Predicting Molecular Properties

[Sgvkamalakar]

Pull Request for ML-Crate 💡

Issue Title: Predicting Molecular Properties #636

• Info about the related issue (Aim of the project) : This project aims to develop predictive models to estimate scalar coupling constants between atom pairs in molecules based on their structural features and conduct exploratory data analysis (EDA) to uncover relationships between molecular properties and structural characteristics.
• Name: S G V Kamalakar
• Email ID for further communication: [email protected]
• GitHub ID: Sgvkamalakar
• Idenitfy yourself: SSOC S3

Closes: #636

Describe the add-ons or changes you've made 📃

Give a clear description of what have you added or modifications made

Modifications Made:

Added interactive 3D visualizations of molecular structures using Plotly, color-coded by atom type, and implemented animations to highlight atoms and bonds.

Analyzed and visualized the distribution of unique atom counts with histograms. Conducted extensive EDA, visualizing scalar coupling constants, Mulliken charges, dipole moments, and potential energy.

Created new features based on atomic interactions, standardized molecular data, and trained multiple regression models to predict scalar coupling constants, evaluating performance with MAE, MSE, and R².

Type of change ☑️

What sort of change have you made:

• New feature (non-breaking change which adds functionality)

How Has This Been Tested? ⚙️

Describe how it has been tested
Describe how have you verified the changes made

The changes were tested and verified through various steps. Interactive 3D visualizations were rendered and cross-checked against known molecular structures.

Histograms of unique atom counts were generated and validated by comparing them with raw data. EDA visualizations were inspected for accuracy and consistency.

Newly engineered features were analyzed for relevance and contribution to model performance.

Finally, regression models (Linear Regressor, Random Forest Regressor, K Nearest Neighbors, Support Vector Regressor, Decision Tree Regressor, and Simple Feed Forward Neural Network) were evaluated against benchmarks, and the entire workflow was validated for reproducibility in a clean environment.

Checklist: ☑️

• My code follows the guidelines of this project.
• I have performed a self-review of my own code.
• I have commented my code, particularly wherever it was hard to understand.
• I have made corresponding changes to the documentation.
• My changes generate no new warnings.
• I have added things that prove my fix is effective or that my feature works.
• Any dependent changes have been merged and published in downstream modules.

[github-actions]

Our team will soon review your PR. Thanks @Sgvkamalakar :)

[abhisheks008]

Looks good to me. Approved @Sgvkamalakar