Exploring QSAR of FLAP inhibitors using kernel partial least squares modeling: Insights from molecular binary fingerprints

Abstract

FLAP (5-Lipoxygenase Activating Protein) inhibitors, offering targeted intervention in leukotriene biosynthesis and holding therapeutic promise for inflammatory diseases like asthma, are hindered by current inhibitors' off-target effects, limited efficacy, safety concerns, potential drug interactions, and accessibility issues. Given these challenges, computational methods, particularly Quantitative Structure Activity Relationships (QSAR) modeling, are vital for developing novel FLAP inhibitors. This study specifically investigates the QSAR of FLAP inhibitors using Kernel Partial Least Squares (KPLS) modeling. Leveraging a dataset of FLAP inhibitors, we employ KPLS within the Schrödinger Canvas environment to correlate molecular descriptors with biological activity. Out of the eight models developed, the "atom pairs" fingerprint yielded a statistically significant 2D QSAR model with outstanding regression coefficient values (R2=0.9624). The model demonstrated high predictive ability for external test set data (R2pred = 0.7105), underscoring its robustness and reliability in accurately predicting biological activity based on molecular structure. Additionally, we tried to visualize the relative contributions of individual atoms within FLAP inhibitors, providing insights into their favorable and unfavorable characteristics. Through the analysis of atomic contributions, we identify key structural motifs crucial for predicting FLAP inhibitor activity. Our findings not only advance our understanding of FLAP inhibitor SAR but also demonstrate the utility of KPLS modeling and atomic contribution analysis in drug discovery efforts. Furthermore, this study contributes to the development of anti-inflammatory therapeutics by elucidating the structural determinants of FLAP inhibitor activity, with potential applications in the treatment of inflammatory disorders.