Skip to main content
. 2019 Jul 8;8:e43803. doi: 10.7554/eLife.43803

Figure 5. Robustness of cNMF to the number of components (K).

Line plots of the maximum Pearson correlation between each of the cNMF components presented in the main analysis, and the cNMF components that result from multiple choices of K. For the simulated data, for which we have access to ground truth, we plot the correlation between the inferred components for each choice of K and the ground truth 14 components. We highlight components corresponding to activity GEPs with distinct colors and denote the number of identity GEPs contained on the same plot in parenthesis in the legend. A dashed line indicates the K choice that was presented in the main analysis. Pearson correlations are computed considering only the 2000 most over-dispersed genes and on vectors normalized by the computed sample standard deviation of each gene.

Figure 5—source data 1. Application of cNMF to pancreas data and analysis of robusness to the choice of K.
elife-43803-fig5-data1.zip (2.2MB, zip)
DOI: 10.7554/eLife.43803.027

Figure 5.

Figure 5—figure supplement 1. Characterization of GEP usage across biological replicates.

Figure 5—figure supplement 1.

(a) Heatmaps of aggregated GEP profiles for each biological replicate of the Hrvatin et al. (2018) data, derived by summing the GEP usage vectors for all cells from a replicate. We plot the GEP profiles normalized to sum to one for each replicate (row) in the left panel or to sum to one for each GEP (column) in the right panel. The left panel contrasts the composition of the replicates while the right panels visualizes which replicates contributed most to each GEP. We use yellow arrows to highlight the usage of the depolarization-induced GEPs (ERP, LRP-S, and LRP-D) in replicates of mice treated with the stimulus condition (replicate names ending in _1 hr or _4 hr denoting the 1 hr or 4 hr treatments, respectively). For all heatmaps, rows are ordered by hierarchical clustering of the row-normalized matrix using the cosine metric and average linkage method. (b) The same as (a) but for the Quadrato et al. (2017) organoid data. We use yellow arrows to highlight the variability corresponding to the bioreactor from which the organoids derived (indicated in the beginning of the name).

Figure 5—figure supplement 2. Comparison of cNMF usages with the published cell-type clusters from Baron et al. (2016).

Figure 5—figure supplement 2.

Box and whisker plot of the usage of each GEP (column) in cells of each cluster from Baron et al. (2016) (rows) stratified by the donor of origin of each cell (hue). Boxes represent interquartile range, whiskers represent 5th and 95th percentiles.