The demand for high-throughput, cost-effective single-cell sequencing methods has driven technological advancements in the field. Existing droplet-based microfluidic approaches, while streamlining analysis, suffer from low cell concentrations to minimize multi-cell droplets, leading to significant reagent and barcode underutilization. Combinatorial indexing on microplates offers higher efficiency but involves complex and time-consuming protocols. This paper introduces Overloading And Unpacking (OAK), a novel method that leverages the advantages of both approaches. OAK utilizes a droplet-based system for initial cell compartmentalization and barcoding, followed by a second indexing round via aliquoting to achieve combinatorial indexing. This strategy aims to maximize barcode utilization, increase throughput, and simplify the experimental workflow while maintaining sensitivity. The researchers anticipated OAK's versatility across various molecular modalities and its applicability to complex samples, enabling investigation of rare cell populations and cellular responses to perturbations.

The authors review existing single-cell sequencing techniques, highlighting the trade-offs between throughput, cost, and sensitivity. Droplet-based methods like those from 10x Genomics are lauded for their ease of use and scalability but criticized for their underutilization of reagents due to the necessity of low cell loading to minimize multiplets. Conversely, combinatorial indexing techniques, while highly efficient in barcoding, are noted for their labor-intensive nature and complex protocols. This review sets the stage for the introduction of OAK as a method addressing these limitations.

OAK involves two main steps: 1. **Droplet-based Overloading and Initial Indexing:** The method begins with overloading a commercially available droplet-based microfluidic system (like the 10x Genomics Chromium system) with a high concentration of fixed cells or nuclei. This results in a higher proportion of droplets containing cells, maximizing reagent utilization. The first round of indexing is performed within these droplets. 2. **Aliquoting and Combinatorial Indexing:** After reverse transcription (for RNA-seq) within the droplets, the emulsions are broken, and the cells are recovered. These cells are then re-distributed into multiple aliquots, each receiving a unique secondary index. This second indexing step, combined with the initial droplet index, creates a combinatorial barcode system capable of identifying individual cells even within initially multi-cell droplets. The aliquots are then processed independently into sequencing libraries. The study demonstrates OAK's compatibility with various modalities including scRNA-seq, paired snRNA-seq and snATAC-seq, and antibody-based cell hashing for sample multiplexing. Formaldehyde and methanol fixation were compared for multiome experiments. Detailed protocols for cell preparation, library preparation, sequencing, and data analysis are described in the Methods section. Simulations are used to model the expected cell distribution in droplets based on Poisson distribution. Theoretical estimates of multiplet rates are also calculated.

The study demonstrates that OAK significantly increases throughput compared to standard droplet-based methods. Using a 1:1 mixture of mouse and human cells, they assessed multiplet rates, showing that even with higher cell loading, the multiplet rate remained within acceptable ranges. Compared to existing ultra-high throughput methods, OAK showed higher sensitivity (measured by the number of genes and transcripts detected per cell). OAK was shown to be compatible with cell hashing for sample multiplexing, achieving high hashtag assignment rates and maintaining accurate cellular composition. The method's suitability for paired snRNA-seq and snATAC-seq was established, with formaldehyde fixation preferred over methanol for multiome experiments. Using human retinal tissue, OAK generated paired snRNA-seq and snATAC-seq data for a large number of cells, allowing the identification of unique open chromatin regions in different retinal cell subtypes and the inference of transcription factor activity. Finally, using melanoma cells treated with a RAF inhibitor, OAK revealed the emergence of a rare resistant cell population (0.12% at baseline) by tracking cell lineages over time. This resistant lineage showed activation of specific pathways (EGFR and TGF-β) and upregulation of Fibronectin 1 (FN1), indicative of a transition towards an undifferentiated state. The study also indicated that pre-existing transcriptional differences between cells influence the response to the RAF inhibitor. The number of cells processed ranged from tens of thousands to hundreds of thousands depending on the experiment.

OAK successfully addresses limitations of current single-cell sequencing methods. The combination of droplet-based technology and combinatorial indexing yields high throughput and sensitivity while simplifying the experimental procedures. The ability to perform paired multiomic profiling on complex tissues like the retina and to identify rare resistant clones in drug treatment studies highlights the method's power. The compatibility with established analysis pipelines and the potential for cost reduction due to reagent efficiency and reduced need for custom oligo synthesis further enhance its attractiveness. The researchers acknowledge that while the method is primarily validated on the Chromium platform, it has the potential to be adapted to other platforms.

OAK presents a significant advancement in single-cell multiomic profiling, offering high throughput, versatility, and cost-effectiveness. Its successful application across diverse experimental setups and modalities demonstrate its potential to become a powerful tool for large-scale molecular studies. Future research could explore its optimization for various cell types and the integration of additional molecular modalities.

The current OAK protocol shows limitations in generating high-complexity libraries for fragile cells such as PBMCs, potentially due to sensitivity to detergents. Optimization of fixation and detergent use is needed to broaden its applicability. The study focuses primarily on validation using the Chromium platform, and future work will investigate its adaptability to other droplet-based systems. While formaldehyde fixation is preferred for paired snRNA-Seq and snATAC-Seq, methanol fixation may have its own advantages in certain contexts.