High-Throughput vs. High-Accuracy in Single-Cell Omics

While being extremely efficient in terms of the number of processed cells, these high-throughput approaches have major limitations.

• Low Single-Cell Accuracy. One limitation of these technologies is that they capture a relatively low number of true single cells in each droplet or microwell. Instead, there is a high likelihood of capturing multiple cells together, known as multiplets, which can complicate or interfere with downstream analyses. In the case of 10X Genomics chemistries, up to 90% of droplets may either be empty or contain multiplets.

• Lack of Sensitivity. High-throughput methods can cause gene dropout, where certain transcripts are not effectively captured, leading to reduced transcriptome coverage and biased results [6].

• No Sorting Capabilities. None of these high throughput methods allow the targeting of subpopulations, by selection of specific morphology (size or shape) and/or fluorescence patterns. When studying a cell subpopulation, such as those expressing a fluorescent protein or with a particular size range, a preliminary sorting step is needed to select this subpopulation prior to the single-cell analysis. However, when studying transcriptomes, which consist of fragile molecules with short half-lives, swift workflows should be prioritized to obtain more biologically relevant data.

• The Need for High Sample Volume and Cell Number. By definition, high-throughput systems require large sample volumes and very high cell numbers. However, they typically have low recovery rates, meaning only a small percentage of cells are usable for single-cell sequencing analysis, and they also generate a significant amount of dead volume. While this is suitable for some applications, it is not compatible with more delicate sample types, like patient-derived microbiopsies, that have a low number of cells and/or volume, or rare cells like CTCs.

• One-Size-Fits-All Approach. As in other scientific fields, high-throughput methods in single-cell analysis are primarily designed for mainstream applications—that is, for studying “normal” cells of “normal” size and “normal” shape with “normal” membranes. However, the diversity of cell types and biological questions is infinite, and the above-cited methods cannot apply to many research objects, for example cells that have different cell walls (e.g., yeasts), extra-large cells (e.g., cardiomyocytes), or atypical morphologies (e.g., neurons or adipocytes).

• What Happens Inside… Stays Inside. While high-throughput workflows generally are simple to use, they also can appear as “black boxes”. The user is virtually powerless to what is happening in the microwells or droplets they are analyzing and is unable to make changes to cell capture and molecular biology parameters. This lack of flexibility can be very limiting when facing technical challenges linked to the biological properties of the samples (e.g., the absence of a poly-A tail in procaryotic mRNA that prevents their capture by barcoded poly-T oligonucleotides) or to the specificity of some research questions (e.g., single mitochondrion sequencing).