_primary	0	1	1	unknown	0	77	0	1	30	17	0	0	67
+subject12	55	visit_1	0	0	0	59	2004	unilateral	benign	positive	negative	unknown	unknown	unknown	stage 1	unknown	localized	unstaged	N1	MX	NA	grade 1	grade 1	grade 2	0	none	0	1	1	pharmaceutical	1	74	1	1	26	15	0	0	59
+subject13	52	visit_1	0	0	0	55	2010	unilateral	borderline	negative	positive	unknown	unknown	stage 2	stage 2	unstaged	localized	regional	N1	MX	NA	grade 2	grade 1	grade 1	0	unknown	0	1	0	surgical	1	53	0	1	25	12	0	0	66
+subject14	66	visit_1	1	0	0	68	2007	unilateral	benign	unknown	negative	positive	stage 4	stage 2	stage 3	in_situ	distant	unknown	N1	M0	NA	grade 2	grade 3	grade 1	1	recurrence_primary	1	1	1	both	0	82	0	2	22	11	0	0	72
+subject15	57	visit_1	0	0	0	64	2008	bilateral	invasive	positive	positive	positive	unstaged	stage 2	stage 4	distant	in_situ	regional	N0	M1	NA	grade 3	grade 3	grade 3	0	recurrence_second_primary	0	1	1	both	1	66	0	1	22	12	0	0	80
+subject16	57	visit_1	0	1	0	62	2003	bilateral	invasive	unknown	positive	unknown	unstaged	stage 4	stage 4	unknown	unknown	unstaged	N0	M0	NA	grade 2	grade 1	grade 1	1	recurrence_second_primary	0	0	0	both	0	57	1	1	26	19	1	0	70
+subject17	67	visit_1	1	0	0	69	2014	unilateral	in_situ	positive	negative	unknown	stage 1	unknown	stage 2	localized	regional	localized	N1	MX	NA	grade 2	grade 1	grade 3	1	none	0	1	1	none	0	74	0	1	32	14	0	0	85
+subject18	59	visit_1	0	0	0	62	2009	bilateral	borderline	positive	positive	negative	stage 2	stage 2	unknown	unknown	regional	distant	N3	M0	NA	grade 3	grade 2	grade 2	0	recurrence_primary	0	1	0	pharmaceutical	0	59	0	1	23	12	0	0	74
+subject19	67	visit_1	1	0	0	69	2001	bilateral	borderline	negative	negative	negative	stage 1	stage 2	stage 1	unknown	unstaged	unknown	N0	MX	NA	grade 2	grade 3	grade 2	0	recurrence_second_primary	1	1	1	both	0	78	0	1	32	18	0	0	73
+subject20	62	visit_1	0	0	0	63	2013	unilateral	benign	unknown	unknown	positive	stage 4	stage 2	unknown	in_situ	regional	distant	N0	M0	NA	grade 1	grade 2	grade 3	0	recurrence_second_primary	0	0	0	surgical	0	75	0	2	28	16	1	0	64

