Cluster analysis of the 55 genes used in the prediction model (ASD55).

<p>The dendrogram and heatmap on top show hierarchical clustering (average linkage) of the 99 samples in the training set (P1) and the 55 genes used in our prediction model. The first 2 lines in the graph on bottom indicate whether each sample is from the patient group or the control group. Finally, the bottom line shows the distribution of Fisher's linear discriminant scores (dots) based on ASD55 with moving average (line). The distributions of linear discriminant scores are shown on the right (blue solid line for controls and black broken line for patients). ASD cases and controls are well separated using linear discriminant analysis on the ASD55 genes.