Fig. 2

Deconvolution of RNA-seq human reads into cell-type fractions using cell-type signatures. a Schematic showing the estimation of a cell-type fraction table and validation steps. b Cell-type fractions were validated using single-cell RNA-seq. Sputum from asthmatics and controls were clustered with reference cell-type expression to label clusters. c The abundances of scRNA-seq cell-type fractions are highly variable between individuals. d Correlation between the deconvolved cell-type fraction table and cell counts by microscopy. e Pearson correlation of the cell-type fraction table, F, and the LDA topic components. Only significant correlations after FDR correction are shown. f Correlation between the cell-type fraction table and the clinical table, C. Only significant correlations after FDR correction are shown. ACT, asthma control test score; Age.DX, age of asthma diagnosis; Age.SX.Onset, age of symptom onset; BDR, bronchodilator response, BMI, body mass index; FENO, forced expiratory nitric oxide; FEV1.FVC.postBD, the ratio of forced expiratory volume in 1 s to the forced vital capacity after treatment with a bronchodilator; FEV1.FVC.preBD, the ratio of forced expiratory volume in 1 s to the forced vital capacity before treatment with a bronchodilator; HIL, hospitalizations in lifetime; HPY, hospitalizations per year; ICS, average daily inhaled corticosteroid use; Number.of.OCS, average number of oral corticosteroids used; OCS.Total, lifetime total oral corticosteroid use