Skip to content

Commit

Permalink
Merge pull request #101 from cherman2/peds-assoc-ref
Browse files Browse the repository at this point in the history 
REF: PEDs/PEDS Associated Methods
  • Loading branch information
@ebolyen
ebolyen authored Nov 8, 2024
2 parents 1cd4738 + a69bb48 commit ed022ec
Show file tree
Hide file tree
Showing 8 changed files with 2,589 additions and 1,458 deletions.
222 changes: 7 additions & 215 deletions q2_fmt/_engraftment.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -8,10 +8,13 @@

import pandas as pd
import itertools
import numpy as np

import qiime2

from q2_fmt._util import (_get_to_baseline_ref, _get_series_from_col,
_sort_multi_index, _ordered_dists,
_independent_dists)


def cc(
ctx, diversity_measure, metadata, compare, distance_to, time_column,
Expand Down Expand Up @@ -83,11 +86,9 @@ def group_timepoints(
original_measure_name = diversity_measure.name
diversity_measure.name = 'measure'
diversity_measure.index.name = 'id'

ordered_df = _ordered_dists(diversity_measure, is_beta, used_references,
time_col, subject_col=subject_col,
group_col=group_col)

id_annotation = {
'title': used_references.index.name,
'description': '...'
Expand Down Expand Up @@ -214,9 +215,10 @@ def _data_filtering(diversity_measure: pd.Series, metadata: qiime2.Metadata,
used_references = reference_col[~time_col.isna()]
elif distance_to == 'baseline':
used_references = \
_get_to_baseline_ref(time_col=time_col, time_column=time_column,
_get_to_baseline_ref(time_col=time_col,
time_column_name=time_column,
baseline_timepoint=baseline_timepoint,
subject_column=subject_column,
subject_column_name=subject_column,
metadata=metadata)
if used_references.isna().any():
if filter_missing_references:
Expand Down Expand Up @@ -268,213 +270,3 @@ def _data_filtering(diversity_measure: pd.Series, metadata: qiime2.Metadata,

return (is_beta, used_references, time_col, subject_col,
group_col, used_controls)


# HELPER FUNCTION FOR DATA Filtering
def _get_series_from_col(md, col_name, param_name, expected_type=None,
drop_missing_values=False):
try:
column = md.get_column(col_name)
except ValueError as e:
raise ValueError("There was an issue with the argument for %r. %s"
% (param_name, e)) from e

if expected_type is not None and not isinstance(column, expected_type):
if type(expected_type) is tuple:
exp = tuple(e.type for e in expected_type)
else:
exp = expected_type.type

raise ValueError("Provided column for %r is %r, not %r."
% (param_name, column.type, exp))

if drop_missing_values:
column = column.drop_missing_values()

return column.to_series()


# HELPER FUNCTION FOR sorting a multi-index (for dist matrix and metadata)
def _sort_multi_index(index):
sorted_levels = list(map(sorted, index))
sorted_multi = pd.MultiIndex.from_tuples(sorted_levels)
return sorted_multi


# HELPER FUNCTION FOR Dists1D[Ordered | NestedOrdered, Matched | Independent]
def _ordered_dists(diversity_measure: pd.Series, is_beta,
used_references, time_col, subject_col, group_col):
if is_beta:
idx = pd.MultiIndex.from_frame(
used_references.to_frame().reset_index())
idx = _sort_multi_index(idx)
idx.names = ['id', 'reference']
diversity_measure.index.names = ['id', 'reference']
else:
idx = used_references.index
idx.name = 'id'

try:
sliced_df = (diversity_measure[idx]
.to_frame()
.reset_index()
.set_index('id')
)
except KeyError:
raise KeyError('Pairwise comparisons were unsuccessful. Please double'
' check that your chosen reference column contains'
' values that are also present in the ID column for'
' the associated metadata.')

if is_beta:
sliced_df.index = used_references.index
sliced_df.index.name = 'id'

ordinal_df = sliced_df[['measure']]
ordinal_df['group'] = time_col

if subject_col is not None:
ordinal_df['subject'] = subject_col

if group_col is not None:
ordinal_df['class'] = group_col.name
ordinal_df['level'] = group_col

return ordinal_df.reset_index()


# HELPER FUNCTION FOR GroupDists[Unordered, Independent]
def _independent_dists(diversity_measure, metadata,
used_references, is_beta, used_controls):
unique_references = sorted(used_references.unique())

if is_beta:
if len(unique_references) > 1:
ref_idx = pd.MultiIndex.from_tuples(
itertools.combinations(unique_references, 2))
ref_idx.names = ['A', 'B']

else:
ref_idx = pd.MultiIndex(levels=[[], []],
codes=[[], []],
names=['A', 'B'])

diversity_measure.index.names = ['A', 'B']

if used_controls is not None:
grouped_md = (metadata
.to_dataframe()
.loc[used_controls.index]
.groupby(used_controls)
)
ctrl_list = list()
for group_id, grouped_ctrls in grouped_md:
if len(grouped_ctrls.index) < 2:
continue
ctrl_combos = list(
itertools.combinations(
grouped_ctrls.index, 2)
)
ctrl_idx = pd.MultiIndex.from_tuples(ctrl_combos)
ctrl_series = pd.Series(group_id, index=ctrl_idx)
ctrl_list.append(ctrl_series)

if len(ctrl_list) >= 1:
ctrl_series = pd.concat(ctrl_list)
ctrl_series.index.names = ['A', 'B']

else:
ctrl_series = pd.Series()

ctrl_series.name = 'group'

else:
ref_idx = list(unique_references)
if used_controls is not None:
ctrl_series = used_controls
ctrl_series.index.name = 'id'

try:
nominal_df = diversity_measure[ref_idx].to_frame().reset_index()
except KeyError:
raise KeyError('Pairwise comparisons were unsuccessful. Please double'
' check that your chosen reference column contains'
' values that are also present in the ID column for'
' the associated metadata.')

