Analysis of good and bad fingerprint for identification of NIR based optical frameworks using Monte Carlo method

Abstract

NIR-based optical materials are widely used in bio-imaging, photodynamic treatment, optoelectronics, sensing and anti-counterfeiting applications. Their advancement has significantly increased the potential for non-invasive imaging, targeted therapy and optical communication systems. Medicine, biology, material science, and telecommunications have all benefited from this progress. Considering this, we herein developed QSAR models using absorption maxima of 292 organic compounds to identify molecular features that can be utilized to shift their λmax towards NIR region. Four splits were developed using Monte Carlo Optimization and index of ideality of correlation (TF2) with CORAL software. Predictability of four models created by these splits was carried out by numerous validation metrics. Model built from second split was observed to be the best with R2validation = 0.8252 and IIC = 0.8423. Mechanistic interpretation was also performed by extracting structural attributes that are responsible for increase and decrease of λmax using correlation weights. Positive structural attributes have been utilized for designing new organic chromophores with red shifted λmax ranging between 791 and 936 nm. The present study offers new horizons for the development of NIR based chromophores.