Skip to content

Commit

Permalink
overhauling to visualise the correct merge outputs and inputs
Browse files Browse the repository at this point in the history
  • Loading branch information
@vreuter
vreuter committed Oct 20, 2024
1 parent e246ab6 commit d284ff3
Show file tree
Hide file tree
Showing 8 changed files with 314 additions and 119 deletions.
8 changes: 4 additions & 4 deletions looptrace_regionals_vis/examples/P0001_rois.merge_contributors.csv
View file
Original file line number Diff line number Diff line change
@@ -1,9 +1,9 @@
index,fieldOfView,timepoint,spotChannel,zc,yc,xc,intensityMean,zMin,zMax,yMin,yMax,xMin,xMax,mergeIndex
index,fieldOfView,timepoint,spotChannel,zc,yc,xc,intensityMean,zMin,zMax,yMin,yMax,xMin,xMax,mergeIndices
0,P0001.zarr,79,0,3.907628987532479,231.9874778925304,871.9833511648726,118.26726920593931,-2.092371012467521,9.90762898753248,219.9874778925304,243.9874778925304,859.9833511648726,883.9833511648726,101
1,P0001.zarr,79,0,17.994259347453493,24.042015416774795,1360.0069098862991,117.1394688491732,11.994259347453491,23.994259347453493,12.042015416774795,36.0420154167748,1348.0069098862991,1372.0069098862991,102
1,P0001.zarr,79,0,17.994259347453493,24.042015416774795,1360.0069098862991,117.1394688491732,11.994259347453491,23.994259347453493,12.042015416774795,36.0420154167748,1348.0069098862991,1372.0069098862991,102;103
2,P0001.zarr,79,0,23.00910242218976,231.98008711401275,871.9596645390719,116.14075915047448,17.009102422189756,29.00910242218976,219.98008711401275,243.98008711401275,859.9596645390719,883.9596645390719,101
3,P0001.zarr,79,0,16.527137137619988,422.5165732477932,667.2129969610728,157.0530303030303,10.527137137619988,22.527137137619988,410.5165732477932,434.5165732477932,655.2129969610728,679.2129969610728,102
4,P0001.zarr,79,0,16.950424004488077,118.88259896330818,349.6540530977019,161.81065410400436,10.950424004488076,22.95042400448808,106.88259896330818,130.88259896330817,337.6540530977019,361.6540530977019,102
3,P0001.zarr,79,0,16.527137137619988,422.5165732477932,667.2129969610728,157.0530303030303,10.527137137619988,22.527137137619988,410.5165732477932,434.5165732477932,655.2129969610728,679.2129969610728,102;103
4,P0001.zarr,79,0,16.950424004488077,118.88259896330818,349.6540530977019,161.81065410400436,10.950424004488076,22.95042400448808,106.88259896330818,130.88259896330817,337.6540530977019,361.6540530977019,102;103
5,P0001.zarr,79,0,18.177805530982404,445.4564669785048,607.9657421380375,160.33961681087763,12.177805530982404,24.177805530982404,433.4564669785048,457.4564669785048,595.9657421380375,619.9657421380375,100
6,P0001.zarr,79,0,17.83959674876146,1006.0753359579252,306.5263466292306,160.1254275940707,11.839596748761458,23.83959674876146,994.0753359579252,1018.0753359579252,294.5263466292306,318.5263466292306,100
7,P0001.zarr,79,0,17.70877472362621,1040.482813665982,290.6567022086824,163.12094117647058,11.70877472362621,23.70877472362621,1028.482813665982,1052.482813665982,278.6567022086824,302.6567022086824,100
Expand Down
2 changes: 1 addition & 1 deletion looptrace_regionals_vis/point.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@
import numpy as np
from numpydoc_decorator import doc # type: ignore[import-untyped]

