Skip to content

Commit

Permalink
Added data-driven test for rna_distribution_data
Browse files Browse the repository at this point in the history
  • Loading branch information
@jaclynbeck-sage
jaclynbeck-sage committed Jun 29, 2023
1 parent 9ef8c86 commit e26e735
Show file tree
Hide file tree
Showing 7 changed files with 272 additions and 0 deletions.
7 changes: 7 additions & 0 deletions ...test_assets/rna_distribution_data/input/test_rna_distribution_data_bad_input_keyerror.csv
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,7 @@
model,tissue,comparison,ensembl_gene_id,logfc,ci_l,ci_r,aveexpr,t,p_value,adj_p_val,gene_biotype,chromosome_name,direction,hgnc_symbol,percentage_gc_content,gene_length,sex,study
,CBE,AD-CONTROL,ENSG00000228521,1.646715414,1.219378549,2.074052279,-2.452298708,7.544599094,2.21E-13,2.26E-10,lncRNA,7,UP,AC099552.3,60.6903164,1043,ALL,MAYO
,CBE,AD-CONTROL,ENSG00000286872,-0.181247795,-0.341187573,-0.021308017,0.032732484,-2.226401902,0.026439738,0.06693269,lncRNA,15,NONE,AC024270.5,46.5131873,13422,ALL,MAYO
,CBE,AD-CONTROL,ENSG00000144228,0.080531879,-0.016465937,0.177529695,4.913442808,1.63093857,0.10355023,0.193939662,protein_coding,2,NONE,SPOPL,36.67609796,71747,ALL,MAYO
Diagnosis.AOD,,AD-CONTROL,ENSG00000005339,0.074072272,0.042805132,0.105339413,7.626719604,4.656313999,4.15E-06,8.32E-05,protein_coding,16,NONE,CREBBP,45.00523533,155673,ALL,MAYO
Diagnosis.AOD,,AD-CONTROL,ENSG00000188994,0.03375383,-0.010318552,0.077826212,6.703806497,1.501322866,0.133920382,0.239511453,protein_coding,6,NONE,ZNF292,37.67626491,113111,ALL,MAYO
Diagnosis.AOD,,AD-CONTROL,ENSG00000261823,-0.008434191,-0.077650231,0.060781849,-0.970547902,-0.239149259,0.811089637,0.873760793,lncRNA,15,NONE,AC084782.2,42.13615023,1704,ALL,MAYO
2 changes: 2 additions & 0 deletions ...est_assets/rna_distribution_data/input/test_rna_distribution_data_bad_input_typeerror.csv
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,2 @@
model,tissue,comparison,ensembl_gene_id,logfc,ci_l,ci_r,aveexpr,t,p_value,adj_p_val,gene_biotype,chromosome_name,direction,hgnc_symbol,percentage_gc_content,gene_length,sex,study
Diagnosis,CBE,AD-CONTROL,ENSG00000228521,No change,1.219378549,2.074052279,-2.452298708,7.544599094,2.21E-13,2.26E-10,lncRNA,7,UP,AC099552.3,60.6903164,1043,ALL,MAYO
26 changes: 26 additions & 0 deletions tests/test_assets/rna_distribution_data/input/test_rna_distribution_data_good_input.csv
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,26 @@
model,tissue,comparison,ensembl_gene_id,logfc,ci_l,ci_r,aveexpr,t,p_value,adj_p_val,gene_biotype,chromosome_name,direction,hgnc_symbol,percentage_gc_content,gene_length,sex,study
Diagnosis,CBE,AD-CONTROL,ENSG00000228521,1.646715414,1.219378549,2.074052279,-2.452298708,7.544599094,2.21E-13,2.26E-10,lncRNA,7,UP,AC099552.3,60.6903164,1043,ALL,MAYO
