Intratumoral heterogeneity in microsatellite instability status at single-cell resolution

Figures

• 

  Figure 1. MSIsensor-RNA and simulation results (A and B) Boxplots show the distribution of (A) MSI score for individuals calculated using the aggregate expression of all cells, and (B) MSI score for individuals calculated using the aggregate expression of only cancer cells. (C and D) Also shown are the mean values of (C) the F-statistic and (D) the number of subclones for the different cell mixes shown on the x axes (with increasing proportions of MSI-H cells ranging from 0.1 in mix M1 to 0.9 in mix M9). The error bars in (C) and (D) correspond to plus/minus twice the standard error around the mean. The MSS and MSI-H samples in panels C and D are the obtained values for all cells in those samples and do not represent an average. See Figure S1 showing ROC and precision-recall curves for MSIsensor-RNA as well as Tables S1 and S2 which contain summary statistics and the raw results for the mixing experiments, respectively.

• 

  Figure 2. Distributions of summary statistics in single-cell RNA sequencing data (A and B) Boxplots show the distribution of (A) F-statistics grouped by PCR/IHC MSI status and (B) the proportion of MSI-H to MSS cells grouped by PCR/IHC MSI status. (C and D) Histograms display the frequency of (C) the number of cancer cell clusters and (D) the number of tumor samples for all individuals. See Table S3, which includes the ANOVA test results for each individual.

• 

  Figure 3. MSI-H and MSS subclone compositions (A and B) Stacked bar plots of (A) the number of subclones for each individual in the analysis and (B) the proportion of subclone types for each individual. Individuals who had a PCHR/IHC test result of MSS are indicated with an asterisk.

• 

  Figure 4. Clustering of cells for the MSI-H individual (A–D) UMAP plots for MSI-H individual P24 show (A) tumor versus normal cell classification, (B) MSI scores for each cell, (C) MSI scores for re-clustered cancer cells, and (D) MSI score for the aggregated pseudobulk expression of each cancer cell cluster. Any gray colors indicate an NA value. See also Tables S4, S6, and S8, which contain the results of Tukey HSD analysis between all cancer cell clusters, differential gene expression analysis between all cancer cell clusters, and differential gene expression analysis between MSI-H and MSS cells, respectively.

• 

  Figure 5. Clustering of cells for the MSS individual (A–D) UMAP plots for MSS individual CRC2786 show (A) tumor versus normal cell classification, (B) MSI scores for each cell, (C) MSI scores for re-clustered cancer cells, and (D) MSI score for the aggregated pseudobulk expression of each cancer cell cluster. Any gray colors indicate an NA value. See also Tables S5, S7, and S9, which contain the results of Tukey’s HSD analysis between all cancer cell clusters, differential gene expression analysis between all cancer cell clusters, and differential gene expression analysis between MSI-H and MSS cells, respectively.

• 

  Figure 6. Distribution of MSI scores and difference in means for cancer cell clusters (A–D) Boxplots show the distribution of MSI scores for each cluster of cancer cells in (A) individual CRC2786 and (B) individual P24. Also shown are the 95% confidence intervals for the difference in mean MSI scores between each cluster pair for (C) individual CRC2786 and (D) individual P24.