Attractor-Based Analysis of Gene Regulatory Networks Along a Human Testicular Germ Cell Aging Trajectory

This dataset contains processed single-cell RNA sequencing data and Boolean network analysis results from human testicular tissue spanning three donors aged 26, 56, and 67 years. The original sequencing data was obtained from Deng et al. (2024) via the Human Cell Atlas Data Explorer (GEO accession GSE215754). The source study profiled five donors across three groups; the two late-onset hypogonadism donors were excluded here, and the donors retained correspond to the young adult and healthy aged groups defined by the original authors. Aged testicular tissue in the source study was obtained via surgical castration performed as prostate cancer treatment, with donors screened for antiandrogen medication history, diabetes mellitus, metastatic testicular malignancy, and abnormal BMI. The dataset includes quality-controlled and normalized expression data for 13,651 cells across 30 cell types, with germ cells (annotated as Gametocytes by SingleR against the HumanPrimaryCellAtlas reference) representing 63.2% of the population. One germ cell aging trajectory was identified through pseudotime analysis: Y_185, comprising 760 cells with a Spearman correlation of 0.679 between pseudotime and donor chronological age, representing aging as convergence toward a stable aged state. For this trajectory, the dataset provides GeneSwitches results identifying 1,451 genes with binary expression transitions during aging, a SCENIC-derived gene regulatory network of 3,036 edges across 50 transcription factors, a 1,295-gene Boolean network model with its attractor landscape, and systematic single-gene and paired-gene perturbation analysis results. Key findings include a strongly asymmetric switch profile, with 1,450 of 1,451 switch genes upregulated along the aging trajectory; an attractor landscape dominated by a single stable aged state occupying 95.7% of the sampled basin against three youthful attractors sharing the remainder; and identification of combined CREB1 and CREM knockdown as the intervention producing complete reversal of the network aging score, with strong synergy between the two genes relative to either alone. Users reanalyzing this dataset should note several constraints. The network derives from three donors along a single trajectory. No Boolean rule exceeded a quality score of 0.75. The inferred regulatory network contains only 2 repressive edges out of 3,036, and the resulting Boolean model uses no AND or NOT logic, making it a monotone system whose convergence toward a high-expression aged state may be partly structural. Attractor identification proceeded by random sampling rather than exhaustive search. The highest-ranked single-gene perturbation, AMFR knockdown, appears in the network only as a regulatory target with no outgoing edges and is most likely an artefact of state-space reduction rather than regulatory causation; it should not be treated as a candidate intervention. Full discussion of these limitations can be found here: https://maypoplabs.org/projects/prism/results_testis Furthermore, a detailed description of the methods used can be found here: https://maypoplabs.org/projects/prism/methods Source Data: Deng, Z., et al. (2024). Targeting dysregulated phago-/auto-lysosomes in Sertoli cells to ameliorate late-onset hypogonadism. Nature Aging, 4(5), 647-663. https://doi.org/10.1038/s43587-024-00614-2