STarFish

STarFish is a Stacked Ensemble Target Fishing tool for small molecule target identification. STarFish takes a chemical structure as an input and provides a ranked list of potential protein targets along with a probability score for each.

Prerequisites

STarFish was built using python and anaconda was used to manage the required packages and dependencies. To get started, download and install the python3 version of miniconda:

https://docs.conda.io/en/latest/miniconda.html

Installing and Running STarFish

Download the STarFish repository and extract files to desired location. Open Anaconda Prompt (This was installed with anaconda).

Note: After extraction, make sure that opening the "STarFish-master" folder that you see folders and files named "build_model", "web_app", "environment.yml", "LICENSE", and "README.md". If you see another folder named "STarFish-master" then move that folder to your desired location and confirm that you see the aforementioned folders and files inside of that one. Otherwise the following commands will not work.

In the prompt type:

cd /your/path/to/STarFish-master

For windows use "\" instead of "/", for example:

cd \your\path\to\STarFish-master

Replace "/your/path/to/" with the pathname of the directory where you extracted the repository files to.

So if you extracted to your Downloads folder you would type in the prompt:

cd Desktop/STarFish-master

or for windows:

cd Desktop\STarFish-master

If you extracted to a different location and are unfamiliar with naviagation via pathnames, then open the file explore GUI, navigate to the directory, and copy the path name from the address bar into the anaconda prompt. For example:

cd C:\Users\cockr\Documents\STarFish-master

Once you have navigated to the repository in the Anaconda Prompt install the anaconda environment from the environment.yml file.

In the prompt type:

conda env create -f environment.yml

The creation of the conda environment may take several minutes.

Following successulf creation of the conda environment, activate the environment:

conda activate STarFish

Change to the web_app directory:

cd web_app

Set the FLASK_APP environmental variable:

export FLASK_APP=app.py

or for windows:

set FLASK_APP=app.py

Run the STarFish web application:

flask run

In your web browser, navigate to the "Running on" url that was listed in the prompt:

* Running on http://127.0.0.1:5000/ (Press CTRL+C to quit)

Building the models used in STarFish

The build_model folder contains the python scripts that were used to generate the models in STarFish. The models were trained using HPC cluster resources and it is not realistic to recreate them on a single machine. Bash scripts and pbs files submission files are provided which can be used to re-create the models. However, these require the use of a bash shell and access to a HPC cluster that uses a PBS batch scheduling system. It is extremely likely the .sh and .pbs files would need to be modified for your environment prior to running them.