Embracing Poisson Encapsulation Statistics for Improved Droplet Digital Immunoassay.

Abstract

Digital immunoassays enable the detection of protein biomarkers with very low concentrations, but the analysis stringently requires single-bead encapsulation. Low bead density has been adopted to minimize multiple-bead encapsulations, but the trade-off is the low droplet effectiveness (∼10%) in droplet-based assays. Here we report the method of inclusive droplet digital ELISA (iddELISA) that embraces all types of encapsulations by factoring in their varied "on-off" probabilities in the statistical inference. We derived the statistical model, optimized the bead encapsulation and immunoreaction, and developed an image analysis pipeline for accurate droplet and bead recognition, showing that approximately 40% of the droplets could be used in the analysis. Using the detection of SARS-CoV-2 nucleocapsid protein as a demonstration, the iddELISA achieved a limit of detection of 0.71 fg/mL, which was much lower than conventional ELISA as well as droplet digital ELISA. By effectively incorporating multiple bead encapsulations, the iddELISA simplified the digital immunoassay while improving the counting efficiency and sensitivity, representing a unique concept in digital immunoassays.