Supplemental Information Figure 1 - ASCP

D. Ng et al.

Supplemental Information Figure 1

1

D. Ng et al.

Figure 2 The green line represents the normalized feature importance, sorted from highest to lowest. The blue line is the cumulative sum of the green line and demonstrates 80% of the power is located in the first 1000 features.

FeatureImportanceRFC.png

2

D. Ng et al.

Figure 3 Bins/Features that were found to be most useful in discriminating positive and negative cases

3

D. Ng et al.

Figure 4 Receiver operating curves (ROC) for each cross validation iteration (6 fold in this case). The mean ROC shows an AUC of 0.97.

ROC with folds.png

4

D. Ng et al.

Figure 5a First Principle Component back plotted into 78 projections

5

D. Ng et al.

Figure 5b Second Principle Component back plotted into 78 projections

6

D. Ng et al.

Figure 5c Third Principle Component back plotted into 78 projections

7

D. Ng et al.

Figure 5a Fourth Principle Component back plotted into 78 projections

8

D. Ng et al.

Figure 5a Fifth Principle Component back plotted into 78 projections

9

D. Ng et al.

Figure 6 Sorted eigenvalues derived from the PCA of the full data matrix (144x196,000)

10

D. Ng et al.

11