From fb7a49e85467dd406ad96bd962ceaafd69243da8 Mon Sep 17 00:00:00 2001
From: pswain <peter.swain@ed.ac.uk>
Date: Thu, 2 Nov 2023 18:01:22 +0000
Subject: [PATCH] feat(babysitter): removed pixel_size; docs(bud_metric)
---
src/agora/utils/lineage.py | 17 ----
src/aliby/baby_client.py | 2 -
src/aliby/baby_sitter.py | 5 +-
src/aliby/pipeline.py | 2 +-
src/postprocessor/core/lineageprocess.py | 5 +-
.../core/reshapers/bud_metric.py | 98 +++++++++----------
src/postprocessor/core/reshapers/picker.py | 12 +--
7 files changed, 54 insertions(+), 87 deletions(-)
delete mode 100644 src/agora/utils/lineage.py
diff --git a/src/agora/utils/lineage.py b/src/agora/utils/lineage.py
deleted file mode 100644
index 05c161a5..00000000
--- a/src/agora/utils/lineage.py
+++ /dev/null
@@ -1,17 +0,0 @@
-#!/usr/bin/env python3
-
-import numpy as np
-
-from agora.io.bridge import groupsort
-
-
-def mb_array_to_dict(mb_array: np.ndarray):
- """
- Convert a lineage ndarray (trap, mother_id, daughter_id)
- into a dictionary of lists ( mother_id ->[daughters_ids] )
- """
- return {
- (trap, mo): [(trap, d[0]) for d in daughters]
- for trap, mo_da in groupsort(mb_array).items()
- for mo, daughters in groupsort(mo_da).items()
- }
diff --git a/src/aliby/baby_client.py b/src/aliby/baby_client.py
index 1ced0c76..a94fe6ff 100644
--- a/src/aliby/baby_client.py
+++ b/src/aliby/baby_client.py
@@ -204,7 +204,6 @@ def choose_model_from_params(
"""
# cameras prime95 has become sCMOS and evolve has EMCCD
valid_models = list(modelsets().keys())
- breakpoint()
# Apply modelset filter if specified
if modelset_filter is not None:
@@ -218,7 +217,6 @@ def choose_model_from_params(
]
params_re = re.compile("^" + "_".join(params) + "$")
valid_models = list(filter(params_re.search, valid_models))
- breakpoint()
# Check that there are valid models
if len(valid_models) == 0:
raise KeyError(
diff --git a/src/aliby/baby_sitter.py b/src/aliby/baby_sitter.py
index 386e312b..2bd930ba 100644
--- a/src/aliby/baby_sitter.py
+++ b/src/aliby/baby_sitter.py
@@ -15,15 +15,14 @@ class BabyParameters(ParametersABC):
def __init__(
self,
modelset_name,
- pixel_size,
clogging_thresh,
min_bud_tps,
isbud_thresh,
session,
):
"""Initialise parameters for BABY."""
+ # pixel_size is specified in BABY's model sets
self.modelset_name = modelset_name
- self.pixel_size = pixel_size
self.clogging_thresh = clogging_thresh
self.min_bud_tps = min_bud_tps
self.isbud_thresh = isbud_thresh
@@ -34,7 +33,6 @@ class BabyParameters(ParametersABC):
"""Define default parameters; kwargs choose BABY model set."""
return cls(
modelset_name=get_modelset_name_from_params(**kwargs),
- pixel_size=0.182,
clogging_thresh=0.75,
min_bud_tps=3,
isbud_thresh=0.5,
@@ -84,7 +82,6 @@ class BabyRunner(StepABC):
else:
brain = modelsets.get(
modelset_name,
- pixel_size=parameters.pixel_size,
clogging_thresh=parameters.clogging_thresh,
min_bud_tps=parameters.min_bud_tps,
isbud_thresh=parameters.isbud_thresh,
diff --git a/src/aliby/pipeline.py b/src/aliby/pipeline.py
index aa74fdfd..46b786c7 100644
--- a/src/aliby/pipeline.py
+++ b/src/aliby/pipeline.py
@@ -476,7 +476,7 @@ class Pipeline(ProcessABC):
result = pipe["steps"][step].run_tp(
i, **run_kwargs.get(step, {})
)
- # write to h5 file using writers
+ # write result to h5 file using writers
# extractor writes to h5 itself
if step in loaded_writers:
loaded_writers[step].write(
diff --git a/src/postprocessor/core/lineageprocess.py b/src/postprocessor/core/lineageprocess.py
index c359df89..90378a47 100644
--- a/src/postprocessor/core/lineageprocess.py
+++ b/src/postprocessor/core/lineageprocess.py
@@ -7,7 +7,6 @@ import numpy as np
import pandas as pd
from agora.abc import ParametersABC
-from agora.utils.kymograph import get_index_as_np
from postprocessor.core.abc import PostProcessABC
@@ -57,10 +56,10 @@ class LineageProcess(PostProcessABC):
)
def get_lineage_information(self, signal=None, merged=True):
- """Get lineage as an array with tile IDs, mother labels, and corresponding bud labels."""
