import pandas as pd
from IPython.display import display
from pkdb_analysis import PKData, PKFilter
from pkdb_analysis.test import TESTDATA_CONCISE_FALSE_ZIP
Filter data¶
A recurring task is to filter data for a certain question. E.g. to compare two groups, or get the subset of data for all healthy smokers.
We work again with our test data set and will filter various subsets from it.
test_data = PKData.from_archive(TESTDATA_CONCISE_FALSE_ZIP)
print(test_data)
------------------------------
PKData (139890849755408)
------------------------------
studies 127 ( 127)
groups 429 ( 3733)
individuals 3163 (28219)
interventions 409 ( 409)
outputs 19407 (29053)
timecourses 722 ( 722)
scatters 80 ( 80)
------------------------------
Filter functions¶
The main principle for filtering PKData is by using the filter_*
and exclude_* functionality.
A key principle are hereby filter functions which for a given DataFrame return a logical index.
Depending on which subset of information this should be applied the
groups, individuals, subjects (groups and individuals),
outputs and timecourses.
Filter by study_sid¶
A first example is filtering by study_sid, i.e. we only want the
subset of data from a single study. An overview over the existing study
sids in the dataset is available via
test_data.study_sids
{'1033273',
'10634135',
'11839447',
'13053413',
'15022032',
'16158445',
'16261361',
'18703023',
'2060878',
'20853468',
'21591074',
'2185297',
'22673010',
'25853045',
'25891161',
'26862045',
'28929443',
'2921843',
'29230348',
'30387917',
'32071850',
'3356110',
'3557314',
'3741730',
'4027137',
'6135578',
'6712142',
'6734698',
'7371463',
'7371698',
'7742721',
'7775610',
'8445222',
'Arnaud1981',
'Arold2005',
'Bchir2006',
'Becker1984',
'Bozikas2004',
'Broughton1981',
'Callahan1982',
'Cattarossi1988',
'Christensen2002',
'Cysneiros2007',
'Djordjevic2008',
'He2017',
'Kamimori1999',
'Laizure2017',
'Lane1992',
'Lennard1982',
'Matthaei2016',
'Nakazawa1988',
'PKDB00001',
'PKDB00002',
'PKDB00003',
'PKDB00004',
'PKDB00005',
'PKDB00006',
'PKDB00007',
'PKDB00008',
'PKDB00009',
'PKDB00010',
'PKDB00011',
'PKDB00012',
'PKDB00013',
'PKDB00014',
'PKDB00015',
'PKDB00016',
'PKDB00017',
'PKDB00018',
'PKDB00019',
'PKDB00032',
'PKDB00033',
'PKDB00034',
'PKDB00035',
'PKDB00036',
'PKDB00037',
'PKDB00038',
'PKDB00039',
'PKDB00040',
'PKDB00041',
'PKDB00042',
'PKDB00043',
'PKDB00044',
'PKDB00045',
'PKDB00046',
'PKDB00047',
'PKDB00048',
'PKDB00049',
'PKDB00050',
'PKDB00051',
'PKDB00052',
'PKDB00053',
'PKDB00054',
'PKDB00055',
'PKDB00056',
'PKDB00057',
'PKDB00058',
'PKDB00059',
'PKDB00060',
'PKDB00061',
'PKDB00062',
'PKDB00063',
'PKDB00065',
'PKDB00126',
'PKDB00127',
'PKDB00128',
'PKDB00129',
'PKDB00136',
'PKDB00137',
'PKDB00138',
'PKDB00210',
'PKDB00328',
'PKDB00338',
'PKDB00339',
'PKDB00341',
'PKDB00378',
'PKDB00380',
'PKDB00381',
'PKDB00382',
'PKDB00383',
'Sandberg1988',
'Scott1989',
'Tanaka1993',
'Trang1985'}
Filtering a subset of data works by providing filter/selection functions
which select a subset of the data. The filters are written on the
groups, individuals
def is_PKDB99999(d):
"""Filter for specific study_sid. """
return d.study_sid == "PKDB99999"
data = test_data.filter_intervention(is_PKDB99999)
print(data)
------------------------------
PKData (139891787702672)
------------------------------
studies 0 ( 0)
groups 0 ( 0)
