Figure 3
Figure 3. Statistical extraction of candidate therapeutic targets. The 14 064 nonredundant peptides detected were subjected to a 4-group Kruskal-Wallis test (ND, AC, ATL, and HAM), resulting in identification of 1170 first candidates (P < .01). ND, normal donors; AC, asymptomatic carriers. (A) Next, the Expressionist ranking method further narrowed down the candidates to 91 peptides based on SVM-REF so that the misclassification rate in the cross-validation test became minimum, 7.8%. (B) The predicted classification result by leave-one-out cross-validation test. The 27 out of 29 cases were successfully classified into the true classes. (C) The three-dimensional plot shows the additional assessment for the classification power of 91 classifiers by principal component analysis. Comp. 1 to 3 indicate principal components 1 to 3.

Statistical extraction of candidate therapeutic targets. The 14 064 nonredundant peptides detected were subjected to a 4-group Kruskal-Wallis test (ND, AC, ATL, and HAM), resulting in identification of 1170 first candidates (P < .01). ND, normal donors; AC, asymptomatic carriers. (A) Next, the Expressionist ranking method further narrowed down the candidates to 91 peptides based on SVM-REF so that the misclassification rate in the cross-validation test became minimum, 7.8%. (B) The predicted classification result by leave-one-out cross-validation test. The 27 out of 29 cases were successfully classified into the true classes. (C) The three-dimensional plot shows the additional assessment for the classification power of 91 classifiers by principal component analysis. Comp. 1 to 3 indicate principal components 1 to 3.

Close Modal

or Create an Account

Close Modal
Close Modal