From 89a62047f85cac63894e9d761b7e203ea080e66a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Al=C3=A1n=20Mu=C3=B1oz?= <amuoz@ed.ac.uk>
Date: Wed, 16 Jun 2021 19:56:14 +0100
Subject: [PATCH] bugfixes everywhere
Former-commit-id: 9642c99b5623ba9b46bcba2cbff45afbc26450f2
---
core/functions/test_hdf.py | 12 ++++++++----
core/functions/tracks.py | 7 +++++--
core/io/signal.py | 4 +++-
core/processes/merger.py | 24 ++++++++++++++++++++----
core/processes/picker.py | 33 +++++++++++++++++----------------
core/processor.py | 23 ++++++++++++++++-------
6 files changed, 69 insertions(+), 34 deletions(-)
diff --git a/core/functions/test_hdf.py b/core/functions/test_hdf.py
index 683c005b..5bd3b1a1 100644
--- a/core/functions/test_hdf.py
+++ b/core/functions/test_hdf.py
@@ -5,11 +5,15 @@ from core.cells import Cells
import pandas as pd
# f = h5py.File("/home/alan/Documents/sync_docs/PhD/tmp/DO6MS2_003store.h5")
-f = h5py.File(
- "/shared_libs/pipeline-core/scripts/data/20191026_ss_experiments_01/DO6MS2_003store.h5"
-)
+fname = "/shared_libs/pipeline-core/scripts/data/20191026_ss_experiments_01/DO6MS2_003store.h5"
+f = h5py.File(fname)
tracks = f["/extraction/general/None/area"][()]
-cell = Cells.from_source("/home/alan/Documents/sync_docs/PhD/tmp/DO6MS2_003store.h5")
+cell = Cells.from_source(fname)
from postprocessor.core.processes.picker import Picker, PickerParameters
picker = Picker(cells=cell, parameters=PickerParameters.default())
+
+from postprocessor.core.processor import PostProcessor, PostProParameters
+
+pp = PostProcessor(filename=fname, parameters=PostProParameters.default())
+pp.run()
diff --git a/core/functions/tracks.py b/core/functions/tracks.py
index ecc4e0f5..2895b676 100644
--- a/core/functions/tracks.py
+++ b/core/functions/tracks.py
@@ -47,7 +47,7 @@ def max_nonstop_ntps(track: pd.Series) -> int:
def get_tracks_ntps(tracks: pd.DataFrame) -> pd.Series:
- return tracks.apply(get_ntps, axis=1)
+ return tracks.apply(max_ntps, axis=1)
def get_avg_gr(track: pd.Series) -> int:
@@ -56,7 +56,7 @@ def get_avg_gr(track: pd.Series) -> int:
:param tracks: Series with volume and timepoints as indices
"""
- ntps = get_ntps(track)
+ ntps = max_ntps(track)
vals = track.dropna().values
gr = (vals[-1] - vals[0]) / ntps
return gr
@@ -138,6 +138,9 @@ def get_joint_ids(merging_seqs) -> dict:
output {a:a, b:a, c:a, d:a}
"""
+ if not merging_seqs:
+ return {}
+
targets, origins = list(zip(*merging_seqs))
static_tracks = set(targets).difference(origins)
diff --git a/core/io/signal.py b/core/io/signal.py
index 5f3e8c6d..91907bd5 100644
--- a/core/io/signal.py
+++ b/core/io/signal.py
@@ -1,3 +1,5 @@
+import pandas as pd
+
from postprocessor.core.io.base import BridgeH5
@@ -12,7 +14,7 @@ class Signal(BridgeH5):
def __getitem__(self, dataset):
dset = self._hdf[dataset][()]
attrs = self._hdf[dataset].attrs
- first_dataset = "/" + dataset.split("/")[0] + "/"
+ first_dataset = dataset.split("/")[1] + "/"
timepoints = self._hdf[first_dataset].attrs["processed_timepoints"]
if "cell_label" in self._hdf[dataset].attrs:
diff --git a/core/processes/merger.py b/core/processes/merger.py
index 5c27dd0b..51a91bf9 100644
--- a/core/processes/merger.py
+++ b/core/processes/merger.py
@@ -1,4 +1,8 @@
-class MergerParameters:
+from postprocessor.core.processes.base import ParametersABC, ProcessABC
+from postprocessor.core.functions.tracks import clean_tracks, merge_tracks, join_tracks
+
+
+class MergerParameters(ParametersABC):
"""
:param tol: float or int threshold of average (prediction error/std) necessary
to consider two tracks the same. If float is fraction of first track,
@@ -26,14 +30,26 @@ class MergerParameters:
self.min_avg_delta = min_avg_delta
+ @classmethod
+ def default(cls):
+ return cls.from_dict(
+ {
+ "smooth": False,
+ "tolerance": 0.1,
+ "window": 5,
+ "degree": 3,
+ "min_avg_delta": 0.9,
+ }
+ )
+
-class Merger:
+class Merger(ProcessABC):
"""
TODO Integrate functions/tracks.py inside this class?
