Table 1 Questions in plant biology to which single cell profiling could be applied: analytical problems and algorithmic solutions
Biological problem or plant-specific question | Analytical problems for single-cell data | Potential approaches |
|---|---|---|
Distinguish genes that show true biological variation | Significant technical noise is present | Hypothesis testing based on identification of variation that exceeds empirical estimations of technical noise [11] |
What genes vary among physiologically distinct cells of seemingly homogenous tissues? | Profiles have no replicates and exhibit zero-biased expression distribution, so traditional statistical methods are inappropriate | |
Model-driven deconvolution of biological variation using estimations of technical noise [20] | ||
Identify transcriptional signature of rare cell types | Linear dimensionality reduction can obscure close relationships and produce misleading clusters | Non-linear t-SNE to minimize joint probability distribution distance and draw similar cells together [29] |
What is the transcriptional profile of root initials? | Clustering methods might miss small sets of cells | |
Find subsets of cells with a unique environmental response | Separation of a continuous cell expression space into types is subjective | RaceID to identify new cell types by detecting a significant number of biological gene outliers [30] |
What is the early response of pathogen-susceptible vs. pathogen-resistant cells of the leaf epidermis? | ||
Assemble dissociated cells into a developmental sequence | Missing data-points exist owing to false negatives and misleading false positives | De novo trajectory reconstruction to order cells using Monocle [39] or diffusion-like dynamics [40] |
What is the ordered profile of specific cell types from initial to differentiated cells? | Variation in individual plants can create artificial groupings | Seurat to map cells using a priori data and imputation of missing data-points [41] |
ICI to map cells to known reference types using many markers [12, 38] | ||
Follow identity transitions during wound repair or in vitro regeneration | Detecting transitional and multiple identities must be robust in single-cell data with many false positives and false negatives | ICI to classify cells using a priori knowledge of identity markers for detecting mixed or diminished cell identity [12, 38] |
Do plant cells follow a course of de- or trans- differentiation during regeneration? |