Decision tree for atypical Alzheimer’s disease. Binary decision tree created by receiver operator characteristic (ROC) analysis is shown. Branching points represent the variable and cutting point which best predicts whether or not an individual will be diagnosed with an atypical form of Alzheimer’s disease (AtD). Shaded boxes depict the variable used to separate each subgroup and unshaded boxes provide summary data characterizing each subgroup. For more information on the genes depicted, please see Additional file 1. NC - normal control.