Technical variability can have a significant impact on single-cell experiments, potentially affecting the accuracy and reproducibility of the results. Unlike biological variability, which reflects true differences in gene expression between individual cells, technical variability arises from experimental and processing inconsistencies. In this feature, we explore the common sources of variability.
Want to optimise your experiments? Check out more features in this series here.
Technical variability in single-cell experiments can arise from several steps in the experimental workflow, each potentially introducing errors or biases that obscure true biological differences between cells. One primary source of variability is the cell isolation and capture process, where different methods (e.g., droplet-based or microwell-based) have varying levels of efficiency in isolating individual cells. Droplet-based methods, for example, can capture many cells quickly, but may also introduce biases because of uneven efficiency, leading to underrepresentation of certain cell types or subpopulations. Moreover, some cells are more fragile than others, increasing the likelihood that they will be lost or damaged during isolation, which can further skew the dataset.
Another significant source of technical variability is RNA capture efficiency and sequencing depth. The capture and conversion of mRNA to complementary DNA (cDNA) is inherently inefficient, and some transcripts may be lost or poorly represented, leading to ‘dropouts,’ where certain genes appear to be unexpressed in a cell even when they are present. This problem can be compounded by sequencing depth, which refers to the total number of reads allocated to each cell. Lower sequencing depth often results in sparse data, where only the most highly expressed genes are reliably detected, while low-abundance genes are missed, increasing the noise in the data. PCR amplification steps, often used to increase the amount of starting material, can introduce additional bias by amplifying certain transcripts over others, causing some genes to appear more highly expressed than they truly are. This can distort the observed gene expression profile, affecting downstream analysis and interpretation.
Lastly, batch effects contribute to technical variability, often arising from processing samples across multiple runs or days, using different reagents, or operating under slight environmental changes. These systematic differences can lead to inconsistencies, causing cells from different batches to exhibit artificial differences in gene expression. To mitigate these effects, researchers commonly apply batch correction techniques, though this can sometimes be challenging in single-cell data, where batch effects and biological differences may be difficult to separate. Recognising these common sources of technical variability and incorporating experimental controls, standardised protocols and computational correction methods can significantly enhance the accuracy and reproducibility of single-cell studies.
The above issues tend to stem from a lack of standardisation across labs and a lack of quality control protocols, but can also be attributed to data storage issues, the time-consuming nature of sample prep (which can often involve a lot of trial and error) and hesitancy around upfront costs for new tools.
The following information contains intelligence contributed by Ashleigh Lister (Senior Research Assistant, Earlham Institute), Nancy Zhang (Professor of Statistics, University of Pennsylvania), Kristen Beaumont (Icahn School of Medicine) and Aridaman Pandit (Senior Research Scientist, AbbVie) during our recent FOG Live: Single-Cell and Spatial event.