FloatLike = float | np.float64
from .types import FloatLike


@doc(
Expand Down
205 changes: 131 additions & 74 deletions looptrace_regionals_vis/reader.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -8,27 +8,29 @@
from pathlib import Path
from typing import Literal, Optional

import numpy as np
import pandas as pd
from numpydoc_decorator import doc # type: ignore[import-untyped]

from .bounding_box import BoundingBox3D
from .colors import IBM_BLUE, IBM_ORANGE, IBM_PINK, IBM_PURPLE, IBM_YELLOW
from .point import FloatLike, Point3D
from .types import LayerParams, PathOrPaths
from .roi import MergeContributorRoi, MergedRoi, NonNuclearRoi, ProximityRejectedRoi, SingletonRoi
from .types import Channel, LayerParams, NucleusNumber, PathOrPaths, RoiIndex, Timepoint

# Aliases
FullDataLayer = tuple[list[list[list[int | FloatLike]]], LayerParams, Literal["shapes"]]
IdAndContributors = tuple[RoiIndex, list[RoiIndex]]
Reader = Callable[[PathOrPaths], list[FullDataLayer]]


SHAPE_PARAMS_KEY = "shape_type"
COLOR_PARAMS_KEY = "edge_color"
# Constants
Z_COLUMN = "zc"
Y_COLUMN = "yc"
X_COLUMN = "xc"
BOX_CENTER_COLUMN_NAMES = [Z_COLUMN, Y_COLUMN, X_COLUMN]
TIME_COLUMN = "timepoint"
CHANNEL_COLUMN = "spotChannel"
COLOR_PARAMS_KEY = "edge_color"
SHAPE_PARAMS_KEY = "shape_type"
TEXT_SIZE = 5
TIME_COLUMN = "timepoint"


class InputFileContentType(Enum):
Expand Down Expand Up @@ -59,21 +61,6 @@ def from_filepath(cls, fp: Path) -> Optional["InputFileContentType"]:
return cls.from_filename(fp.name)


class RoiType(Enum):
"""The type of ROI to display, and in which color"""

MergeContributor = IBM_BLUE
DiscardForProximity = IBM_PURPLE
DiscardForNonNuclearity = IBM_ORANGE
AcceptedSingleton = IBM_PINK
AcceptedMerger = IBM_YELLOW

@property
def color(self) -> str:
"""More reader-friendly alias for accessing the color associated with the ROI type"""
return self.value


@doc(
summary="The main interface for the napari plugin reader contribution",
parameters=dict(path="Path to file with data to visualise"),
Expand Down Expand Up @@ -124,44 +111,75 @@ def build_layers(folder) -> list[FullDataLayer]: # type: ignore[no-untyped-def]

for file_type, file_path in file_by_kind.items():
logging.debug("Processing data for file type %s: %s", file_type.name, file_path)
time_channel_location_trios: list[
tuple[int | FloatLike, int | FloatLike, BoundingBox3D]
]
if file_type == InputFileContentType.NucleiLabeled:
time_channel_location_trios_by_roi_type = _parse_non_contributor_non_proximal_rois(
file_path
)
if file_type == InputFileContentType.MergeContributors:
rois_by_type = {
MergeContributorRoi: [
_parse_merge_contributor_record(row)
for _, row in pd.read_csv(file_path).iterrows()
]
}
elif file_type == InputFileContentType.NucleiLabeled:
rois_by_type = {}
for roi in _parse_non_contributor_non_proximal_rois(file_path):
rois_by_type.setdefault(type(roi), []).append(roi)
elif file_type == InputFileContentType.ProximityRejects:
rois_by_type = {
ProximityRejectedRoi: [
ProximityRejectedRoi(timepoint=t, channel=c, bounding_box=b)
for t, c, b in parse_boxes(file_path)
]
}
else:
time_channel_location_trios = parse_boxes(file_path)
if file_type == InputFileContentType.MergeContributors:
rt = RoiType.MergeContributor
elif file_type == InputFileContentType.ProximityRejects:
rt = RoiType.DiscardForProximity
else:
raise RuntimeError(
f"Unexpected file type (can't determine ROI type)! {file_type}"
)
time_channel_location_trios_by_roi_type = {rt: time_channel_location_trios}

