Droplet microfluidics has demonstrated immense potential in microbiological studies due to its unique features, such as miniaturization, compartmentalization, and parallelization. Multiplexing droplet content allows the investigation of various experimental conditions in a highly parallelized manner. Yet, droplet library generation and tracking remain challenging in high-throughput screening. The introduction of distinct reagents into droplets necessitates precise control over droplet flow in a microfluidic chip, limiting the throughput to a few reagents. Additionally, tracking individual droplets is complex due to their fast dynamics. To address these challenges, we have developed a multiplexing platform for automated sample preparation, enabling on-demand merging and mixing of reagents for fine-tuning the sample compositions for droplet generation. A coding space with 169 optical barcodes can be realized by the pairwise combination of four fluorescence dyes at six concentration levels to encode droplet populations as required by the experimental design. A machine-learning algorithm has been employed to identify distinct droplet populations. As proof of concept, we conducted an antibiotic susceptibility assay on an E. coli strain to screen for susceptibility of four antibiotics and determine minimum inhibitory concentrations in one experiment. Utilizing the on-demand sample preparation, optical barcodes, and machine-learning analysis, our setup provides a rapid, straightforward, and reliable multiplexing capability for numerous microbial and biochemical applications.
Read full abstract