individuals 0 ( 0)
interventions 0 ( 0)
outputs 0 ( 0)
timecourses 0 ( 0)
scatters 80 ( 80)
------------------------------
The PKData now only contains data for the given study_sid:
print(data.study_sids)
{'Lennard1982', '25853045', 'PKDB00002', '25891161', '6712142', 'PKDB00045', 'Trang1985', 'PKDB00210', 'PKDB00012', '1033273', '4027137', 'PKDB00126', '7371463', '15022032', 'Sandberg1988', '3557314', 'Nakazawa1988', '2921843', '2185297', '28929443', 'PKDB00015', '26862045'}
# for instance interventions
display(data.interventions)
| Unnamed: 0 | study_sid | study_name | intervention_pk | raw_pk | normed | name | route | route_label | form | ... | substance_label | value | mean | median | min | max | sd | se | cv | unit |
|---|
0 rows × 31 columns
Empty DataFrame
Columns: [Unnamed: 0, study_sid, study_name, intervention_pk, raw_pk, normed, name, route, route_label, form, form_label, application, application_label, time, time_end, time_unit, measurement_type, measurement_type_label, choice, choice_label, substance, substance_label, value, mean, median, min, max, sd, se, cv, unit]
Index: []
[0 rows x 31 columns]
One could also define this as a simple lambda function
data = test_data.filter_intervention(lambda d: d.study_sid == "PKDB99999")
print(data)
------------------------------
PKData (139891787682832)
------------------------------
studies 0 ( 0)
groups 0 ( 0)
individuals 0 ( 0)
interventions 0 ( 0)
outputs 0 ( 0)
timecourses 0 ( 0)
scatters 80 ( 80)
------------------------------
Concise data¶
All operations on PKData leave the data in a consistent manner. E.g.
if an intervention is filtered out also all the outputs using this
intervention are filtered out. This behavior is controlled by the
concise flag on most operations.
t1 = test_data.filter_intervention(is_PKDB99999)
t2 = test_data.filter_intervention(is_PKDB99999, concise=False)
print(t1)
print(t2)
------------------------------
PKData (139890839716880)
------------------------------
studies 0 ( 0)
groups 0 ( 0)
individuals 0 ( 0)
interventions 0 ( 0)
outputs 0 ( 0)
timecourses 0 ( 0)
scatters 80 ( 80)
------------------------------
------------------------------
PKData (139890839717712)
------------------------------
studies 127 ( 127)
groups 429 ( 3733)
individuals 3163 (28219)
interventions 0 ( 0)
outputs 19407 (29053)
timecourses 722 ( 722)
scatters 80 ( 80)
------------------------------
# FIXME: only normed data
t1.interventions_mi
t2.interventions_mi
t2.outputs
| Unnamed: 0 | study_name | measurement_type | tissue | sd | se | min | group_pk | output_pk | time_unit | ... | max | substance | label | individual_pk | unit | cv | median | mean | time | choice | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 0 | Abernethy1985 | thalf | plasma | NaN | NaN | NaN | -1 | 210625 | NaN | ... | NaN | caf | NaN | 23952 | hour | NaN | NaN | NaN | NaN | NaN |
| 1 | 1 | Abernethy1985 | thalf | plasma | NaN | NaN | NaN | -1 | 210628 | NaN | ... | NaN | caf | NaN | 23955 | hour | NaN | NaN | NaN | NaN | NaN |
| 2 | 2 | Abernethy1985 | thalf | plasma | NaN | NaN | NaN | -1 | 210631 | NaN | ... | NaN | caf | NaN | 23958 | hour | NaN | NaN | NaN | NaN | NaN |
| 3 | 3 | Abernethy1985 | thalf | plasma | NaN | NaN | NaN | -1 | 210635 | NaN | ... | NaN | caf | NaN | 23962 | hour | NaN | NaN | NaN | NaN | NaN |
