T2D Research

Gu et al. 2024 × Li et al. 2025

Properties2
tagssynthesis, type-2-diabetes, pbmc, immunology, transcriptomics
aliasesGu et al. 2024 × Li et al. 2025

3 min read

Gu et al. 2024 × Li et al. 2025

The Connection

Gu et al. 2024 generated and published the GSE268210 PBMC scRNA-seq dataset from Korean non-diabetes and T2D participants. Li et al. 2025 reused the same T2D single-cell data from GSE268210, added healthy controls from a different accession (GSE244515), and applied a distinct analytical pipeline focused on immunometabolic subtyping rather than the broader cell-composition and cell-state characterization Gu et al. performed. The two papers are not independent replicates — they are complementary analyses of overlapping primary data, each revealing different biological dimensions.

Where They Co-occur

Both references appear together in the PBMC immune changes hub, the single-cell PBMC profiling concept page, the cytotoxic T-cell expansion and monocyte inflammatory signature concept pages, the type-2-diabetes concept page, the GSE268210 entity page, the project page, and the evidence map synthesis. Their co-occurrence across 8 pages reflects the central role both studies play in the paper’s T2D PBMC background, while also creating a recurring interpretive challenge: claims from Li et al. should not be cited as independent replication of Gu et al.’s T2D findings.

Cross-cutting Insight

The value of this pair lies in understanding what each analytical lens reveals that the other misses. Gu et al.’s original analysis emphasizes cell-composition shifts (monocyte subsets, cytotoxic T-cell expansion, B-cell differentiation) and receptor-repertoire features (TCR clonality, BCR isotype diversity), providing a broad phenotypic map of T2D PBMC remodeling. Li et al.’s reanalysis emphasizes T-cell metabolic heterogeneity, inferred cell-cell communication, inferred transcription-factor activity, and drug-enrichment hypotheses — computationally inferred layers not emphasized by Gu et al.’s original analytical scope. Together they show that a single well-characterized PBMC dataset can support multiple conceptual framings of T2D immune dysregulation, but neither study should be treated as an independent validation of the other’s T2D-vs-control comparisons when the T2D cases are shared.

Tensions and Trade-offs

  • Li et al. report increased monocyte proportions in T2D while Gu et al. report lower CD14 monocyte and higher CD16/intermediate monocyte proportions. This apparent discrepancy spans different resolution levels and control designs, and the overlap in T2D samples means it cannot be resolved by appealing to independent cohort replication.
  • Li et al.’s immunometabolic subtypes (A, B, C) are new analyses on overlapping T2D cases, but they are not independent validation of T2D-vs-control effects from Gu et al.
  • Li et al. add GSE244515 controls from a different GEO accession, introducing a potential dataset-effect confound that Gu et al.’s internal-control design avoids.
  • For manuscript use, Gu et al. provides the strongest single-study evidence for T2D PBMC remodeling, while Li et al. provides mechanistic hypotheses that should be attributed to the same underlying cohort with transparent methodological differences.

Open Questions

  • Would Li et al.’s immunometabolic subtypes replicate if computed on an independent T2D PBMC scRNA-seq cohort?
  • Do the CellChat-inferred communication patterns Li et al. report reflect T2D biology or are they sensitive to the cross-accession control design?
  • Which of Li et al.’s drug-enrichment hypotheses would survive if computed on subtype DEGs from an independent dataset?

Graph View

Backlinks