test_data/cancer_prostate.tsv

@@ -0,0 +1,21 @@
+subject_id	age_at_obs	visit	prostate_cancer_status_emerge_1	prostate_cancer_status_registry_1	prostate_cancer_status_survey_1	age_at_diagnosis_1	year_at_diagnosis_1	cancer_behavior_1	T_stage_clinical_1	T_stage_pathological_1	T_stage_uknown_1	T_stage_clinical_2	T_stage_pathological_2	T_stage_unknown_2	nodal_involvement_1	distant_metastasis_1	stage_system_1	gleason_score_clinical_1	gleason_score_pathological_1	gleason_score_unknown_1	psa_1	psa_at_diagnosis_1	screening_history_1	recurrence_1	surgery_1	radiotherapy_1	chemotherapy_1	hormone_therapy_1	NSAID_1	deceased_1	cause_of_death_prostate_cancer_1	age_at_death_1
+subject1	59	visit_1	1	1	1	63	2017	borderline	stage 1	stage 4	unknown	localized	unstaged	in_situ	N1	M1	NA	9	9	7	3.3273850651366184	0.6910954617090586	0	none	0	0	0	0	1	0	0	68
+subject2	46	visit_1	0	1	0	46	2013	in_situ	stage 3	unstaged	unknown	in_situ	unstaged	in_situ	N0	MX	NA	5	6	3	0.1117048051118088	1.114277099266359	1	recurrence_second_primary	1	1	0	0	0	0	0	70
+subject3	55	visit_1	0	1	1	62	2011	borderline	unknown	stage 2	stage 3	regional	distant	unstaged	NX	M1	NA	8	3	8	1.1467807022659444	1.650565194078115	0	recurrence_primary	0	1	1	0	0	0	0	67
+subject4	55	visit_1	0	0	0	64	2017	in_situ	stage 2	unknown	stage 3	unknown	distant	regional	N2	MX	NA	2	3	5	0.6244635030052077	1.385998555747178	1	none	0	1	1	1	0	0	1	62
+subject5	62	visit_1	1	1	1	63	2016	borderline	stage 1	stage 2	stage 2	in_situ	distant	unstaged	N3	MX	NA	3	10	3	1.0444848894466576	1.9458136918517794	1	recurrence_second_primary	0	0	0	0	0	1	0	63
+subject6	55	visit_1	0	1	0	59	2002	in_situ	unstaged	stage 3	stage 1	regional	localized	in_situ	NX	MX	NA	2	10	9	0.959323014536773	1.2909907535132685	1	recurrence_second_primary	1	1	1	1	1	0	0	59
+subject7	61	visit_1	1	0	1	61	2002	in_situ	unknown	stage 4	stage 1	unstaged	distant	in_situ	N0	MX	NA	8	10	2	0.08122257203704342	0.302979697472334	0	none	1	1	1	1	0	0	0	62
+subject8	65	visit_1	1	1	1	68	2021	benign	stage 2	unknown	stage 1	regional	distant	distant	N2	M0	NA	2	10	2	1.6165848336901831	0.3081623612021196	1	none	0	1	1	1	0	0	0	86
+subject9	66	visit_1	1	1	1	72	2006	invasive	stage 4	unstaged	stage 1	unstaged	in_situ	distant	N3	MX	NA	4	4	9	3.215384566425846	3.4306225959208776	0	recurrence_primary	1	1	0	0	1	0	0	70
+subject10	51	visit_1	0	1	1	62	2017	invasive	stage 3	stage 3	unstaged	unknown	localized	unknown	N0	M1	NA	5	9	8	0.21256352824871372	1.1323419133081138	0	recurrence_primary	1	0	1	0	0	0	0	66
+subject11	61	visit_1	1	1	0	63	2006	borderline	stage 3	stage 3	stage 1	distant	localized	in_situ	N3	M1	NA	4	8	2	1.6517128367933704	0.9405233862074993	0	none	0	0	0	1	1	1	0	85
+subject12	55	visit_1	0	0	1	59	2004	borderline	stage 2	stage 1	stage 4	unknown	unknown	unknown	NX	M1	NA	2	8	6	0.2398326754427973	0.7250777214867976	0	unknown	0	1	0	1	1	0	0	85
+subject13	52	visit_1	0	1	1	55	2010	borderline	stage 3	stage 4	stage 2	unknown	unknown	regional	NX	M1	NA	3	3	6	1.1806309466433595	1.9446144610218248	0	recurrence_primary	0	1	0	1	0	0	0	54
+subject14	66	visit_1	1	1	1	68	2007	borderline	stage 2	unknown	stage 4	regional	in_situ	regional	N0	M0	NA	6	9	6	1.6857904625831341	1.52517841165081	1	recurrence_primary	1	0	1	1	1	0	0	86
+subject15	57	visit_1	0	0	1	64	2008	invasive	unstaged	stage 2	stage 1	regional	unstaged	regional	N2	M0	NA	7	4	3	2.1816372213400363	2.002552414007943	1	unknown	0	0	0	1	1	0	0	73
+subject16	57	visit_1	0	1	0	62	2003	in_situ	unknown	unknown	stage 1	unstaged	unstaged	in_situ	N2	M1	NA	10	8	4	1.8889410806023983	1.223594159764647	0	recurrence_primary	1	0	0	0	0	0	0	57
+subject17	67	visit_1	1	0	0	69	2014	borderline	unstaged	stage 2	stage 4	unknown	localized	unknown	N0	MX	NA	4	2	5	0.8706631343253838	1.8269750432423142	0	none	1	1	0	1	1	0	0	78
+subject18	59	visit_1	1	1	0	62	2009	invasive	stage 2	stage 2	stage 2	distant	regional	regional	N3	M1	NA	4	5	10	1.5219466926525085	2.223313852420131	0	unknown	0	0	0	1	0	0	0	60
+subject19	67	visit_1	1	0	1	69	2001	invasive	stage 3	stage 3	stage 3	distant	localized	regional	NX	M1	NA	9	9	5	0.61399603928097	1.4170698016606862	1	unknown	0	0	1	1	1	0	0	82
+subject20	62	visit_1	1	0	1	63	2013	benign	stage 1	unstaged	unstaged	localized	in_situ	regional	N3	M0	NA	3	5	4	1.211037802492477	0.8747108557841603	1	none	0	0	0	0	0	0	0	81