Diagnosis,CBE,AD-CONTROL,ENSG00000286872,-0.181247795,-0.341187573,-0.021308017,0.032732484,-2.226401902,0.026439738,0.06693269,lncRNA,15,NONE,AC024270.5,46.5131873,13422,ALL,MAYO
Diagnosis,CBE,AD-CONTROL,ENSG00000144228,0.080531879,-0.016465937,0.177529695,4.913442808,1.63093857,0.10355023,0.193939662,protein_coding,2,NONE,SPOPL,36.67609796,71747,ALL,MAYO
Diagnosis,CBE,AD-CONTROL,ENSG00000285778,0.011774265,-0.225427841,0.248976371,-1.286000104,0.097352193,0.922486354,0.950960873,lncRNA,1,NONE,AL591463.1,36.23399051,127971,ALL,MAYO
Diagnosis.AOD,CBE,AD-CONTROL,ENSG00000005339,0.074072272,0.042805132,0.105339413,7.626719604,4.656313999,4.15E-06,8.32E-05,protein_coding,16,NONE,CREBBP,45.00523533,155673,ALL,MAYO
Diagnosis.AOD,CBE,AD-CONTROL,ENSG00000188994,0.03375383,-0.010318552,0.077826212,6.703806497,1.501322866,0.133920382,0.239511453,protein_coding,6,NONE,ZNF292,37.67626491,113111,ALL,MAYO
Diagnosis.AOD,CBE,AD-CONTROL,ENSG00000261823,-0.008434191,-0.077650231,0.060781849,-0.970547902,-0.239149259,0.811089637,0.873760793,lncRNA,15,NONE,AC084782.2,42.13615023,1704,ALL,MAYO
Diagnosis.Sex,CBE,AD-CONTROL,ENSG00000146733,0.245221101,0.088916757,0.401525444,3.6245628,3.081227128,0.002180227,0.025429476,protein_coding,7,NONE,PSPH,46.3530108,40554,FEMALE,MAYO
Diagnosis.Sex,CBE,AD-CONTROL,ENSG00000164574,0.111683536,-0.075953728,0.299320799,4.556619644,1.16706539,0.24377083,0.459714102,protein_coding,5,NONE,GALNT10,44.28655187,230255,FEMALE,MAYO
Diagnosis.Sex,CBE,AD-CONTROL,ENSG00000105127,0.036793733,-0.059745776,0.133333242,4.795637019,0.751265208,0.452857433,0.660329327,protein_coding,19,NONE,AKAP8,53.23691982,26414,FEMALE,MAYO
Diagnosis.Sex,CBE,AD-CONTROL,ENSG00000205790,0.347641938,0.137145533,0.558138342,-0.677376322,3.255063005,0.001212908,0.010745247,lncRNA,19,UP,DPP9-AS1,59.2620369,6667,MALE,MAYO
Diagnosis.Sex,CBE,AD-CONTROL,ENSG00000171469,-0.147816242,-0.269059529,-0.026572954,3.644836924,-2.399058545,0.016813764,0.066411124,protein_coding,19,NONE,ZNF561,44.45249743,16557,MALE,MAYO
Diagnosis.Sex,CBE,AD-CONTROL,ENSG00000064703,-0.055752044,-0.165988677,0.054484588,3.592526425,-0.996676453,0.319418656,0.497558937,protein_coding,1,NONE,DDX20,40.26340546,12756,MALE,MAYO
Diagnosis,FP,AD-CONTROL,ENSG00000260751,-0.247109844,-0.416792039,-0.077427649,0.47017593,-2.856419551,0.004377501,0.037095362,lncRNA,16,NONE,AC008870.2,43.65234375,1024,ALL,MSSM
Diagnosis,FP,AD-CONTROL,ENSG00000170017,0.116124024,0.013752868,0.218495179,7.605164653,2.220609241,0.026685651,0.111182878,protein_coding,3,NONE,ALCAM,35.28991581,209992,ALL,MSSM
Diagnosis,FP,AD-CONTROL,ENSG00000235290,-0.057859204,-0.252212735,0.136494327,-0.16909436,-0.584873994,0.558811376,0.712709557,unprocessed_pseudogene,6,NONE,HLA-W,53.26582278,1975,ALL,MSSM
Diagnosis,FP,AD-CONTROL,ENSG00000177728,0.017353424,-0.086789821,0.12149667,6.788782124,0.32609337,0.744427926,0.845969576,protein_coding,17,NONE,TMEM94,52.00188889,59294,ALL,MSSM
