Figure S1.
Classification of morphotypes captures heterogeneity of cell states in EwS cells. (a) TC32 cell expressing F-Tractin-mRuby2 and cytoplasmic GFP. Note that F-Tractin better captures detailed features of the cell contour. Scale bar, 10 µm. (b) Sphericity single-cell level (n = 15, 22, 15; 19, 23, 18; 19, 41, 37; 18, 25, 28; 25, 26,19 for HBV/PVS/CHT of NIH-3T3, TC71, TC32, TC32 C shRNA, TC32 EF1 shRNA, respectively) or fish-level (n = 13, 11, 10; 12, 11, 12; 11, 16, 13; 10, 7, 4; 14, 12, 10 for HBV/PVS/CHT of NIH-3T3, TC71, TC32, TC32 C shRNA, TC32 EF1 shRNA, respectively) show similar distributions, with low fish-to-fish variability. Significant shifts between experimental groups already for a single feature at the fish level indicates that cell type, molecular condition, and seeding site play a role in the control of cell morphogenesis (ANOVA, P < 0.0001). (c) Violin plots (left) and similarity matrices displaying P values (right, Kruskal-Wallis test, post-hoc Dunn’s test) of two prominent individual features, sphericity (top) and volume (bottom). Individual features alone cannot capture the full heterogeneity of the dataset (n = 52, 60, 97, 71, 70 for NIH-3T3, TC71, TC32, TC32 C shRNA, TC32 EF1 shRNA, respectively). (d) Histogram of the relative contributions of PCs. The first two PCs capture 86% of the heterogeneity in the data and are used for further analysis. (e) A similarity matrix based on pairwise permutation tests of the Tukey median values of the first two PCs shows significantly different distributions between all experimental groups except TC32 and TC32 C shRNA. (f) A biplot displaying the relative contribution of each geometric feature to the first two PCs. All features appear to contribute to the first two PCs, with sphericity and volume strongly influencing PCs 1 and 2, respectively. (g) When excluding outlier group NIH-3T3, the PC space (top) and contribution of features (bottom) remain largely the same while the classes become shifted, suggesting the PCA for the 12 3D global features defined constitutes a robust space for the definition of morphotype classes.

Classification of morphotypes captures heterogeneity of cell states in EwS cells. (a) TC32 cell expressing F-Tractin-mRuby2 and cytoplasmic GFP. Note that F-Tractin better captures detailed features of the cell contour. Scale bar, 10 µm. (b) Sphericity single-cell level (n = 15, 22, 15; 19, 23, 18; 19, 41, 37; 18, 25, 28; 25, 26,19 for HBV/PVS/CHT of NIH-3T3, TC71, TC32, TC32 C shRNA, TC32 EF1 shRNA, respectively) or fish-level (n = 13, 11, 10; 12, 11, 12; 11, 16, 13; 10, 7, 4; 14, 12, 10 for HBV/PVS/CHT of NIH-3T3, TC71, TC32, TC32 C shRNA, TC32 EF1 shRNA, respectively) show similar distributions, with low fish-to-fish variability. Significant shifts between experimental groups already for a single feature at the fish level indicates that cell type, molecular condition, and seeding site play a role in the control of cell morphogenesis (ANOVA, P < 0.0001). (c) Violin plots (left) and similarity matrices displaying P values (right, Kruskal-Wallis test, post-hoc Dunn’s test) of two prominent individual features, sphericity (top) and volume (bottom). Individual features alone cannot capture the full heterogeneity of the dataset (n = 52, 60, 97, 71, 70 for NIH-3T3, TC71, TC32, TC32 C shRNA, TC32 EF1 shRNA, respectively). (d) Histogram of the relative contributions of PCs. The first two PCs capture 86% of the heterogeneity in the data and are used for further analysis. (e) A similarity matrix based on pairwise permutation tests of the Tukey median values of the first two PCs shows significantly different distributions between all experimental groups except TC32 and TC32 C shRNA. (f) A biplot displaying the relative contribution of each geometric feature to the first two PCs. All features appear to contribute to the first two PCs, with sphericity and volume strongly influencing PCs 1 and 2, respectively. (g) When excluding outlier group NIH-3T3, the PC space (top) and contribution of features (bottom) remain largely the same while the classes become shifted, suggesting the PCA for the 12 3D global features defined constitutes a robust space for the definition of morphotype classes.

This Feature Is Available To Subscribers Only

or Create an Account

Close Modal
Close Modal