test_data/cmqt_anthropometry.tsv

@@ -0,0 +1,21 @@
+subject_id	age_at_obs	visit	height_1	weight_1	bmi_1	waist_hip_ratio_1
+subject1	59	visit_1	166.51728404004504	87.70407490435488	31.630141478991852	0.9074966900959498
+subject2	46	visit_1	161.202551994156	80.8258450985496	31.103297082655853	0.8145228307678352
+subject3	55	visit_1	171.23801251550628	86.53811180127285	29.51257188546359	0.9034009869133715
+subject4	55	visit_1	169.17186404031233	86.4412843896079	30.204033404764687	0.6649561139308164
+subject5	62	visit_1	176.4493260077908	82.96448470324013	26.647238715038643	0.7343205137895237
+subject6	55	visit_1	169.82492809343805	78.58528157838867	27.248232409006288	0.7310283084720187
+subject7	61	visit_1	156.0312735892919	86.27942012799926	35.439200098235276	0.8079074951310364
+subject8	65	visit_1	163.50798836505146	84.54919575614312	31.62507251261148	0.7699475884641312
+subject9	66	visit_1	178.2757791033649	75.35985947462825	23.711301552730852	0.8579123324227408
+subject10	51	visit_1	177.43804249577906	86.20090419001572	27.379048176451334	0.656209438514197
+subject11	61	visit_1	168.9662314872622	80.76732089777315	28.290188511041826	0.7469005486739022
+subject12	55	visit_1	165.110036178032	85.25966289494805	31.274953024607136	0.7501018809016656
+subject13	52	visit_1	167.68140136962006	76.22894393591241	27.111285687273472	0.7936294054529471
+subject14	66	visit_1	164.68404018860684	72.58905440174385	26.765053327547104	0.8348499810263527
+subject15	57	visit_1	165.24046335177894	84.30565936248858	30.876205386787987	0.9576720775779393
+subject16	57	visit_1	166.18318741952814	77.97740084621502	28.23544170233956	0.75225930619927
+subject17	67	visit_1	173.1551878731937	78.86297291363192	26.30280885590887	0.7557994230712831
+subject18	59	visit_1	164.6905720192354	84.67048085431246	31.217243374076368	0.8556773306960885
+subject19	67	visit_1	164.29742090189868	77.67052060104524	28.773634591122615	0.7875468828313612
+subject20	62	visit_1	163.01588801788486	76.81228250714892	28.90485548023756	0.9079118556158186

