Skip to main content
. 2019 Apr 4;20(Suppl 2):193. doi: 10.1186/s12864-019-5489-4

Table 3.

Performance of different methods in terms of auPRC scores

Methods H3K4me1 H3K4me3 H3K27me3 H3K36me3 H3K9me3 H3K9ac H3K27ac
DeepHistone (Standard) 0.8116 (0.0751) 0.8432 (0.0801) 0.6655 (0.1032) 0.7551 (0.0688) 0.6779 (0.1329) 0.7271 (0.1044) 0.7714 (0.0822)
DeepHistone (DNA-only) 0.7429 (0.1268) 0.8208 (0.1348) 0.6543 (0.1382) 0.7493 (0.0754) 0.6427 (0.1424) 0.6888 (0.1835) 0.6990 (0.1535)
DeepHistone (DNase-only) 0.6664 (0.0985) 0.5962 (0.0549) 0.3219 (0.1459) 0.4043 (0.0776) 0.3489 (0.1401) 0.5701 (0.1465) 0.6673 (0.1179)
DeepSEA 0.6087 (0.0817) 0.7371 (0.0566) 0.4404 (0.1355) 0.6164 (0.0647) 0.5554 (0.1277) 0.5629 (0.1536) 0.5340 (0.1027)
DanQ 0.5805 (0.0783) 0.7303 (0.0737) 0.4233 (0.1046) 0.6178 (0.0756) 0.5420 (0.1379) 0.5502 (0.1034) 0.5015 (0.0859)
gkm-SVM 0.4237 (0.0722) 0.5858 (0.0981) 0.2774 (0.0945) 0.4069 (0.0898) 0.4221 (0.1208) 0.3889 (0.1187) 0.3340 (0.0700)

Numbers in a cell are the mean (left) and standard deviation (right) of auPRC scores over the 15 epigenomes