Master Project Readme

This repository contains the code for the Master thesis graduation project. In order to run the front and backend with the test dataset, the following software should be installed:

• Python 3.9 and pip
• Node.js (with npm) (v14.x or v16.x)
• Yarn (preferred, but npm (comes with node) works as well)

Installation backend

There are two options to install the backend, either using the requirements.txt file or the pipenv file which is also included. If known, the latter option can be used, else you can install the dependencies via the requirements.txt file.

Move into the backend directory in your terminal and run:

pip install -r requirements.txt

Installation frontend

The frontend can be installed via either Yarn or NPM, we provide both methods. Please choose one and only use the command corresponding with your choice:

• npm install
  

  or

• yarn
  

Starting the application

First, make sure no application is running on ports 3000 and 5000.

In order to run the application, start a terminal session and move into the backend directory. Then run the following commands in order:

1.  export FLASK_ENV=development
   

2.  flask run
   

The backend is now running on 127.0.0.1:5000. For consecutive starts of the backend, run the export and run commands.

Open another terminal session and move into the frontend folder. Run the command corresponding to the choice made in the installation of the frontend:

• npm run dev
  

  or

• yarn dev
  

The frontend is now running on http://localhost:3000/

Retrieving the dataset

The dataset is protected and not public and should therefore be retrieved and recreated by the user. Upon obtaining access to the dataset and more specific, the Google BigQuery instance, the raw sequences can be downloaded using the Retrieve_mimic_colab.ipynb notebook, which should be run in Google Colab. Please check the paths in the beginning of the file, as well as make sure the paths to the required csv files from the ED are correct and present. After running this script, the dataset can be created using scripts (in order):

1. dataset create.ipynb
2. Clustering.ipynb

These scripts allow for interactive creation of the dataset. Please set the number of sequences the dataset should contain using the variable number_of_stays, this defaults to 1000. For a terminal based approach, retrieve the data using the Colab script, run the dataset create.ipynb script, and then run the following scripts in order (in the script as code folder):

1. calc_distance_cython.py NUM_SEQUENCES
2. hier_clust.py NUM_SEQUENCES
3. msa.py
4. labels.py

where NUM_SEQUENCES is the number of sequences the data should be prepared for. IMPORTANT: make sure this number is also set in the variable number_of_stays in the msa and labels script. The output will be stored in the scripts/output folder, where the backend can retrieve this data. When another dataset is used than the default one, change the variable file_suffix in the app.py to _NUM_SEQUENCES, so for 1000 sequences this would be file_suffix=_1000, including the leading underscore

Creating an artificial test dataset

Another possibility is to create an artificial test set using the TestSetCreate.ipynb script in /scripts. Enter your sequences in the variable seqs_1. The default dataset is currently present in this file and variable as an example.