| 4 | 4 | Abernethy1985 | clearance | plasma | NaN | NaN | NaN | -1 | 210640 | NaN | ... | NaN | caf | NaN | 23967 | liter / hour / kilogram | NaN | NaN | NaN | NaN | NaN |
| ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... |
| 29048 | 29048 | Barnett1990 | cmax | plasma | NaN | NaN | NaN | -1 | 263142 | NaN | ... | NaN | caf | NaN | 30567 | gram / liter | NaN | NaN | NaN | NaN | NaN |
| 29049 | 29049 | Barnett1990 | vd-ss | plasma | NaN | NaN | NaN | -1 | 263146 | NaN | ... | NaN | caf | NaN | 30567 | liter | NaN | NaN | NaN | NaN | NaN |
| 29050 | 29050 | Barnett1990 | cmax | plasma | NaN | NaN | NaN | -1 | 263159 | NaN | ... | NaN | caf | NaN | 30567 | gram / liter | NaN | NaN | NaN | NaN | NaN |
| 29051 | 29051 | Barnett1990 | vd | plasma | NaN | NaN | NaN | -1 | 263163 | NaN | ... | NaN | caf | NaN | 30567 | liter | NaN | NaN | NaN | NaN | NaN |
| 29052 | 29052 | Barnett1990 | vd-ss | plasma | NaN | NaN | NaN | -1 | 263164 | NaN | ... | NaN | caf | NaN | 30567 | liter | NaN | NaN | NaN | NaN | NaN |
29053 rows × 27 columns
Unnamed: 0 study_name measurement_type tissue sd se min 0 0 Abernethy1985 thalf plasma NaN NaN NaN
1 1 Abernethy1985 thalf plasma NaN NaN NaN
2 2 Abernethy1985 thalf plasma NaN NaN NaN
3 3 Abernethy1985 thalf plasma NaN NaN NaN
4 4 Abernethy1985 clearance plasma NaN NaN NaN
... ... ... ... ... .. .. ...
29048 29048 Barnett1990 cmax plasma NaN NaN NaN
29049 29049 Barnett1990 vd-ss plasma NaN NaN NaN
29050 29050 Barnett1990 cmax plasma NaN NaN NaN
29051 29051 Barnett1990 vd plasma NaN NaN NaN
29052 29052 Barnett1990 vd-ss plasma NaN NaN NaN
group_pk output_pk time_unit ... max substance label 0 -1 210625 NaN ... NaN caf NaN
1 -1 210628 NaN ... NaN caf NaN
2 -1 210631 NaN ... NaN caf NaN
3 -1 210635 NaN ... NaN caf NaN
4 -1 210640 NaN ... NaN caf NaN
... ... ... ... ... ... ... ...
29048 -1 263142 NaN ... NaN caf NaN
29049 -1 263146 NaN ... NaN caf NaN
29050 -1 263159 NaN ... NaN caf NaN
29051 -1 263163 NaN ... NaN caf NaN
29052 -1 263164 NaN ... NaN caf NaN
individual_pk unit cv median mean time choice
0 23952 hour NaN NaN NaN NaN NaN
1 23955 hour NaN NaN NaN NaN NaN
2 23958 hour NaN NaN NaN NaN NaN
3 23962 hour NaN NaN NaN NaN NaN
4 23967 liter / hour / kilogram NaN NaN NaN NaN NaN
... ... ... .. ... ... ... ...
29048 30567 gram / liter NaN NaN NaN NaN NaN
29049 30567 liter NaN NaN NaN NaN NaN
29050 30567 gram / liter NaN NaN NaN NaN NaN
29051 30567 liter NaN NaN NaN NaN NaN
29052 30567 liter NaN NaN NaN NaN NaN
[29053 rows x 27 columns]
Query groups and individuals¶
2.1 Get data for groups with characteristica/keywords X¶
healthy=True, smoking=N, disease=None, individual queries and combinations.
def is_healthy(d):
# healthy is reported and True
return (d.measurement_type == "healthy") & (d.choice == "Y")
def disease(d):
# any disease is reported
return d.measurement_type == "disease"
def smoking(d):
# smoking status is curated for study (this could by Y/N/NR)
return d.measurement_type == "smoking"
def nonsmoker(d):
# smoking is reported and no
return smoking(d) & (d.choice == "N")
def smoker(d):
# smoking is reported and yes
return smoking(d) & (d.choice == "Y")
test_data = PKData.from_archive(TESTDATA_CONCISE_FALSE_ZIP)
f_idx can be a single function, or a list of functions. A list of
functions are applied successively and is equivalent to “AND logic”. “OR
logic” can be directly applied on the index.
