Regulatory Network Inference Results
This note records the figure-level result from Hasan’s REACTOR analysis of binarized SCENIC regulons. The corresponding method is documented in Regulatory Network Inference.
Figure 1. Top cell type-resolved activating regulon responses to T2D. For each donor-cell_type combination, binarized SCENIC regulons were summarized as the percentage of cells with active regulon signal. Panel A shows T2D - Control differences in active-cell percentage points for selected activating regulons across Chechen, Tatar, and Yakut donors. Points represent ancestry-specific estimates, horizontal bars show donor-level 95% confidence intervals, and filled points indicate REACTOR FDR <= 0.05. Panel B summarizes direction and evidence by ancestry, with color indicating the T2D - Control activity difference and asterisks marking FDR <= 0.05 (*) or FDR <= 0.01 (**). Rows were selected by direction-blind Fisher-combined evidence among 162 activating regulons, and each cell type was analyzed separately using donors with at least 10 cells in the subtype.
Observed Pattern From Figure
- The figure supports a cell-type-resolved regulatory response to T2D rather than a uniform PBMC-wide shift.
- Tatar donors show the clearest visual pattern of increased regulon activity in T2D, especially in B Naive Resting cells and NK CD56+ cells.
- B Naive Resting cells show prominent Tatar increases across NF-kB/AP-1-related regulons, including JUND(+), REL(+), RELB(+), NFKB1(+), CREM(+), and SOX5(+).
- NK CD56+ cells show Tatar-positive activity shifts across multiple regulons, including FOSL2(+), KLF2(+), GATA3(+).
- T CD4+ cm and T CD8+ GZMK+ populations also show ancestry-separated regulon responses, with several Tatar-positive shifts and mixed or weaker Chechen/Yakut directions.
- Mono CD14+ cells show mixed ancestry-specific responses rather than a single shared T2D direction across all selected regulons.
- Yakut responses appear weaker, mixed, or negative for several selected regulons, especially when contrasted with the Tatar-positive pattern in lymphoid compartments.
- Chechen responses are generally more modest or mixed in the displayed rows, with fewer visually dominant positive shifts than the Tatar pattern. todo we should maybe take a look at some regulons that reach significance in Chechens
Interpretation
The REACTOR results suggest that T2D-associated transcription-factor regulon activity differs by both cell type and ancestry. The strongest signal is a Tatar-enriched increase in active-cell percentages for activating regulons, particularly in B Naive Resting and NK CD56+ compartments. This pattern is consistent with the broader working model that T2D immune remodeling is not only abundance- or expression-level, but also includes ancestry-dependent regulatory-state changes. The result should be framed as computational regulon activity inferred from binarized SCENIC output, not as direct measurement of transcription-factor binding or protein activity.
Result Caveats
- The Figure 1 says rows were selected by direction-blind Fisher-combined evidence, so displayed rows are not an unbiased list of all tested regulons.
- Donor groups with fewer cells in a subtype were excluded using the >=10-cell threshold; donor balance by ancestry and condition should be checked. todo we checked that, we have the figure
- “Population” groups are ancestry strata within cell type, and additional covariates were not included upstream or in the ROTS comparison. Unfortunately, not correcting the batch effect as well.