TigerChem: An Interactive Computational Platform for Molecular Property Research

Presentation Type

Visual Display

Conference Date

Spring 4-17-2026

Abstract

TigerChem is an interactive platform that enables efficient and accessible molecular property prediction for researchers and students. Experimental determination of molecular properties is time-consuming, resource-intensive, and occasionally prone to inaccuracies. The growing scale and complexity of molecular datasets present significant challenges for pattern recognition, impacting critical applications in drug discovery, materials science, and chemical manufacturing.

Today, performing end-to-end molecular property prediction and interpretation typically requires programming expertise to combine multiple cheminformatics and machine-learning steps into a single workflow. TigerChem lowers this barrier by providing an interactive, unified interface that guides users through the full analysis process. Users can upload molecular datasets as SMILES strings, which are transformed into numerical descriptors to support unsupervised learning for structural pattern discovery, regression modeling for property prediction, and interpretability analysis using SHAP (Shapley Additive Explanations).

Beyond integrating these capabilities, TigerChem improves efficiency by automating preprocessing, reducing analysis time, and enabling users to quickly identify structural trends. The platform highlights key descriptors that drive model outcomes, helping users make better-informed decisions. By making end-to-end analysis accessible without requiring coding experience, TigerChem supports both education and research in molecular property prediction.

Presenter

Hannah Akala

Faculty Mentor

Jose Romagnoli

Award

Top 5 Individual Presenter, LSU College of Engineering

Academic Major

Chemical Engineering

This document is currently not available here.

Share

COinS