Data Frame

Description

Data Frame is a two dimensional tabular data. This data structure contains labeled axes (rows and columns).
The corresponding class in gnomon is gnomonDataFrameDataPandas.

Default reader plugin

The reader of data frame form in gnomon is gnomonDataFrameReaderPandas. It loads a CSV file with named columns.
This reader supports one extension csv.

Default writer plugin

The default writer of point cloud form is gnomonDataFrameWriterPandas

Gnomon Data Frame Example

Plugins which produce Image as output

Here is a non exhaustive list of some algorithms which produce data frame as output.

• cellFeaturesTissueImage

• signalQuantificationImageSignal

• surfaceCellCurvature

A Use Case, Signal Quantification

from copy import deepcopy

import numpy as np
import pandas as pd

from dtkcore import d_inliststringlist
from dtkcore import d_real

from gnomon.utils import algorithmPlugin
from gnomon.utils.decorators import dataFrameOutput
from gnomon.utils.decorators import imageInput
from gnomon.utils.decorators import pointCloudInput
from gnomon.utils.decorators import pointCloudOutput

from gnomon.core import gnomonAbstractPointCloudQuantification

from timagetk.algorithms.signal_quantification import quantify_nuclei_signal_intensity


@algorithmPlugin(version="0.3.1", coreversion="0.81.1")
@imageInput('img', data_plugin='gnomonImageDataMultiChannelImage')
@pointCloudInput("df", data_plugin="gnomonPointCloudDataPandas")
@pointCloudOutput('out_df', data_plugin="gnomonPointCloudDataPandas")
@dataFrameOutput('data_df', data_plugin="gnomonDataFrameDataPandas")
class nucleiSignalQuantificationTimagetk(gnomonAbstractPointCloudQuantification):
    """Quantify image signal intensities on a 3D nuclei point cloud.
   """

    def __init__(self):
        super().__init__()

        self.img = {}
        self.df = {}
        self.out_df = {}
        self.data_df = {}

        self._parameters = {}
        self._parameters['signal_channels'] = d_inliststringlist("Channels", [""], [""], "Channels on which to compute the signal intensity")
        self._parameters['gaussian_sigma'] = d_real("Sigma", 0.5, 0., 50., 2, "Standard deviation of the Gaussian kernel used to smooth signal")


    def run(self):
        self.data_df = {}
        self.out_df = {}

        for time in self.df.keys():
            df = self.df[time]
            out_df = deepcopy(df)

            nuclei_points = df[['center_'+dim for dim in 'xyz']].values

            assert (time in self.img) and (self.img[time] is not None)

            img = self.img[time]

            for signal_name in self['signal_channels']:
                signal_img = img[signal_name]
                out_df[signal_name] = quantify_nuclei_signal_intensity(signal_img, nuclei_points, nuclei_sigma=self['gaussian_sigma'])

            self.out_df[time] = out_df
            
            self.data_df[time] = deepcopy(out_df)
            self.data_df[time]['time'] = time

        if len(self.data_df)>0:
            self.data_df = {0:pd.concat([self.data_df[time] for time in self.data_df.keys()])}