Table 2. Prediction the database on Pupdb 10-fold with AAIndex BLOSUM62 PCA.

Subset	Sn(%)	Sp(%)	Acc(%)	Mcc	AUC
1	99.81	80.48	90.15	0.8183	0.8127
2	95.57	80.11	87.84	0.7660	0.8157
3	99.27	76.79	88.03	0.7806	0.8287
4	99.72	83.54	91.63	0.8437	0.7903
5	99.34	81.59	90.46	0.8224	0.8107
6	99.43	85.75	92.59	0.8599	0.8102
7	99.52	77.53	88.52	0.7898	0.8167
8	99.62	76.42	88.02	0.7817	0.8397
9	99.75	80.47	90.11	0.8175	0.8576
10	87.57	80.06	83.82	0.6782	0.8162
Average	97.96	80.28	89.12	0.7949	0.8199

The first column records sensitivity of these ten subsets of the Pupdb. The second column records the specialty of such subsets. And the 3th and 4th column record the accuracy and the Markovian correlation coefficient, AUC of these data, respectively.