healthy_nonsmoker = test_data.filter_subject(f_idx=[is_healthy, nonsmoker])
print(healthy_nonsmoker)
healthy_nonsmoker.groups_mi
------------------------------
PKData (139890849157456)
------------------------------
studies 81 ( 81)
groups 143 ( 1396)
individuals 1150 (10810)
interventions 238 ( 238)
outputs 10578 (14843)
timecourses 438 ( 438)
scatters 80 ( 80)
------------------------------
| Unnamed: 0 | study_name | study_sid | measurement_type | group_count | group_name | max | substance | count | group_parent_pk | sd | unit | se | min | cv | median | mean | choice | value | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| group_pk | characteristica_pk | |||||||||||||||||||
| 3463 | 67383 | 10 | Abernethy1985 | PKDB00001 | smoking | 9 | OCS | NaN | nan | 18 | 3462 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | N | NaN |
| 67384 | 11 | Abernethy1985 | PKDB00001 | age | 9 | OCS | 30.0 | nan | 18 | 3462 | NaN | year | 1.0 | 23.0 | NaN | NaN | 26.0 | NaN | NaN | |
| 67385 | 12 | Abernethy1985 | PKDB00001 | species | 9 | OCS | NaN | nan | 18 | 3462 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | homo sapiens | NaN | |
| 67386 | 13 | Abernethy1985 | PKDB00001 | healthy | 9 | OCS | NaN | nan | 18 | 3462 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | Y | NaN | |
| 67387 | 14 | Abernethy1985 | PKDB00001 | sex | 9 | OCS | NaN | nan | 18 | 3462 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | F | NaN | |
| ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... |
| 4001 | 80932 | 3677 | Tian2019 | 30387917 | CYP1A2 genotype | 12 | men | NaN | nan | 1 | 3999 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | *1a/*1a | NaN |
| 80933 | 3678 | Tian2019 | 30387917 | CYP1A2 genotype | 12 | men | NaN | nan | 1 | 3999 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | *1c/*1f | NaN | |
| 80934 | 3681 | Tian2019 | 30387917 | CYP1A2 genotype | 12 | men | NaN | nan | 2 | 3999 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | *1c*1f/*1c*1f | NaN | |
| 80935 | 3682 | Tian2019 | 30387917 | CYP1A2 genotype | 12 | men | NaN | nan | 6 | 3999 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | *1a/*1f | NaN | |
| 80936 | 3683 | Tian2019 | 30387917 | CYP1A2 genotype | 12 | men | NaN | nan | 2 | 3999 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | *1f/*1f | NaN |
1396 rows × 19 columns
Unnamed: 0 study_name study_sid group_pk characteristica_pk
3463 67383 10 Abernethy1985 PKDB00001
67384 11 Abernethy1985 PKDB00001
67385 12 Abernethy1985 PKDB00001
67386 13 Abernethy1985 PKDB00001
67387 14 Abernethy1985 PKDB00001
... ... ... ...
4001 80932 3677 Tian2019 30387917
80933 3678 Tian2019 30387917
80934 3681 Tian2019 30387917
80935 3682 Tian2019 30387917
80936 3683 Tian2019 30387917
measurement_type group_count group_name max group_pk characteristica_pk
3463 67383 smoking 9 OCS NaN
67384 age 9 OCS 30.0
67385 species 9 OCS NaN
67386 healthy 9 OCS NaN
67387 sex 9 OCS NaN
... ... ... ... ...
4001 80932 CYP1A2 genotype 12 men NaN
80933 CYP1A2 genotype 12 men NaN
80934 CYP1A2 genotype 12 men NaN
80935 CYP1A2 genotype 12 men NaN
80936 CYP1A2 genotype 12 men NaN
substance count group_parent_pk sd unit se group_pk characteristica_pk
3463 67383 nan 18 3462 NaN NaN NaN
67384 nan 18 3462 NaN year 1.0
67385 nan 18 3462 NaN NaN NaN
67386 nan 18 3462 NaN NaN NaN
67387 nan 18 3462 NaN NaN NaN
... ... ... ... .. ... ...
