Multivariable models to predict hematologic variables from driver mutations. (A) Multivariable model to predict hemoglobin value from driver mutations. The step curve shows the cumulative proportion of variance (y-axis, left) in hemoglobin levels explained by each of the genetic variables. The gray shaded area represents the 95% confidence interval (CI) for this curve. Coefficient estimates for each gene in the model including all variables (y-axis, right) are shown as circles (coefficients >0 indicate positive correlation with hemoglobin levels, ie, the covariate is associated with higher hemoglobin levels; coefficients <0 indicate negative correlation with hemoglobin levels, ie, the covariate is associated with lower hemoglobin levels). (B) Multivariable model to predict WBC from driver mutations, as for (A). (C) Multivariable model to predict BM blast count from driver mutations, as for panels A-B.