Dyes

About us

We are computational chemists that created this program at the University of Mississippi. The goal of our project is to create and predict the properties of exotic molecular dyes for dye sensitized solar cells. The Benchmark folder contains our training set and the JSON folder has the all of our data. If you would like to use this program for your research please email tsantaloci@gmail.com.

How The Code Works

This program performs four tasks:

• (Build) Combines SMILES strings and creates gaussian input files with there corresponding coordinates.
• (Manage) Submits Jobs and resubmits jobs on the Mississippi Center Supercomputer account which runs on an PBS workload manager
• (Gather) Gathers data from Gaussian output files and places them into JSON files to determine the best candidates.
• (Analysis) Perform analysis on generated JSON files to find trends, improve computational predictions, and determine best candidates for experimental testing.

Steps

Select what task you would like to perform Build, Manage, Gather, or Analysis

Build

1. input SMILES strings (copied from ChemDraw):

   • python3 inputs_v2.py
   • Follow inputs_v2.py instructions and SMILE strings will go in either the eAcceptor, backbone or eDonor directory

2. submit final.py script!

   • SMILES strings will be combined and input files will be placed in the results directory.

3. python3 final.py -B True

Manage

qsub_queue will keep track of what jobs need to be submitted to the cluster next

1. python3 final.py -M True

Gather

Parses output files to generate JSON and .pkl files

1. python3 final.py -G True

Analysis

Create figures and score computed structures

1. Relevant functions are in src/gather_results.py

Combinatorial Approach

Examples

Selecting the Specific Type

This code is designed for four types of Dye Sensitized Solar Cells:

• Single Donor 𝝅-conjugated dye sensitized solar cell (D-𝝅-A)
• Double Donor 𝝅-conjugated dye sensitized solar cell (D-D-𝝅-A)
• Triple Donor 𝝅-conjugated dye sensitized solar cell (D-D-D-𝝅-A)
• Donor Acceptor Donor dye sensitized solar cell (D-A-D)

Dependencies

Package	Version
matplotlib	3.3.3
networkx	3.0
numba	0.55.1
numpy	1.19.4
pandas	1.2.3
rdkit	2022.9.5
requests	2.28.2
scikit_learn	1.2.2
scipy	1.6.1

For a simple installation of packages, run conda create --name <env> --file requirements.txt

System Package Requirements

• OpenBabel- Creates the SMILE strings and converts them to Cartesian Coordinates (Mandatory)
• ChemDraw - The user inputs chemdraw generated SMILE strings (Recommended for substructure generation)

Future Ideas

These are future ideas for the project. If anyone would like to do them feel free to contact us and we are willing to help. Everyone in this project has split up but I think someday atleast one of us will come back to it.

• Update scoring for high voltage dye sensitized solar cells
• Setting up code for porphyrin solar cell molecules
• Benchmark study on porphyrin solar cells to create a scoring methodology
• Setting up code for perovskite solar cell molecules
• Benchmark study on perovskite solar cell molecules to create a scoring methodology
• Explore the possibility of finding new catalysts
• Setting up code for the ruthenium dye database that already exists