scripts.Prediction

This script is used within the CellScanner application to predict species in coculture samples, apply gating for live/dead or debris classification, and perform heterogeneity analysis.

Key Features: - Predicts species in flow cytometry data using a trained neural network model. - Applies gating to distinguish live, inactive, and debris states based on specified thresholds. - Saves prediction and gating results, including visualizations in 3D scatter plots. - Performs heterogeneity analysis using simple and MiniBatchKMeans clustering approaches. - Generates and saves heterogeneity plots as HTML files.

Authors:

• Ermis Ioannis Michail Delopoulos

• Haris Zafeiropoulos

Date: 2024 - 2025

Classes

PredictionPanel	The prediction panel of the CellScanner GUI enabling the user to provide files and parameters for predicting
WorkerPredict	Worker class for running predict() for each co-culture file loaded, and if more than one, merge the findings

Module Contents

class scripts.Prediction.PredictionPanel(file_panel, train_panel, parent=None)

Bases: PyQt5.QtWidgets.QWidget, scripts.GUIhelpers.GatingMixin, scripts.GUIhelpers.GatingCheckBox, scripts.GUIhelpers.LiveDeadDebrisSelectors

The prediction panel of the CellScanner GUI enabling the user to provide files and parameters for predicting co-culture data.

build_stain_inputs()

Function to support user-labeled stains A stain needs to have: - a channel - a sign - a threshold - a label CellScanner will set the label as True when the threshold holds.

channels_on_stain_buttons()

choose_coculture_file()

fire_predict()

on_error(message)

prediction_completed()

start_loading_cursor()

stop_loading_cursor()

toggle_uncertainty_filterint_options()

update_uncertainty_threshold()

axis_selection_label

axis_selectors

choose_coculture_file_button

file_panel

layout

new_stain_button

predict_panel

predict_panel_layout

run_prediction_button

train_panel

uncertainty_filtering_checkbox

uncertainty_threshold

uncertainty_threshold_label

uncertainty_threshold_layout

x_axis_selector

y_axis_selector

z_axis_selector

class scripts.Prediction.WorkerPredict(PredictPanel=None)

Bases: PyQt5.QtCore.QObject

Worker class for running predict() for each co-culture file loaded, and if more than one, merge the findings

run_predict()

Main function to call the predict() function of CellScanner for each and every co-culture file provided

PredictPanel = None

error_signal

finished_signal