Image-activated pico-injection for single-cell analysis.

The addition of reagents into preformed droplets is a crucial yet intricate task in droplet-based applications where sequential reactions is required. Pico-injection offers high throughput and robustness in accomplishing this task, but the existing pico-injection techniques work in an indiscriminate manner, making it difficult to target particular groups of droplets. Here we report image-activated pico-injection (imgPico) for label-free, on-demand reagent supplementation into droplets. The imgPico detects the droplets of interest by real-time image analysis and makes decisions for the downstream pico-injection operation. We studied the performance of different algorithms for the image analysis and optimized the experimental settings of the imgPico. In the validation experiment, the imgPico successfully injected fluorescent dyes into droplets encapsulating one, two, and three cells, respectively, as expected. We further demonstrated the utility of imgPico by targeting droplets encapsulating single cells in droplet-based single-cell RNA sequencing (scRNA-seq) using exceedingly high cell density, and the results showed that the imgPico effectively reduced the presence of doublets in the scRNA-seq data. With the merits of being label-free and versatile, the imgPico represents a technical advance with potential applications in single-cell analysis.

CoID-LAMP: Color-Encoded, Intelligent Digital LAMP for Multiplex Nucleic Acid Quantification.

Multiplex, digital nucleic acid tests have important biomedical applications, but existing methods mostly use fluorescent probes that are target-specific and difficult to optimize, limiting their widespread applications. Here, we report color-encoded, intelligent digital loop-mediated isothermal amplification (CoID-LAMP) for the coidentification of multiple nucleic acid targets. CoID-LAMP supplements different primer solutions with different dyes, generates primer droplets and sample droplets, and collectively pairs these two types of droplets in a microwell array device to perform LAMP. After imaging, the droplet colors were analyzed to decode the primer information, and the precipitate byproducts within droplets were detected to determine the target occupancy and calculate the concentrations. We first established an image analysis pipeline based on a deep learning algorithm for reliable droplet detection and validated the analytical performance in nucleic acid quantification. We then implemented CoID-LAMP using fluorescent dyes as the coding materials and established an 8-plex digital nucleic acid assay, confirming the reliable coding performance and the capability of multiplex nucleic acid quantification. We further implemented CoID-LAMP using brightfield dyes for a 4-plex assay, suggesting that the assay could be realized solely by brightfield imaging with minimal demand on the optics. Leveraging the advantages of droplet microfluidics in multiplexing and deep learning in intelligent image analysis, CoID-LAMP offers a useful tool for multiplex nucleic acid quantification.

Combinatorial perturbation sequencing on single cells using microwell-based droplet random pairing.

Combinatorial drug therapy reduces drug resistance and disease relapse, but informed drug combinations are lacking due to the high scale of possible combinations and the relatively simple phenotyping strategies. Here we report combinatorial perturbation sequencing (CP-seq) on single cells using microwell-base droplet random pairing. CP-seq uses oligonucleotides to barcode drugs, encapsulates drugs and cells in separate droplets, and pairs cell droplets with two drug droplets randomly on a microwell array chip to complete combinatorial drug treatment and barcode-tagging on cells. The subsequent single-cell RNA sequencing simultaneously detects the single-cell transcriptomes and drug barcodes to demultiplex the corresponding drug treatment. The microfluidic droplet operations had robust performance, with the overall utilization rate of the microwells being up to 83%. We then progressively validated the CP-seq by performing single-drug treatments and then combinatorial-drug treatments, confirming the CP-seq's capability in the collection and analysis of drug-perturbed transcriptomes. Leveraging the advantage of droplet microfluidics in massive multiplexing, the CP-seq represents a great technology for combinatorial perturbation screening with high throughput and comprehensive profiling.