"""
def __init__(self, parameters):
- self.parameters = parameters
+ super().__init__(parameters)
def run(self, signal):
- merged, joint_pairs = merge_tracks(signal, min_len=self.window + 1)
+ merged, joint_pairs = merge_tracks(signal) # , min_len=self.window + 1)
diff --git a/core/processes/picker.py b/core/processes/picker.py
index 9342ff17..367a88a1 100644
--- a/core/processes/picker.py
+++ b/core/processes/picker.py
@@ -88,7 +88,7 @@ class Picker(ProcessABC):
return signals.loc[idx]
def pick_by_condition(self, signals):
- idx = switch_case(self.condition[0], signals, self.condition[1])
+ idx = self.switch_case(self.condition[0], signals, self.condition[1])
return signals.loc[idx]
def run(self, signals):
@@ -96,22 +96,23 @@ class Picker(ProcessABC):
self.signals = getattr(self, "pick_by_" + alg)(signals)
return self.signals
+ @staticmethod
+ def switch_case(
+ condition: str,
+ signals: pd.DataFrame,
+ threshold: Union[float, int],
+ ):
+ threshold_asint = _as_int(threshold, signals.shape[1])
+ case_mgr = {
+ "present": signals.apply(max_ntps, axis=1) > threshold_asint,
+ "nonstoply_present": signals.apply(max_nonstop_ntps, axis=1)
+ > threshold_asint,
+ "quantile": [np.quantile(signals.values[signals.notna()], threshold)],
+ }
+ return case_mgr[condition]
+
-def as_int(threshold: Union[float, int], ntps: int):
+def _as_int(threshold: Union[float, int], ntps: int):
if type(threshold) is float:
threshold = threshold / ntps
return threshold
-
-
-def switch_case(
- condition: str,
- signals: pd.DataFrame,
- threshold: Union[float, int],
-):
- threshold_asint = as_int(threshold, signals.shape[1])
- case_mgr = {
- "present": signals.apply(max_ntps, axis=1) > threshold_asint,
- "nonstoply_present": signals.apply(max_nonstop_ntps, axis=1) > threshold_asint,
- "quantile": [np.quantile(signals.values[signals.notna()], threshold)],
- }
- return case_mgr[condition]
diff --git a/core/processor.py b/core/processor.py
index 9e433076..41e3c630 100644
--- a/core/processor.py
+++ b/core/processor.py
@@ -42,25 +42,34 @@ class PostProParameters(ParametersABC):
},
)
+ def to_dict(self):
+ return {k: _if_dict(v) for k, v in self.__dict__()}
+
class PostProcessor:
- def __init__(self, fname, parameters, signals):
+ def __init__(self, filename, parameters):
self.parameters = parameters
- self._signals = Signals(fname)
- self._writer = Writer(fname)
+ self._signals = Signal(filename)
+ self._writer = Writer(filename)
self.datasets = parameters["datasets"]
self.merger = Merger(parameters["merger"])
self.picker = Picker(
- parameters=parameters["picker"], cell=Cells.from_source(fname)
+ parameters=parameters["picker"], cells=Cells.from_source(filename)
)
self.processes = [
self.get_process(process) for process in parameters["processes"]
]
def run(self):
- self.merger.run(signals[self.datasets["merger"]])
- self.picker.run(signals[self.datasets["picker"]])
+ self.merger.run(self._signals[self.datasets["merger"]])
+ self.picker.run(self._signals[self.datasets["picker"]])
for process, dataset in zip(self.processes, self.datasets["processes"]):
- process_result = process.run(signals.get_dataset(dataset))
+ process_result = process.run(self._signals.get_dataset(dataset))
self.writer.write(process_result, dataset)
+
+
+def _ifdict(item):
+ if hasattr(item, "to_dict"):
+ item = item.to_dict()
+ return item
--
GitLab