Fig. 6

Deconvolution of cellular mixtures based on their DNA methylation profiles. A Bar plots of relative proportions of cell subsets in each in vitro mixture and lysed blood sample. For the in vitro cellular mixtures, ‘expected’ refers to the known proportions of FACS-purified cell types used to generate the mixtures. For the lysed blood samples, ‘expected’ refers to the proportions of each cell subset estimated by flow cytometry. The subsequent four bars for each mixture show the proportions of each cell type estimated by CIBERSORTx using reference samples from specified breeds, where HF denotes Holstein Friesian, ND denotes N’Dama and Ne denotes Nelore. Of note, the Holstein Friesian lysed blood samples were collected from different cattle from which the reference profiles were derived, while the N’Dama and Nelore lysed blood samples were collected from the same cattle as the reference profiles. B, C Scatterplots comparing flow cytometry with CIBERSORTx estimates where only the Holstein Friesian (B) or Nelore (C) reference samples were used to predict the compositions of the mixtures. Concordance was determined by Spearman's rank correlation (rs) and linear regression (solid line). gdT cell denotes γδ T cell