highly significantp-values<0.001
The differences in both cases are highly significant (p-values<0.001), and it is evident that PRODIV-TMHMM is highly unreliable for discrimination purposes, a fact noticed already in [ 13 ], followed by MEMSAT and UMDHMM TMHP , whereas HMM-TM and HMMTOP form a group of intermediate reliability just before the highly reliable for this task methods, TMHMM, Phobius, S-TMHMM and PRO-TMHMM.