Fig. 8: HDMA results translate to humans.

A Distribution of loadings for cell-type-specific transcripts in adipose tissue (purple). Numbers in parentheses indicate the number of transcripts in each group. B Distribution of loadings for cell-type-specific transcripts in pancreatic islets (green). Each box in this panel represents 10 transcripts. C Null distributions from 10,000 permutations for the mean loading of randomly selected transcripts in each cell type compared with the observed mean loading of each group of transcripts (red asterisk). Violin plot colors indicate the tissue of each cell type and match (A and B) (purple = adipose; green = islet) (D). Predictions of metabolic disease index (MDI) in four adipose transcription data sets downloaded from GEO. In each study the obese/diabetic patients were predicted to have greater MDI than the lean/non-diabetic patients based on the HDMA results from DO mice. 1) two-sided Welch’s t test: t = 2.28, df = 58.1, 95% CI of difference = 2.1 × 10⁻⁵ to 3.2 × 10⁻⁴ A.U.; p = 0.027 2) two-sided Welch’s t test: t = 3.04, df = 11.84, 95% CI of difference = 9.3 × 10⁻⁴ to 5.7 × 10⁻³ A.U.; p = 0.01 3) linear mixed effects model: fixed effect diabetic = 1.4 × 10⁻³ ± 5.6 × 10⁻⁴ Std. Error; t = 2.4, df = 4, p = 0.075 4) two-sided Welch’s t test: t = 2.95, df = 11.89, 95% CI of difference = 6.8 × 10⁻⁵ to 4.6 × 10⁻⁴ A.U.; p = 0.012 Lines in boxes correspond to the median; lower and upper edges of boxes indicate the first and third quartiles; whiskers indicate either the minimum and maximum values if no outliers, or the first and third quartiles  ± 1.5 times the interquartile range if there are outliers; dots indicate outliers beyond 1.5 times the interquartile range. The number of patients in each group is indicated by numbers in parentheses with the number of technical replicates (reps) if relevant. No adjustments were made for multiple comparisons. Source data are provided as a Source Data file.