test_data/cmqt_blood_pressure.tsv

@@ -0,0 +1,21 @@
+subject_id	age_at_obs	visit	systolic_bp_1	diastolic_bp_1	hypertension_1
+subject1	59	visit_1	124.33509725727151	95.40814980870978	0
+subject2	46	visit_1	109.15014855473149	81.65169019709921	0
+subject3	55	visit_1	137.82289290144655	93.0762236025457	0
+subject4	55	visit_1	131.91961154374948	92.88256877921579	0
+subject5	62	visit_1	152.71236002225947	85.92896940648028	0
+subject6	55	visit_1	133.78550883839446	77.17056315677733	0
+subject7	61	visit_1	94.37506739797686	92.5588402559985	0
+subject8	65	visit_1	115.73710961443275	89.09839151228624	0
+subject9	66	visit_1	157.93079743818544	70.7197189492565	0
+subject10	51	visit_1	155.53726427365447	92.40180838003143	1
+subject11	61	visit_1	131.33208996360634	81.53464179554629	0
+subject12	55	visit_1	120.3143890800914	90.5193257898961	0
+subject13	52	visit_1	127.66114677034301	72.45788787182482	0
+subject14	66	visit_1	119.09725768173382	65.17810880348769	0
+subject15	57	visit_1	120.68703814793986	88.61131872497717	0
+subject16	57	visit_1	123.38053548436613	75.95480169243004	0
+subject17	67	visit_1	143.30053678055347	77.72594582726384	0
+subject18	59	visit_1	119.11592005495825	89.34096170862492	0
+subject19	67	visit_1	117.9926311482819	75.3410412020905	0
+subject20	62	visit_1	114.33110862252818	73.62456501429784	0

test_data/cmqt_flags.tsv

@@ -0,0 +1,21 @@
+subject_id	age_at_obs	visit	flag_pregnancy_1	flag_acute_illness_1	flag_bld_1	flag_anemia_1	flag_hiv_1	flag_eskd_1	flag_splenectomy_1	flag_cirrhosis_1	flag_fasting_1	flag_lipids_med_1	flag_bp_med_1	flag_cvd_1	flag_t2d_1	flag_t1d_1	flag_diabetes_other_1
+subject1	59	visit_1	unknown	unknown	data not collected	unknown	data not collected	yes	no	unknown	data not collected	no	data not collected	unknown	yes	data not collected	data not collected
+subject2	46	visit_1	no	data not collected	no	no	unknown	unknown	yes	unknown	data not collected	unknown	data not collected	unknown	yes	no	yes
+subject3	55	visit_1	yes	unknown	unknown	data not collected	unknown	yes	data not collected	yes	yes	no	yes	data not collected	no	unknown	data not collected
+subject4	55	visit_1	yes	yes	no	data not collected	data not collected	unknown	yes	data not collected	yes	data not collected	yes	yes	unknown	no	no
+subject5	62	visit_1	data not collected	unknown	data not collected	no	data not collected	yes	no	no	yes	unknown	yes	data not collected	data not collected	no	no
+subject6	55	visit_1	yes	data not collected	no	unknown	no	unknown	unknown	yes	no	no	no	unknown	no	no	no
+subject7	61	visit_1	unknown	yes	data not collected	no	no	unknown	data not collected	unknown	no	unknown	yes	unknown	yes	no	unknown
+subject8	65	visit_1	data not collected	data not collected	data not collected	unknown	data not collected	no	no	data not collected	no	yes	yes	yes	unknown	unknown	no
+subject9	66	visit_1	data not collected	unknown	unknown	data not collected	no	data not collected	unknown	data not collected	no	data not collected	unknown	yes	data not collected	yes	no
+subject10	51	visit_1	no	unknown	unknown	data not collected	data not collected	data not collected	yes	yes	data not collected	data not collected	no	data not collected	yes	data not collected	yes
+subject11	61	visit_1	data not collected	unknown	yes	yes	no	yes	yes	yes	no	yes	no	unknown	data not collected	data not collected	yes
+subject12	55	visit_1	yes	no	unknown	unknown	no	yes	no	no	unknown	data not collected	data not collected	data not collected	no	data not collected	no
+subject13	52	visit_1	no	data not collected	data not collected	yes	unknown	yes	yes	unknown	no	data not collected	data not collected	no	no	unknown	no
+subject14	66	visit_1	data not collected	unknown	data not collected	yes	yes	yes	no	data not collected	unknown	no	unknown	no	yes	unknown	yes
+subject15	57	visit_1	yes	yes	data not collected	data not collected	yes	data not collected	data not collected	yes	no	yes	data not collected	yes	unknown	yes	unknown
+subject16	57	visit_1	yes	data not collected	yes	unknown	no	unknown	data not collected	data not collected	no	unknown	unknown	unknown	unknown	data not collected	data not collected
+subject17	67	visit_1	data not collected	yes	no	yes	data not collected	yes	unknown	no	unknown	data not collected	no	data not collected	yes	no	yes
+subject18	59	visit_1	unknown	unknown	no	yes	yes	data not collected	yes	yes	no	yes	no	yes	data not collected	data not collected	yes
+subject19	67	visit_1	data not collected	yes	data not collected	yes	no	data not collected	yes	yes	yes	yes	no	no	no	data not collected	data not collected
+subject20	62	visit_1	data not collected	no	unknown	no	unknown	no	no	data not collected	unknown	no	unknown	yes	data not collected	unknown	no

