search

AMP-SCZ / outcome_calculations

Repository includes the scripts to calculate outcomes (predictors) for the AMP-SCZ study.
Apache License 2.0
0 stars 0 forks source link

readme

outcome_calculations

In this repository we calculate the outcomes (predictors) for the AMP-SCZ study based on calculations provided by experts in the field.

We calculate the outcomes for the AMP-SCZ study.

Nora Penzel (March 30th, 2023)

All the outcomes are based on calculations or suggestions from experts in the field, e.g., Jean Addington, Scott Woods, Dominic Oliver etc. Outcomes should be used to predict endpoints. In this repository we have 2 main scripts and four additional csv files.

1. outcome_calculations.python

This is the main script. Here all the outcome-calculations are implemented. This script should be used in a two(three)-stage procedure. The actual commands to run the script are given below along with the explanation what happens depending on the input variables given:

  1. First, it is recommended to run a test version. Especially after code has been changed and new updates have been made within the script. 1a. To run it for Pronet: python outcome_calculations.py "pronet" "test" 1b. To run it for Prescient: python outcome_calculations.py "prescient" "test" Hereby, we create 2 new csv files. These include all the outcomes based on the script for 6 subjects per network that have been previously checked. These csv files should then be used by the second script.

  2. Run the second script "test_outcome.py". IF it displays that: "The data is the same as when we have checked it. Thus, go ahead and run the code" You can go back to outcome_calculations.py and run it with the following input variables: 3a. To run it for Pronet: python outcome_calculations.py "pronet" "run_outcome" 3b. To run it for Prescient: python outcome_calculations.py "prescient" "run_outcome" This now will create the outcomes within each subject folder. Developer mode

  3. If the function "test_outcome.py" produced that there was an error it should be further investigated what was different between the control files and the new files generated. In case code is updated and everything is checked the "control subject" should be re-generated running the code with: python outcome_calculations.py "pronet" "create_check" python outcome_calculations.py "prescient" "create_check"

  4. After the creation of all outcomes etc. we create a dictionary using the R code create_dictionary.R. This R-code loads the csv files that are created when running the outcome_calculations.py "run_outcome" version. This script loads all the subject data and outcome data and provides overall scores. It should be run everytime the outcome_calculations.py was run to have the most up-to date values.