Skip to main content
. 2018 Jun 27;34(13):i254–i262. doi: 10.1093/bioinformatics/bty275

Fig. 3.

Fig. 3.

The average elapsed time of five trials to predict families of 1000 protein sequences for each method. Three independent models were built for each method, which were trained to predict one of 1074, 1796 and 2892 families by using COG-500-1074, COG-250 1796 and COG-100-2892 dataset respectively