Diagnosis.Sex,FP,AD-CONTROL,ENSG00000172780,-0.434385421,-0.717838177,-0.150932665,0.396064873,-3.00527216,0.00271497,0.065139157,protein_coding,3,NONE,RAB43,50.78477564,35233,MALE,MSSM
Diagnosis.Sex,FP,AD-CONTROL,ENSG00000269054,-0.155588882,-0.396743565,0.0855658,0.942597545,-1.265404041,0.206045462,0.491878373,lncRNA,19,NONE,AC012313.6,56.27619238,5011,MALE,MSSM
Diagnosis.Sex,FP,AD-CONTROL,ENSG00000197497,0.079706407,-0.081429533,0.240842348,2.881636183,0.973898187,0.33048948,0.611179707,protein_coding,19,NONE,ZNF665,45.25955405,34174,MALE,MSSM
Diagnosis.Sex,FP,AD-CONTROL,ENSG00000155827,0.02916353,-0.045914974,0.104242034,5.680912284,0.768042101,0.442712225,0.699692125,protein_coding,9,NONE,RNF20,38.16967313,29492,MALE,MSSM
Diagnosis,ACC,AD-CONTROL,ENSG00000125841,-0.181579939,-0.256616609,-0.10654327,6.725621038,-4.743548393,2.22E-06,6.90E-05,protein_coding,20,NONE,NRSN2,51.93726221,12879,ALL,ROSMAP
Diagnosis,ACC,AD-CONTROL,ENSG00000135898,-0.074241909,-0.471057434,0.322573617,0.222262803,-0.366638152,0.713920945,0.827016898,protein_coding,2,NONE,GPR55,48.82695492,53749,ALL,ROSMAP
Diagnosis,ACC,AD-CONTROL,ENSG00000151834,-0.027222811,-0.121567916,0.067122295,6.530023004,-0.56550242,0.571784411,0.72424515,protein_coding,4,NONE,GABRA2,35.62900125,226804,ALL,ROSMAP
Diagnosis.AOD,ACC,AD-CONTROL,ENSG00000099365,-0.093331151,-0.124320202,-0.0623421,8.150447477,-5.901609746,4.10E-09,4.84E-08,protein_coding,16,NONE,STX1B,55.21208437,21383,ALL,ROSMAP
26 changes: 26 additions & 0 deletions tests/test_assets/rna_distribution_data/input/test_rna_distribution_data_missing_values.csv
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,26 @@
model,tissue,comparison,ensembl_gene_id,logfc,ci_l,ci_r,aveexpr,t,p_value,adj_p_val,gene_biotype,chromosome_name,direction,hgnc_symbol,percentage_gc_content,gene_length,sex,study
,CBE,AD-CONTROL,ENSG00000228521,1.646715414,1.219378549,2.074052279,-2.452298708,7.544599094,2.21E-13,2.26E-10,lncRNA,7,UP,AC099552.3,60.6903164,1043,ALL,MAYO
Diagnosis,CBE,AD-CONTROL,ENSG00000286872,-0.181247795,-0.341187573,-0.021308017,0.032732484,-2.226401902,0.026439738,0.06693269,lncRNA,15,NONE,AC024270.5,46.5131873,13422,ALL,MAYO
Diagnosis,CBE,AD-CONTROL,ENSG00000144228,0.080531879,-0.016465937,0.177529695,4.913442808,1.63093857,0.10355023,0.193939662,protein_coding,2,NONE,SPOPL,36.67609796,71747,ALL,MAYO
Diagnosis,CBE,AD-CONTROL,ENSG00000285778,0.011774265,-0.225427841,0.248976371,-1.286000104,0.097352193,0.922486354,0.950960873,lncRNA,1,NONE,AL591463.1,36.23399051,127971,ALL,MAYO
Diagnosis.AOD,CBE,AD-CONTROL,ENSG00000005339,0.074072272,0.042805132,0.105339413,7.626719604,4.656313999,4.15E-06,8.32E-05,protein_coding,16,NONE,CREBBP,45.00523533,155673,ALL,MAYO
Diagnosis.AOD,,AD-CONTROL,ENSG00000188994,0.03375383,-0.010318552,0.077826212,6.703806497,1.501322866,0.133920382,0.239511453,protein_coding,6,NONE,ZNF292,37.67626491,113111,ALL,MAYO