nominal_df['group'] = 'reference'

if used_controls is not None:
ctrl_group = diversity_measure[ctrl_series.index].to_frame()
ctrl_group['group'] = ctrl_series
ctrl_group = ctrl_group.reset_index()
nominal_df = pd.concat([nominal_df, ctrl_group])
nominal_df = nominal_df.reset_index(drop=True)

if 'A' in nominal_df.columns:
if not nominal_df.empty:
nominal_df['id'] = nominal_df[['A', 'B']].agg('..'.join, axis=1)
else:
nominal_df['id'] = []
nominal_df = nominal_df[['id', 'measure', 'group', 'A', 'B']]

return nominal_df


# Helper Function For to-baseline datafiltering
def _get_to_baseline_ref(time_col, baseline_timepoint, time_column,
subject_column, metadata):

temp_baseline_ref = []
reference_list = []
baseline_ref_df = pd.DataFrame()
# All valid FMT samples have to have a time column
metadata = metadata.to_dataframe()[~time_col.isna()]
if float(baseline_timepoint) not in metadata[time_column].values:
raise AssertionError('The provided baseline timepoint'
f' {baseline_timepoint} was not'
f' found in `metadata` '
f' column {time_column}.')
for sub, samples in metadata.groupby([subject_column]):
reference = \
samples[samples[
time_column] == float(baseline_timepoint)].index.to_list()
if len(reference) > 1:
raise ValueError('More than one baseline sample was found per'
' subject. Only one baseline sample can be'
' used as a reference. Please group baseline'
' replicates.')
elif len(reference) == 0:
# If there is no baseline for a subject,
# This will either drop with filter-missing-references or
# or error and say that they need to pass
# filter-missing-references
reference = [np.nan]
temp_baseline_ref = temp_baseline_ref + samples.index.to_list()
reference_list = \
reference_list + (reference * len(samples.index.to_list()))
# I dont see any way that this hits because of my above assertion but
# I think its a good check so I am leavig it in.
if len(reference_list) == 0:
raise AssertionError('No baseline samples'
' were found in the metadata.'
' Please confirm that a valid'
' baseline timepoint was given.')
if pd.Series(reference_list).isnull().all():
raise AssertionError('No baseline samples'
' were connected via subject.'
' Confirm that all subjects have a'
' baseline timepoint')
baseline_ref_df['sample_name'] = temp_baseline_ref
baseline_ref_df['relevant_baseline'] = reference_list
baseline_ref_df = \
baseline_ref_df[~baseline_ref_df['sample_name'].isin(
reference_list)].set_index('sample_name')
reference_col = _get_series_from_col(
md=qiime2.Metadata(baseline_ref_df), col_name='relevant_baseline',
param_name='reference_column',
expected_type=qiime2.CategoricalMetadataColumn)
# this is so the variables for distance to donor and distance to
# baseline have the same variable name
used_references = reference_col
return used_references
27 changes: 19 additions & 8 deletions q2_fmt/_examples.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -11,6 +11,10 @@

import qiime2

from qiime2.plugin.util import transform

from q2_stats.types import TabularDataResourceDirFmt


def _get_data_from_tests(path):
return pkg_resources.resource_filename('q2_fmt.tests',
Expand Down Expand Up @@ -62,9 +66,13 @@ def peds_md_factory():


def peds_dist_factory():
import pandas as pd
df = transform(_get_data_from_tests('peds_dist'),
from_type=TabularDataResourceDirFmt,
to_type=pd.DataFrame)
return qiime2.Artifact.import_data(
"Dist1D[Ordered, Matched] % Properties('peds')",
_get_data_from_tests('peds_dist')
df
)


Expand Down Expand Up @@ -210,20 +218,19 @@ def peds_pipeline_sample(use):
subject_column='SubjectID'
),
use.UsageOutputNames(
visualization='heatmap',
heatmap='heatmap',

)
)

peds_heatmap.assert_output_type('Visualization')


# Heatmap pipeline
def peds_heatmap(use):
def heatmap(use):
peds_dist = use.init_artifact('peds_dist', peds_dist_factory)

peds_heatmap, = use.action(
use.UsageAction('fmt', 'peds_heatmap'),
use.UsageAction('fmt', 'heatmap'),
use.UsageInputs(
data=peds_dist,
),
Expand All @@ -240,16 +247,19 @@ def simulation_peds_method(use):
md = use.init_metadata('md', peds_md_factory)
table = use.init_artifact('table', feature_table_factory)

peds_stats, global_stats = use.action(
actual_sample_peds, peds_stats, global_stats = use.action(
use.UsageAction('fmt', 'peds_simulation'),
use.UsageInputs(
table=table,
metadata=md,
time_column='time_point',
reference_column='Donor',
subject_column='SubjectID'
subject_column='SubjectID',
num_resamples=99,
sampling_depth=400
),
use.UsageOutputNames(
actual_sample_peds='actual_sample_peds',
per_subject_stats='per_subject_stats',
global_stats='global_stats'
)
Expand All @@ -271,7 +281,8 @@ def pprs_method(use):
metadata=md,
time_column='time_point',
baseline_timepoint='1',
subject_column='SubjectID'
subject_column='SubjectID',
filter_missing_references=False
),
use.UsageOutputNames(
pprs_dists='pprs_dist'
Expand Down
Loading

0 comments on commit ed022ec

Please sign in to comment.