+ """Get lineage as an array with tile IDs, mother and bud labels."""
if signal is not None and "mother_label" in signal.index.names:
# from kymograph
- lineage = get_index_as_np(signal)
+ lineage = np.array(signal.index.to_list())
elif hasattr(self, "lineage"):
lineage = self.lineage
elif hasattr(self, "cells"):
diff --git a/src/postprocessor/core/reshapers/bud_metric.py b/src/postprocessor/core/reshapers/bud_metric.py
index 57c01069..eee5c0b4 100644
--- a/src/postprocessor/core/reshapers/bud_metric.py
+++ b/src/postprocessor/core/reshapers/bud_metric.py
@@ -3,11 +3,26 @@ import typing as t
import numpy as np
import pandas as pd
-from agora.utils.lineage import mb_array_to_dict
+from agora.io.bridge import groupsort
from postprocessor.core.lineageprocess import (
LineageProcess,
LineageProcessParameters,
)
+import logging
+
+
+def mother_bud_array_to_dict(mb_array: np.ndarray):
+ """
+ Convert a lineage into a dict of lists.
+
+ A lineage is an array (trap, mother_id, daughter_id) and
+ becomes a dictionary of lists (mother_id->[daughters_ids])
+ """
+ return {
+ (trap, mo): [(trap, d[0]) for d in daughters]
+ for trap, mo_da in groupsort(mb_array).items()
+ for mo, daughters in groupsort(mo_da).items()
+ }
class BudMetricParameters(LineageProcessParameters):
@@ -18,12 +33,13 @@ class BudMetricParameters(LineageProcessParameters):
class BudMetric(LineageProcess):
"""
- Requires mother-bud information to create a new dataframe where the
- indices are mother ids and values are the daughters' values for a
- given signal.
+ Create a dataframe with indices mother IDs and values from buds.
+
+ Requires mother-bud information.
"""
def __init__(self, parameters: BudMetricParameters):
+ """Initialise using LineageProcess."""
super().__init__(parameters)
def run(
@@ -31,23 +47,27 @@ class BudMetric(LineageProcess):
signal: pd.DataFrame,
lineage: t.Dict[pd.Index, t.Tuple[pd.Index]] = None,
):
+ """Calculate a metric for all buds."""
if lineage is None:
# define lineage
if hasattr(self, "lineage"):
lineage = self.lineage
else:
- # lineage information in the Signal dataframe
+ # lineage information in the Signal data frame
assert "mother_label" in signal.index.names
lineage = signal.index.to_list()
- return self.get_bud_metric(signal, mb_array_to_dict(lineage))
+ return self.get_bud_metric(signal, mother_bud_array_to_dict(lineage))
@staticmethod
def get_bud_metric(
- signal: pd.DataFrame, md: t.Dict[t.Tuple, t.Tuple[t.Tuple]] = None
+ signal: pd.DataFrame,
+ lineage_dict: t.Dict[t.Tuple, t.Tuple[t.Tuple]] = None,
):
"""
- Generate a dataframe of a Signal for buds indexed by their mothers,
- concatenating data from all the buds for each mother.
+ Generate a dataframe of a Signal for buds.
+
+ The data frame is indexed by the buds' mothers and concatenates
+ data from all the buds for each mother.