Diagnosis.AOD,CBE,AD-CONTROL,ENSG00000261823,-0.008434191,-0.077650231,0.060781849,-0.970547902,-0.239149259,0.811089637,0.873760793,lncRNA,15,NONE,AC084782.2,42.13615023,1704,ALL,MAYO
Diagnosis.Sex,CBE,AD-CONTROL,ENSG00000146733,0.245221101,0.088916757,0.401525444,3.6245628,3.081227128,0.002180227,0.025429476,protein_coding,7,NONE,PSPH,46.3530108,40554,FEMALE,MAYO
Diagnosis.Sex,CBE,AD-CONTROL,ENSG00000164574,0.111683536,-0.075953728,0.299320799,4.556619644,1.16706539,0.24377083,0.459714102,protein_coding,5,NONE,GALNT10,44.28655187,230255,FEMALE,MAYO
Diagnosis.Sex,CBE,AD-CONTROL,ENSG00000105127,0.036793733,-0.059745776,0.133333242,4.795637019,0.751265208,0.452857433,0.660329327,protein_coding,19,NONE,AKAP8,53.23691982,26414,FEMALE,MAYO
Diagnosis.Sex,CBE,AD-CONTROL,ENSG00000205790,0.347641938,0.137145533,0.558138342,-0.677376322,3.255063005,0.001212908,0.010745247,lncRNA,19,UP,DPP9-AS1,59.2620369,6667,MALE,MAYO
Diagnosis.Sex,CBE,AD-CONTROL,ENSG00000171469,-0.147816242,-0.269059529,-0.026572954,3.644836924,-2.399058545,0.016813764,0.066411124,protein_coding,19,NONE,ZNF561,44.45249743,16557,MALE,MAYO
Diagnosis.Sex,CBE,AD-CONTROL,ENSG00000064703,-0.055752044,-0.165988677,0.054484588,3.592526425,-0.996676453,0.319418656,0.497558937,protein_coding,1,NONE,DDX20,40.26340546,12756,MALE,MAYO
Diagnosis,FP,AD-CONTROL,ENSG00000260751,-0.247109844,-0.416792039,-0.077427649,0.47017593,-2.856419551,0.004377501,0.037095362,lncRNA,16,NONE,AC008870.2,43.65234375,1024,ALL,MSSM
Diagnosis,FP,AD-CONTROL,ENSG00000170017,,0.013752868,0.218495179,7.605164653,2.220609241,0.026685651,0.111182878,protein_coding,3,NONE,ALCAM,35.28991581,209992,ALL,MSSM
Diagnosis,FP,AD-CONTROL,ENSG00000235290,-0.057859204,-0.252212735,0.136494327,-0.16909436,-0.584873994,0.558811376,0.712709557,unprocessed_pseudogene,6,NONE,HLA-W,53.26582278,1975,ALL,MSSM
Diagnosis,FP,AD-CONTROL,ENSG00000177728,0.017353424,-0.086789821,0.12149667,6.788782124,0.32609337,0.744427926,0.845969576,protein_coding,17,NONE,TMEM94,52.00188889,59294,ALL,MSSM
Diagnosis.Sex,FP,AD-CONTROL,ENSG00000172780,,-0.717838177,-0.150932665,0.396064873,-3.00527216,0.00271497,0.065139157,protein_coding,3,NONE,RAB43,50.78477564,35233,MALE,MSSM
Diagnosis.Sex,FP,AD-CONTROL,ENSG00000269054,,-0.396743565,0.0855658,0.942597545,-1.265404041,0.206045462,0.491878373,lncRNA,19,NONE,AC012313.6,56.27619238,5011,MALE,MSSM
Diagnosis.Sex,FP,AD-CONTROL,ENSG00000197497,,-0.081429533,0.240842348,2.881636183,0.973898187,0.33048948,0.611179707,protein_coding,19,NONE,ZNF665,45.25955405,34174,MALE,MSSM
Diagnosis.Sex,FP,AD-CONTROL,ENSG00000155827,,-0.045914974,0.104242034,5.680912284,0.768042101,0.442712225,0.699692125,protein_coding,9,NONE,RNF20,38.16967313,29492,MALE,MSSM
