cellPLATO: an unsupervised method for identifying cell behaviour in heterogeneous cell trajectory data
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (Arcadia Science)
Abstract
Advances in imaging, cell segmentation, and cell tracking now routinely produce microscopy datasets of a size and complexity comparable to transcriptomics or proteomics. New tools are required to process this ‘phenomics’ type data. Cell PLasticity Analysis TOol (cellPLATO) is a Python-based analysis software designed for measurement and classification of diverse cell behaviours based on clustering of parameters of cell morphology and motility. cellPLATO is used after segmentation and tracking of cells from live cell microscopy data. The tool extracts morphological and motility metrics from each cell per timepoint, before being using them to segregate cells into behavioural subtypes with dimensionality reduction. Resultant cell tracks have a ‘behavioural ID’ for each cell per timepoint corresponding to their changing behaviour over time in a sequence. Similarity analysis allows the grouping of behavioural sequences into discrete trajectories with assigned IDs. Trajectories and underlying behaviours generate a phenotypic finger-print for each experimental condition, and representative cells are mathematically identified and graphically displayed for human understanding of each subtype. Here, we use cellPLATO to investigate the role of IL-15 in modulating NK cell migration on ICAM-1 or VCAM-1. We find 8 behavioural subsets of NK cells based on their shape and migration dynamics, and 4 trajectories of behaviour. Therefore, using cellPLATO we show that IL-15 increases plasticity between cell migration behaviours and that different integrin ligands induce different forms of NK cell migration.
Article activity feed
-
-
cellPLATO performs UMAP on morphological/motility parameters then uses HDBSCAN cluster analysis to define behavioural clusters
It is hard to tell from the text if HDBSCAN is run on the behavioral parameters or on the UMAP output. If the latter, then I would take extreme caution in thinking about the generalizability of the method given the numerous issues with clustering on nonlinear manifolds. Either way, it would also be helpful to report more information on what the specific morphological/motility parameters are and any normalizations/manipulations that were done on them prior to UMAP and clustering.
Also, any justification for choosing of UMAP and HDBSCAN would be useful.
-
UMAPs 1, 2 and 3
This might be a slightly confusing way to refer to UMAP dimensions (is it accepted that a UMAP dimension = a single 'UMAP'?)
-
We first investigated two fundamental measurements of cell migration and morphology, namely cell speed and cell area. When comparing conditions, the median migration speed of NK cells on VCAM-1 was 3.48 μm/min and 2.54 μm/min on ICAM-1 (Fig. 2A). The effect size distribution for VCAM-1 was greater, demonstrating statistical significance (p <0.00001) (41), and its distribution did not overlap with the control condition (ICAM-1). NK cells migrating on VCAM-1 also had smaller median cell area (114 μm2) compared with ICAM-1 (175 μm2) (Fig. 2B), with nonoverlapping effect size distribution (p < 0.00001).
Does donor identify have any effect here? Do the donors differ at all in their speed/area distributions and effect sizes? This would be useful to know here and for many other analyses presented in the manuscript. More broadly, it is a …
We first investigated two fundamental measurements of cell migration and morphology, namely cell speed and cell area. When comparing conditions, the median migration speed of NK cells on VCAM-1 was 3.48 μm/min and 2.54 μm/min on ICAM-1 (Fig. 2A). The effect size distribution for VCAM-1 was greater, demonstrating statistical significance (p <0.00001) (41), and its distribution did not overlap with the control condition (ICAM-1). NK cells migrating on VCAM-1 also had smaller median cell area (114 μm2) compared with ICAM-1 (175 μm2) (Fig. 2B), with nonoverlapping effect size distribution (p < 0.00001).
Does donor identify have any effect here? Do the donors differ at all in their speed/area distributions and effect sizes? This would be useful to know here and for many other analyses presented in the manuscript. More broadly, it is a little hard to assess the generalizability of the behavioral results presented here (including the cellPLATO analyses) without knowing more about the influence of experimental variables like this.
-