4001 80932 nan 1 3999 NaN NaN NaN
80933 nan 1 3999 NaN NaN NaN
80934 nan 2 3999 NaN NaN NaN
80935 nan 6 3999 NaN NaN NaN
80936 nan 2 3999 NaN NaN NaN
min cv median mean choice value
group_pk characteristica_pk
3463 67383 NaN NaN NaN NaN N NaN
67384 23.0 NaN NaN 26.0 NaN NaN
67385 NaN NaN NaN NaN homo sapiens NaN
67386 NaN NaN NaN NaN Y NaN
67387 NaN NaN NaN NaN F NaN
... ... .. ... ... ... ...
4001 80932 NaN NaN NaN NaN *1a/*1a NaN
80933 NaN NaN NaN NaN *1c/*1f NaN
80934 NaN NaN NaN NaN *1c*1f/*1c*1f NaN
80935 NaN NaN NaN NaN *1a/*1f NaN
80936 NaN NaN NaN NaN *1f/*1f NaN
[1396 rows x 19 columns]
Often attributes are mixed for groups so we have to exclude the
opposites. In the example, the group 20 consists of 5 smokers and 1
nonsmoker. So for a subset of the group smoking is No. We can exclude
groups via
healthy_nonsmoker = test_data.filter_subject([is_healthy, nonsmoker]).exclude_subject([smoker])
print(healthy_nonsmoker)
display(healthy_nonsmoker.groups_mi)
------------------------------
PKData (139890842843216)
------------------------------
studies 74 ( 74)
groups 124 ( 1144)
individuals 927 ( 8619)
interventions 221 ( 221)
outputs 9539 (13730)
timecourses 397 ( 397)
scatters 80 ( 80)
------------------------------
| Unnamed: 0 | study_name | study_sid | measurement_type | group_count | group_name | max | substance | count | group_parent_pk | sd | unit | se | min | cv | median | mean | choice | value | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| group_pk | characteristica_pk | |||||||||||||||||||
| 3463 | 67383 | 10 | Abernethy1985 | PKDB00001 | smoking | 9 | OCS | NaN | nan | 18 | 3462 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | N | NaN |
| 67384 | 11 | Abernethy1985 | PKDB00001 | age | 9 | OCS | 30.0 | nan | 18 | 3462 | NaN | year | 1.0 | 23.0 | NaN | NaN | 26.0 | NaN | NaN | |
| 67385 | 12 | Abernethy1985 | PKDB00001 | species | 9 | OCS | NaN | nan | 18 | 3462 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | homo sapiens | NaN | |
| 67386 | 13 | Abernethy1985 | PKDB00001 | healthy | 9 | OCS | NaN | nan | 18 | 3462 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | Y | NaN | |
| 67387 | 14 | Abernethy1985 | PKDB00001 | sex | 9 | OCS | NaN | nan | 18 | 3462 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | F | NaN | |
| ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... |
| 4001 | 80932 | 3677 | Tian2019 | 30387917 | CYP1A2 genotype | 12 | men | NaN | nan | 1 | 3999 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | *1a/*1a | NaN |
| 80933 | 3678 | Tian2019 | 30387917 | CYP1A2 genotype | 12 | men | NaN | nan | 1 | 3999 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | *1c/*1f | NaN | |
| 80934 | 3681 | Tian2019 | 30387917 | CYP1A2 genotype | 12 | men | NaN | nan | 2 | 3999 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | *1c*1f/*1c*1f | NaN | |
| 80935 | 3682 | Tian2019 | 30387917 | CYP1A2 genotype | 12 | men | NaN | nan | 6 | 3999 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | *1a/*1f | NaN | |
| 80936 | 3683 | Tian2019 | 30387917 | CYP1A2 genotype | 12 | men | NaN | nan | 2 | 3999 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | *1f/*1f | NaN |
1144 rows × 19 columns
Unnamed: 0 study_name study_sid group_pk characteristica_pk
3463 67383 10 Abernethy1985 PKDB00001
67384 11 Abernethy1985 PKDB00001
67385 12 Abernethy1985 PKDB00001
67386 13 Abernethy1985 PKDB00001
67387 14 Abernethy1985 PKDB00001
... ... ... ...