for (
roi_type,
time_channel_location_trios,
) in time_channel_location_trios_by_roi_type.items():
raise RuntimeError(f"Unexpected file type (can't determine ROI type)! {file_type}")

for roi_type, rois in rois_by_type.items():
get_text_color = lambda: rois[0].color # noqa: B023
corners: list[list[list[int | FloatLike]]] = []
shapes: list[str] = []
for timepoint, channel, box in time_channel_location_trios:
for q1, q2, q3, q4, is_center_slice in box.iter_z_slices_nonnegative():
for roi in rois:
for (
q1,
q2,
q3,
q4,
is_center_slice,
) in roi.bounding_box.iter_z_slices_nonnegative():
corners.append(
[[timepoint, channel, *_point_to_list(pt)] for pt in [q1, q2, q3, q4]]
[
[roi.timepoint, roi.channel, *_point_to_list(pt)]
for pt in [q1, q2, q3, q4]
]
)
shapes.append("rectangle" if is_center_slice else "ellipse")
logging.debug("Point count for ROI type %s: %d", roi_type.name, len(corners))
logging.debug("Point count for ROI type %s: %d", roi.typename, len(shapes))
params: dict[str, object] = {
"name": roi_type.name,
"name": roi.typename,
"shape_type": shapes,
"face_color": "transparent",
COLOR_PARAMS_KEY: roi_type.color,
COLOR_PARAMS_KEY: roi.color,
}
if roi_type in [MergeContributorRoi, MergedRoi]:
if roi_type == MergeContributorRoi:
features = {
"index": [roi.index for roi in rois],
"merge_indices": [roi.merge_indices for roi in rois],
}
text = {
"string": "{index} --> {';'.join(sorted(merge_indices))}",
"size": TEXT_SIZE,
"color": get_text_color(),
}
elif roi_type == MergedRoi:
features = {
"index": [roi.index for roi in rois],
"contributors": [roi.contributors for roi in rois],
}
text = {
"string": "{index} <-- {contributors}",
"size": TEXT_SIZE,
"color": get_text_color(),
}
params.update({"features": features, "text": text})
layers.append((corners, params, "shapes"))

return layers
Expand All @@ -180,16 +198,34 @@ def _is_plausible_input_file(path: Path) -> bool:

def _parse_non_contributor_non_proximal_rois(
path: Path,
) -> dict[RoiType, list[tuple[int | FloatLike, int | FloatLike, BoundingBox3D]]]:
rois_by_type: dict[RoiType, list[tuple[int | FloatLike, int | FloatLike, BoundingBox3D]]] = {}
) -> list[NonNuclearRoi | SingletonRoi | MergedRoi]:
rois: list[NonNuclearRoi | SingletonRoi | MergedRoi] = []
for _, row in pd.read_csv(path).iterrows():
time, channel, loc, is_singleton, is_in_nuc = _parse_nucleus_labeled_record(row)
if is_in_nuc:
key = RoiType.AcceptedSingleton if is_singleton else RoiType.AcceptedMerger
else:
key = RoiType.DiscardForNonNuclearity
rois_by_type.setdefault(key, []).append((time, channel, loc))
return rois_by_type
time, channel, box, maybe_nuc_num, maybe_id_and_contribs = _parse_nucleus_labeled_record(
row
)
match (maybe_nuc_num, maybe_id_and_contribs):
case (None, _):
roi = NonNuclearRoi(timepoint=time, channel=channel, bounding_box=box)
case (nuc_num, None):
roi = SingletonRoi(
timepoint=time, channel=channel, bounding_box=box, nucleus_number=nuc_num
)
case (nuc_num, (main_id, contrib_ids)):
roi = MergedRoi(
index=main_id,
timepoint=time,
channel=channel,
bounding_box=box,
nucleus_number=nuc_num,
contributors=contrib_ids,
)
case _:
raise Exception( # noqa: TRY002
f"Could not determine how to build ROI! maybe_nuc_num={maybe_nuc_num}, maybe_id_and_contribs={maybe_id_and_contribs}"
)
rois.append(roi)
return rois