test_data/cmqt_glycemic.tsv

@@ -0,0 +1,21 @@
+subject_id	age_at_obs	visit	fasting_glucose_plasma_1	fasting_glucose_serum_1	fasting_insulin_1	hba1c_1
+subject1	59	visit_1	54.335097257271514	100.40814980870978	44.062251757196236	0.8629538590264505
+subject2	46	visit_1	39.150148554731494	86.65169019709921	37.08921230758763	5.12890149360826
+subject3	55	visit_1	67.82289290144655	98.0762236025457	43.75507401850285	0.3745564709250373
+subject4	55	visit_1	61.9196115437495	97.88256877921579	25.871708544811234	7.12738940450381
+subject5	62	visit_1	82.71236002225946	90.92896940648028	31.074038534214274	3.262766021325356
+subject6	55	visit_1	63.78550883839445	82.17056315677733	30.827123135401397	2.5366231021701244
+subject7	61	visit_1	24.37506739797687	97.5588402559985	36.59306213482773	2.2052889540530853
+subject8	65	visit_1	45.73710961443275	94.09839151228624	33.746069134809844	4.778864164562747
+subject9	66	visit_1	87.93079743818544	75.7197189492565	40.343424931705556	4.052338078993457
+subject10	51	visit_1	85.53726427365447	97.40180838003143	25.215707888564772	2.657453514074762
+subject11	61	visit_1	61.332089963606336	86.53464179554629	32.01754115054266	3.31735379488651
+subject12	55	visit_1	50.3143890800914	95.5193257898961	32.25764106762492	2.0286698676551476
+subject13	52	visit_1	57.66114677034302	77.45788787182482	35.52220540897103	1.0821878500015871
+subject14	66	visit_1	49.09725768173383	70.17810880348769	38.61374857697645	3.36103458421411
+subject15	57	visit_1	50.68703814793985	93.61131872497717	47.825405818345445	4.443468565637837
+subject16	57	visit_1	53.38053548436613	80.95480169243004	32.41944796494524	2.2609190437584035
+subject17	67	visit_1	73.30053678055346	82.72594582726384	32.68495673034623	3.475076625079314
+subject18	59	visit_1	49.11592005495825	94.34096170862492	40.17579980220664	1.6681557751701703
+subject19	67	visit_1	47.99263114828191	80.3410412020905	35.06601621235209	1.4063849803144863
+subject20	62	visit_1	44.33110862252818	78.62456501429784	44.09338917118639	2.8966061373772334