4001 80932 3677 Tian2019 30387917
80933 3678 Tian2019 30387917
80934 3681 Tian2019 30387917
80935 3682 Tian2019 30387917
80936 3683 Tian2019 30387917
measurement_type group_count group_name max group_pk characteristica_pk
3463 67383 smoking 9 OCS NaN
67384 age 9 OCS 30.0
67385 species 9 OCS NaN
67386 healthy 9 OCS NaN
67387 sex 9 OCS NaN
... ... ... ... ...
4001 80932 CYP1A2 genotype 12 men NaN
80933 CYP1A2 genotype 12 men NaN
80934 CYP1A2 genotype 12 men NaN
80935 CYP1A2 genotype 12 men NaN
80936 CYP1A2 genotype 12 men NaN
substance count group_parent_pk sd unit se group_pk characteristica_pk
3463 67383 nan 18 3462 NaN NaN NaN
67384 nan 18 3462 NaN year 1.0
67385 nan 18 3462 NaN NaN NaN
67386 nan 18 3462 NaN NaN NaN
67387 nan 18 3462 NaN NaN NaN
... ... ... ... .. ... ...
4001 80932 nan 1 3999 NaN NaN NaN
80933 nan 1 3999 NaN NaN NaN
80934 nan 2 3999 NaN NaN NaN
80935 nan 6 3999 NaN NaN NaN
80936 nan 2 3999 NaN NaN NaN
min cv median mean choice value
group_pk characteristica_pk
3463 67383 NaN NaN NaN NaN N NaN
67384 23.0 NaN NaN 26.0 NaN NaN
67385 NaN NaN NaN NaN homo sapiens NaN
67386 NaN NaN NaN NaN Y NaN
67387 NaN NaN NaN NaN F NaN
... ... .. ... ... ... ...
4001 80932 NaN NaN NaN NaN *1a/*1a NaN
80933 NaN NaN NaN NaN *1c/*1f NaN
80934 NaN NaN NaN NaN *1c*1f/*1c*1f NaN
80935 NaN NaN NaN NaN *1a/*1f NaN
80936 NaN NaN NaN NaN *1f/*1f NaN
[1144 rows x 19 columns]
In addition often combinations of attributes have to be used to find the
correct subjects. For instance a combination of healthy and reported
disease
def is_healthy(d):
# healthy is reported and True
return (d.measurement_type == "healthy") & (d.choice == "Y")
def disease(d):
# any disease is reported
return d.measurement_type == "disease"
healthy1 = test_data.filter_subject(is_healthy)
healthy2 = test_data.exclude_subject(disease)
healthy3 = test_data.filter_subject(is_healthy).exclude_subject(disease)
print(healthy1)
print(healthy2)
print(healthy3)
------------------------------
PKData (139890842843920)
------------------------------
studies 111 ( 111)
groups 228 ( 2093)
individuals 2165 (17531)
interventions 345 ( 345)
outputs 15746 (25008)
timecourses 607 ( 607)
scatters 80 ( 80)
------------------------------
------------------------------
PKData (139890839534800)
------------------------------
studies 119 ( 119)
groups 234 ( 2107)
individuals 2259 (17906)
interventions 349 ( 349)
outputs 16367 (25485)
timecourses 620 ( 620)
scatters 80 ( 80)
------------------------------
------------------------------
PKData (139890849160720)
------------------------------
studies 110 ( 110)
groups 222 ( 2028)
individuals 2058 (16812)
interventions 337 ( 337)
outputs 15472 (24590)
timecourses 603 ( 603)
scatters 80 ( 80)
------------------------------
3 Query interventions¶
3.1 Get outputs/timecourses for intervention with substance¶
intervention with measurement_type “dosing” and substance “caffeine”
def dosing_and_caffeine(d):
return ((d["measurement_type"]=="dosing") & (d["substance"]=="caffeine"))
3.2 Get outputs/timecourses where multiple interventions were given¶
test_data = PKData.from_archive(TESTDATA_CONCISE_FALSE_ZIP)
caffeine_data = test_data.filter_intervention(dosing_and_caffeine)
print(caffeine_data)
