Gene Prediction in Metagenomic Fragments with Deep Learning

Table 2

The overall accuracy (%) of Orphelia, FragGeneScan, MGC, MetaGUN, Meta-MFSVM, and Meta-MFDL on the 700 independent testing dataset.

Species	Orphelia	Meta-MFSVM	MGC	FragGeneScan	MetaGUN	Meta-MFDL
A. fulgidus	89.49	92.25	92.47	93.14	94.30	94.70
N. pharaonis	84.73	89.87	89.90	91.67	94.40	94.12
M. jannaschii	86.13	91.68	91.69	95.91	95.03	94.65
B. aphidicola	94.16	93.43	93.75	95.60	96.24	96.05
B. pseudomallei	88.19	90.71	90.76	91.53	94.13	94.15
B. subtilis	91.56	92.55	91.78	92.49	93.86	93.89
C. jeikeium	86.03	90.15	90.85	91.32	91.31	91.62
C. tepidum	83.59	88.48	88.43	88.07	90.68	90.66
E. coli	91.50	91.98	92.06	93.45	93.58	93.66
H. pylori	92.88	93.30	93.48	94.98	93.81	94.49
P. aeruginosa	85.05	92.42	92.47	94.42	94.03	93.89
P. marinus	89.16	91.24	91.36	90.82	93.73	93.88
W. endosymbiont	78.66	84.70	84.78	75.67	83.69	87.34
Average	87.78	90.98	91.06	91.47	92.98	93.31