Parameters
---------
@@ -62,7 +82,11 @@ class BudMetric(LineageProcess):
# md_index should only comprise (trap, cell_label)
if "mother_label" not in md_index.names:
# dict with daughter indices as keys and mother indices as values
- bud_dict = {v: k for k, values in md.items() for v in values}
+ bud_dict = {
+ bud: mother
+ for mother, buds in lineage_dict.items()
+ for bud in buds
+ }
# generate mother_label in Signal using the mother's cell_label
# cells with no mothers have a mother_label of 0
signal["mother_label"] = list(
@@ -70,17 +94,21 @@ class BudMetric(LineageProcess):
)
signal.set_index("mother_label", append=True, inplace=True)
# combine mothers and daughter indices
- mothers_index = md.keys()
- daughters_index = [y for x in md.values() for y in x]
+ mothers_index = lineage_dict.keys()
+ daughters_index = [
+ bud for buds in lineage_dict.values() for bud in buds
+ ]
relations = set([*mothers_index, *daughters_index])
- # keep from md_index only cells that are mother or daughters
+ # keep only cells that are mother or daughters
md_index = md_index.intersection(relations)
else:
md_index = md_index.droplevel("mother_label")
if len(md_index) < len(signal):
- print(
- f"Dropped {len(signal) - len(md_index)} cells before applying bud_metric"
- ) # TODO log
+ logging.getLogger("aliby").log(
+ logging.WARNING,
+ f"Dropped {len(signal) - len(md_index)} cells before "
+ "applying bud_metric.",
+ )
# restrict signal to the cells in md_index moving mother_label to do so
signal = (
signal.reset_index("mother_label")
@@ -137,41 +165,3 @@ def combine_daughter_tracks(tracks: pd.DataFrame):
bud_df.iloc[jrow].loc[init_tps[j] :].values
)
return pd.Series(combined_tracks, index=tracks.columns)
-
-
-def _combine_daughter_tracks_original(tracks: pd.DataFrame):
- """
- Combine multiple time series of daughter cells into one time series.
-
- At any one time, a mother cell should have only one daughter.
-
- Two daughters are still sometimes present at the same time point, and we
- then choose the daughter that appears first.
-
- TODO We need to fix examples with more than one daughter at a time point.
-
- Parameters
- ----------
- tracks: a Signal
- Data for all daughters, which are distinguished by different cell_labels,
- for a particular trap and mother_label.
- """
- # sort by daughter IDs
- bud_df = tracks.sort_index(level="cell_label")
- # remove multi-index
- bud_df.index = range(len(bud_df))
- # find which row of sorted_df has the daughter for each time point
- tp_fvt: pd.Series = bud_df.apply(lambda x: x.first_valid_index(), axis=0)
- # combine data for all daughters
- combined_tracks = np.nan * np.ones(tracks.columns.size)
- for bud_row in np.unique(tp_fvt.dropna().values).astype(int):
- ilocs = np.where(tp_fvt.values == bud_row)[0]
- combined_tracks[ilocs] = bud_df.values[bud_row, ilocs]
- # TODO delete old version
- tp_fvt = bud_df.columns.get_indexer(tp_fvt)
- tp_fvt[tp_fvt == -1] = len(bud_df) - 1
- old = np.choose(tp_fvt, bud_df.values)
- assert (
- (combined_tracks == old) | (np.isnan(combined_tracks) & np.isnan(old))
- ).all(), "yikes"
- return pd.Series(combined_tracks, index=tracks.columns)
diff --git a/src/postprocessor/core/reshapers/picker.py b/src/postprocessor/core/reshapers/picker.py
index 14d76446..e283386e 100644
--- a/src/postprocessor/core/reshapers/picker.py
+++ b/src/postprocessor/core/reshapers/picker.py
@@ -24,9 +24,9 @@ class PickerParameters(ParametersABC):
"""
_defaults = {
- "sequence": [
+ "picker_sequence": [
["lineage", "families"],
- ["condition", "present", 7],
+ ["condition", "present", 3],
],
}
@@ -80,9 +80,9 @@ class Picker(LineageProcess):
if len(lineage):
self.mothers = lineage[:, [0, 1]]
self.daughters = lineage[:, [0, 2]]
- for alg, *params in self.sequence:
+ for method, *params in self.picker_sequence:
if indices:
- if alg == "lineage":
+ if method == "lineage":
# pick by lineage
param1 = params[0]
new_indices = self.pick_by_lineage(
@@ -113,7 +113,7 @@ class Picker(LineageProcess):
):
"""Pick indices from signal by any_present, present, and growing."""
if len(threshold) == 1:
- threshold = [_as_int(*threshold, signal.shape[1])]
+ threshold = [as_int(*threshold, signal.shape[1])]
#: is this correct for "growing"?
case_mgr = {
"any_present": lambda s, threshold: any_present(s, threshold),
@@ -127,7 +127,7 @@ class Picker(LineageProcess):
return new_indices
-def _as_int(threshold: t.Union[float, int], ntps: int):
+def as_int(threshold: t.Union[float, int], ntps: int):
"""Convert a fraction of the total experiment duration into a number of time points."""
if type(threshold) is float:
threshold = ntps * threshold
--
GitLab