------------------------------
PKData (139890848739472)
------------------------------
studies 0 ( 0)
groups 0 ( 0)
individuals 0 ( 0)
interventions 0 ( 0)
outputs 0 ( 0)
timecourses 0 ( 0)
scatters 80 ( 80)
------------------------------
4 Query outputs/timecourses¶
4.1 query by measurement_type¶
filter all outputs with measurement_type auc_inf
def is_auc_inf(d):
return (d["measurement_type"]=="auc_inf")
test_data = PKData.from_archive(TESTDATA_CONCISE_FALSE_ZIP)
test_data = test_data.filter_output(is_auc_inf).delete_timecourses()
print(test_data)
------------------------------
PKData (139890841988752)
------------------------------
studies 0 ( 0)
groups 0 ( 0)
individuals 0 ( 0)
interventions 0 ( 0)
outputs 0 ( 0)
timecourses 0 ( 0)
scatters 80 ( 80)
------------------------------
5 Other Query others¶
5.1 Complex¶
get clearance of codeine for all.h5 subjects, which have been phenotyped for cyp2d6.
def is_cyp2d6_phenotyped(d):
cyp2d6_phenotype_substances = ['spar/(2hspar+5hspar)', 'deb/4hdeb', 'dtf/dmt']
return d["measurement_type"].isin(["metabolic phenotype", "metabolic ratio"]) & d["substance"].isin(cyp2d6_phenotype_substances)
def codeine_clearance(d):
return (d["measurement_type"]=="clearance") & (d["substance"]=="codeine")
test_data = PKData.from_archive(TESTDATA_CONCISE_FALSE_ZIP)
phenotyped_data = test_data.filter_output(is_cyp2d6_phenotyped)
test_data.groups = phenotyped_data.groups
test_data.individuals = phenotyped_data.individuals
test_data = test_data.filter_output(codeine_clearance).delete_timecourses()
print(test_data)
------------------------------
PKData (139890848654736)
------------------------------
studies 0 ( 0)
groups 0 ( 0)
individuals 0 ( 0)
interventions 0 ( 0)
outputs 0 ( 0)
timecourses 0 ( 0)
scatters 80 ( 80)
------------------------------
6 Pitfalls¶
test_data = PKData.from_archive(TESTDATA_CONCISE_FALSE_ZIP)
# Wrong
def is_healthy_smoker(d):
""" This will yield zero subjects. No characteristica satisfy measurement_type == 'healthy' and measurement_type == 'disease'.
"""
return ((d["measurement_type"]=="healthy") & (d["choice"]=="Y")) & ((d["measurement_type"]=="smoking") & (d["choice"]=="Y"))
# Correct
def is_healthy_smoker(d):
""" """
return [(d["measurement_type"]=="healthy") & (d["choice"]=="Y"), (d["measurement_type"]=="smoking") & (d["choice"]=="Y")]
# Wrong
def not_smoker_y(d):
""" Be care this might not do what you expect. Excluding a specific characteristica will not eliminate any subject unless it is the only characteristica.
"""
return ~((d["measurement_type"]=="smoking") & (d["choice"]=="Y"))
not_smoker_y_data = test_data.filter_subject(not_smoker_y)
#Correct
# exlcude smoker
def smoker_y(d):
return (d["measurement_type"]=="smoking") & (d["choice"]=="Y")
healthy_data = test_data.exclude_subject(smoker_y)
# Wrong
def not_disease(d):
""" Be care this might not do what you expect. Excluding a specific characteristica will not eliminate any subject unless it is the only characteristica
"""
return ~(d["measurement_type"]=="disease")
healthy_data = test_data.filter_subject(not_disease)
# Correct
# exlcude the disease
def disease(d):
return d["measurement_type"]=="disease"
healthy_data = test_data.exclude_subject(disease)