Transfer Learning Approach to Multitarget QSRR Modeling in RPLC.

Area of Science:

• Analytical Chemistry
• Computational Chemistry
• Cheminformatics

Background:

• Quantitative Structure-Retention Relationship (QSRR) is crucial for predicting small molecule retention times in chromatography.
• Challenges exist in simultaneous target prediction under variable conditions and data scarcity for predictive modeling.
• Bridging molecular structure and chromatographic behavior is vital for analytical chemistry insights.

Purpose of the Study:

• To introduce a transfer learning-based multitarget QSRR approach to enhance retention time prediction.
• To evaluate the performance of single and multitarget QSRR models with and without transfer learning.
• To assess model performance against benchmark studies.

Main Methods:

• Comparative study of four QSRR models (with and without transfer learning).
• Evaluation using Mean Squared Error (MSE) and R-squared (R²) metrics.
• Testing individual models against established benchmark studies.

Main Results:

• Transfer learning-based multitarget QSRR models show potential for enhanced accuracy in retention time prediction.
• Models demonstrated improved predictive capabilities compared to traditional approaches.
• Performance was rigorously evaluated using MSE and R² metrics.

Conclusions:

• Transfer learning-based multitarget QSRR offers a promising avenue for improving small molecule retention time predictions.
• These models can aid in optimizing experimental conditions for Reversed-Phase Liquid Chromatography (RPLC) method development.
• The predictive models serve as valuable tools for pharmaceutical research and development.