Diagnosis,ACC,AD-CONTROL,ENSG00000125841,-0.181579939,-0.256616609,-0.10654327,6.725621038,-4.743548393,2.22E-06,6.90E-05,protein_coding,20,NONE,NRSN2,51.93726221,12879,ALL,ROSMAP
Diagnosis,ACC,AD-CONTROL,ENSG00000135898,-0.074241909,-0.471057434,0.322573617,0.222262803,-0.366638152,0.713920945,0.827016898,protein_coding,2,NONE,GPR55,48.82695492,53749,ALL,ROSMAP
Diagnosis,ACC,AD-CONTROL,ENSG00000151834,-0.027222811,-0.121567916,0.067122295,6.530023004,-0.56550242,0.571784411,0.72424515,protein_coding,4,NONE,GABRA2,35.62900125,226804,ALL,ROSMAP
Diagnosis.AOD,ACC,AD-CONTROL,ENSG00000099365,-0.093331151,-0.124320202,-0.0623421,8.150447477,-5.901609746,4.10E-09,4.84E-08,protein_coding,16,NONE,STX1B,55.21208437,21383,ALL,ROSMAP
74 changes: 74 additions & 0 deletions tests/test_assets/rna_distribution_data/output/rna_distribution_data_good_output.json
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,74 @@
[
{
"model": "AD Diagnosis (males and females)",
"tissue": "ACC",
"min": -0.2437,
"max": 0.065,
"first_quartile": -0.1279,
"median": -0.0742,
"third_quartile": -0.0507
},
{
"model": "AD Diagnosis x AOD (males and females)",
"tissue": "ACC",
"min": -0.0933,
"max": -0.0933,
"first_quartile": -0.0933,
"median": -0.0933,
"third_quartile": -0.0933
},
{
"model": "AD Diagnosis (males and females)",
"tissue": "CBE",
"min": -0.7993,
"max": 1.2349,
"first_quartile": -0.0365,
"median": 0.0462,
"third_quartile": 0.4721
},
{
"model": "AD Diagnosis x AOD (males and females)",
"tissue": "CBE",
"min": -0.0492,
"max": 0.1158,
"first_quartile": 0.0127,
"median": 0.0338,
"third_quartile": 0.0539
},
{
"model": "AD Diagnosis x Sex (females only)",
"tissue": "CBE",
"min": -0.0821,
"max": 0.3348,
"first_quartile": 0.0742,
"median": 0.1117,
"third_quartile": 0.1785
},
{
"model": "AD Diagnosis x Sex (males only)",
"tissue": "CBE",
"min": -0.4734,
"max": 0.5175,
"first_quartile": -0.1018,
"median": -0.0558,
"third_quartile": 0.1459
},
{
"model": "AD Diagnosis (males and females)",
"tissue": "FP",
"min": -0.326,
"max": 0.2629,
"first_quartile": -0.1052,
"median": -0.0203,
"third_quartile": 0.042
},
{
"model": "AD Diagnosis x Sex (males only)",
"tissue": "FP",
"min": -0.6259,
"max": 0.4424,
"first_quartile": -0.2253,
"median": -0.0632,
"third_quartile": 0.0418
}
]
74 changes: 74 additions & 0 deletions ...s/test_assets/rna_distribution_data/output/rna_distribution_data_missing_data_output.json
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,74 @@
[
{
"model": "AD Diagnosis (males and females)",
"tissue": "ACC",
"min": -0.2437,
"max": 0.065,
"first_quartile": -0.1279,
"median": -0.0742,
"third_quartile": -0.0507
},
{
"model": "AD Diagnosis x AOD (males and females)",
"tissue": "ACC",
"min": -0.0933,
"max": -0.0933,
"first_quartile": -0.0933,
"median": -0.0933,
"third_quartile": -0.0933
},
{
"model": "AD Diagnosis (males and females)",
"tissue": "CBE",
"min": -0.2811,
"max": 0.2425,
"first_quartile": -0.0847,
