Towards a 3D spheroid system for modelling leukodystrophies 119 4 Defining valid contrasts in the data set Building on our analysis of heterogeneity, we wanted to shed some light on the variation between samples using the isogenic pairs. The inclusion of isogenic pairs allowed us to control for genetic background and focus on disease-associated changes. We first compared the correlation of log2FC of the genes significant at day 150 in both HD2.2xKI and HD2.2xKO contrast. We observed a high correlation (Pearson r > 0.800) in three out of five investigated clusters, except for OPC and astrocytes (Fig. 4A-E, Pearson r: 0.669 and 0.547 respectively). When including genes significant in only one of the contrasts (Fig. 4F-J) correlation of the findings between the contrasts decreases in all clusters. Similarly, when investigating all genes (Fig. 4K-O) correlation decreases even further in three out of five clusters. Interestingly, the astrocyte (Pearson r = 0.831) and NPC (Pearson r = 0.559) clusters still show high and medium correlation. To conclude, there are quite some differences between the findings of the isogenic lines. However, if a gene is significant in both contrasts it is often correlated. The findings of the clusters NPC and astrocytes seem to correlate more strongly. Possibly indicating disease specific changes in these clusters. Next, we wanted to investigate if significant findings of isogenic comparisons correlate with non-isogenic comparisons with larger sample size (HD x GLD). First, we compared the significant genes for the contrasts HD x GLD and HD2.2 x KI at day 150, we observed a weak correlation in the OPC cluster (Fig. 5a) and a strong correlation (Pearson r > 0.800) in the proliferating OPCs and oligodendrocytes (Fig. 5A-C). In the astrocyte cluster there was a negative correlation (Fig. 5D). The NPCs cluster mainly showed replication of the upregulated genes (Fig. 5E, blue). For the comparison of HD x GLD and HD2.2 x KO we see a similar picture, there is some correlation in the oligodendrocyte lineage, although less compared to the KI comparison (Fig. 5F-H). Again, we find no correlation in astrocytes and again mainly overlay of upregulated genes in the NPC cluster (Fig. 5I-J). Noteworthy, most extreme DEGs such as DPP6 and CHCHD2 in OPC and VAMP2 in NPC are replicated in all contrasts. We conclude that there are, as expected, differences in findings when changing contrasts. Nevertheless, it is reassuring that some of the most significant genes are consistently up or down-regulated independent of the contrast. Hence, we felt confident to use the current data set for the exploration of disease specific findings in the other leukodystrophies as well.
RkJQdWJsaXNoZXIy MTk4NDMw