Table 1.
Demographic data, co-morbid conditions, and use of psychiatric medications in ASD and TD subjects.
Fig 1.
Range of ADOS scores among the ASD boys.
Each dot represents the ADOS total score for a single subject.
Fig 2.
Top-10 predictive proteins identified by three different methods.
(A) Random Forest-based method. (B) T-test based method. A volcano plot of group comparison revealed a total of 86 proteins were downregulated in the ASD boys and 52 proteins were upregulated. All of these proteins had a false discovery rate <0.05. (C) Correlation-based method.
Table 2.
Top-10 predictive proteins identified by three different methods.
The 5 core proteins common to all three methods are highlighted.
Fig 3.
Optimization of the predictive proteins.
(A) Identification of the 5 core proteins. The core proteins were among the top-10 proteins identified by each of the 3 methods. (B) Identification of proteins with additive predictive power when combined with the core proteins (red asterisks). (C) Predictive power (AUC) of the top-10 proteins from the three different methods: RF–random forest, Ttest–t-tests, and Corr–correlation compared with the optimal panel of 9 proteins (AUC_Optimal) representing the 5 core proteins and the 4 additional proteins identified in Fig 3B).
Table 3.
Top-20 GO terms from pathway enrichment analysis.