"median": 0.0118,
"third_quartile": 0.0462
},
{
"model": "AD Diagnosis x AOD (males and females)",
"tissue": "CBE",
"min": -0.0497,
"max": 0.1153,
"first_quartile": 0.0122,
"median": 0.0328,
"third_quartile": 0.0534
},
{
"model": "AD Diagnosis x Sex (females only)",
"tissue": "CBE",
"min": -0.0821,
"max": 0.3348,
"first_quartile": 0.0742,
"median": 0.1117,
"third_quartile": 0.1785
},
{
"model": "AD Diagnosis x Sex (males only)",
"tissue": "CBE",
"min": -0.4734,
"max": 0.5175,
"first_quartile": -0.1018,
"median": -0.0558,
"third_quartile": 0.1459
},
{
"model": "AD Diagnosis (males and females)",
"tissue": "FP",
"min": -0.3508,
"max": 0.1781,
"first_quartile": -0.1525,
"median": -0.0579,
"third_quartile": -0.0203
},
{
"model": "AD Diagnosis x Sex (males only)",
"tissue": "FP",
"min": null,
"max": null,
"first_quartile": null,
"median": null,
"third_quartile": null
}
]
63 changes: 63 additions & 0 deletions tests/transform/test_rna_distribution_data.py
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,63 @@
import os

import pandas as pd
import pytest

from agoradatatools.etl.transform import rna_distribution


class TestTransformRnaDistributionData:
data_files_path = "tests/test_assets/rna_distribution_data"
pass_test_data = [
( # pass with good data
"test_rna_distribution_data_good_input.csv",
"rna_distribution_data_good_output.json",
),
( # pass with missing data
"test_rna_distribution_data_missing_values.csv",
"rna_distribution_data_missing_data_output.json",
),
]
pass_test_ids = [
"Pass with good data",
"Pass with missing data",
]
fail_test_data = [
( # Fail with a TypeError due to string value in logfc
"test_rna_distribution_data_bad_input_typeerror.csv",
TypeError,
),
( # Fail with a KeyError due to too many missing values
"test_rna_distribution_data_bad_input_keyerror.csv",
KeyError,
),
]
fail_test_ids = [
"Fail with bad data type",
"Fail with too many missing values",
]

@pytest.mark.parametrize(
"input_file, expected_output_file", pass_test_data, ids=pass_test_ids
)
def test_transform_rna_distribution_data_should_pass(
self, input_file, expected_output_file
):
input_df = pd.read_csv(os.path.join(self.data_files_path, "input", input_file))
output_df = rna_distribution.transform_rna_distribution_data(
datasets={"diff_exp_data": input_df}
)
expected_df = pd.read_json(
os.path.join(self.data_files_path, "output", expected_output_file),
)
pd.testing.assert_frame_equal(output_df, expected_df)

@pytest.mark.parametrize("input_file, error_type", fail_test_data, ids=fail_test_ids)
def test_transform_rna_distribution_data_should_fail(self, input_file, error_type):
with pytest.raises(error_type):
input_df = pd.read_csv(
os.path.join(self.data_files_path, "input", input_file)
)
rna_distribution.transform_rna_distribution_data(
datasets={"diff_exp_data": input_df}
)

0 comments on commit e26e735

Please sign in to comment.