@doc(
Expand All @@ -199,21 +235,53 @@ def _parse_non_contributor_non_proximal_rois(
)
def parse_boxes( # noqa: D103
path: Path,
) -> list[tuple[int | FloatLike, int | FloatLike, BoundingBox3D]]:
) -> list[tuple[Timepoint, Channel, BoundingBox3D]]:
box_cols = [f.name for f in dataclasses.fields(BoundingBox3D) if f.name != "center"]
spot_data = pd.read_csv(
path, usecols=BOX_CENTER_COLUMN_NAMES + box_cols + [TIME_COLUMN, CHANNEL_COLUMN]
)
time_channel_location_trios: list[tuple[int | FloatLike, int | FloatLike, BoundingBox3D]] = [
time_channel_location_trios: list[tuple[Timepoint, Channel, BoundingBox3D]] = [
_parse_time_channel_box_trio(record) for _, record in spot_data.iterrows()
]
return time_channel_location_trios


def _parse_merge_contributor_record(
record: pd.Series, # type: ignore[type-arg]
) -> MergeContributorRoi:
record: dict[str, int | FloatLike] = record.to_dict() # type: ignore[no-redef]
index: RoiIndex = record["index"]
merge_indices: set[RoiIndex] = {int(i) for i in record["mergeIndices"].split(";")}
time, channel, box = _parse_time_channel_box_trio(record)
return MergeContributorRoi(
index=index, timepoint=time, channel=channel, bounding_box=box, merge_indices=merge_indices
)


def _parse_nucleus_labeled_record(
record: pd.Series, # type: ignore[type-arg]
) -> tuple[Timepoint, Channel, BoundingBox3D, Optional[NucleusNumber], Optional[IdAndContributors]]:
record: dict[str, int | FloatLike] = record.to_dict() # type: ignore[no-redef]
raw_nuc_num: int = record["nucleusNumber"]
maybe_nuc_num: Optional[NucleusNumber] = (
None if raw_nuc_num == 0 else NucleusNumber(raw_nuc_num)
)
raw_merge_rois: object = record["mergeRois"]
id_and_contribs: Optional[IdAndContributors]
if raw_merge_rois is None or pd.isna(raw_merge_rois):
id_and_contribs = None
else:
roi_id: RoiIndex = record["index"]
contribs = [int(i) for i in raw_merge_rois.split(";")]
id_and_contribs = (roi_id, contribs)
time, channel, box = _parse_time_channel_box_trio(record)
return time, channel, box, maybe_nuc_num, id_and_contribs


def _parse_time_channel_box_trio(
record: dict[str, int | FloatLike] | pd.Series, # type: ignore[type-arg]
) -> tuple[int | FloatLike, int | FloatLike, BoundingBox3D]:
record: dict[str, int | float | np.float64] = ( # type: ignore[no-redef]
) -> tuple[Timepoint, Channel, BoundingBox3D]:
record: dict[str, int | FloatLike] = ( # type: ignore[no-redef]
record if isinstance(record, dict) else record.to_dict()
)
time = record.pop(TIME_COLUMN)
Expand All @@ -230,17 +298,6 @@ def _parse_time_channel_box_trio(
return time, channel, box


def _parse_nucleus_labeled_record(
record: pd.Series, # type: ignore[type-arg]
) -> tuple[int | FloatLike, int | FloatLike, BoundingBox3D, bool, bool]:
record: dict[str, int | float | np.float64] = record.to_dict() # type: ignore[no-redef]
nuc_num: int = record["nucleusNumber"]
in_nuc: bool = nuc_num != 0
is_singleton: bool = record["mergeRois"] is None or pd.isna(record["mergeRois"])
time, channel, box = _parse_time_channel_box_trio(record)
return time, channel, box, is_singleton, in_nuc


@doc(
summary="Flatten point coordinates to list",
parameters=dict(pt="Point to flatten"),
Expand Down
Loading

0 comments on commit d284ff3

Please sign in to comment.