Partitioned Polygenic Scores in Type 2 Diabetes

Partitioned polygenic scores (pPGS) are a refinement of standard polygenic risk scores. Instead of collapsing all T2D-associated SNPs into a single risk score, SNPs are grouped into biologically meaningful clusters based on shared patterns of association across multiple phenotype GWAS, so each cluster reflects a distinct T2D-relevant mechanism.

The Smith et al. 2024 Framework

Smith et al. (Multi-Ancestry Polygenic Mechanisms of T2D) identified 12 T2D genetic clusters from GWAS summary statistics for 650 T2D-associated SNPs:

Insulin deficiency:

• Beta Cell 1 — reduced corrected insulin response (CIR) and disposition index (DI)
• Beta Cell 2 — related to beta-cell function
• Proinsulin-negative — proinsulin processing

Insulin resistance:

• Obesity — BMI-driven T2D risk
• Lipodystrophy 1, Lipodystrophy 2 — adipose distribution
• Hyper Insulin — increased CIR and DI (compensatory)
• Cholesterol-negative — lipid metabolism
• Liver-Lipid — hepatic fat metabolism
• ALP-negative — alkaline phosphatase related

Less understood:

• Bilirubin
• SHBG-LpA — reduced sex hormone-binding globulin, elevated lipoprotein (a)

Application in Prior Work (Markelova et al. 2025)

Our prior work (Markelova et al. 2025) applied Smith et al.’s pPGS framework to the same Chechen, Tatar, and Yakut cohort used in the current PBMC project, using 285 of the 650 SNPs (those with weights > 0.7802 in any cluster) to maximize signal-to-noise ratio. The pPGS calculation:

pPGS = sum(weight_snp × genotype_snp) for each cluster

where genotype is coded 0, 1, or 2.

Relevance To The Paper

The pPGS approach demonstrates that genetic ancestry can be resolved to specific mechanism-level risk profiles, not just global ancestry proportions. Because the current PBMC project uses the same cohort, these pPGS results are directly available for integration — they are not an external parallel but a built-in layer of the same dataset. For the PBMC ancestry paper, this enables direct correlation between pPGS-defined T2D mechanism clusters and PBMC immune features within the same individuals.

Open Questions

• Can pPGS for T2D mechanisms be correlated with PBMC cell-type proportions or activation states in the same cohort?
• Do individuals with high Beta Cell 1 pPGS show different PBMC immune profiles than those with high Obesity pPGS?
• How do the pPGS distributions from our prior analysis relate to immune features within the same individuals?

Sources

• Markelova et al. 2025 — our prior work, same cohort as the current project