Table 3.
Number of leave-one-out mispredictions when classifying an organism into two temperature optimum ranges, hyperthermophile versus non-hyperthermophile (hyper vs. non-hyper) and mesophile versus non-mesophile (meso vs. non-meso) using all tetramers frequencies over the entire vitrual coding strand, based on a model from the genomes of the other 194 organisms. The leave-one-out classifications were made for the 195 original genomes (original) and two types of randomly generated control sequences, multinomial (multi) and Markov.
Temperature range | Original | Random control | |
Multi | Markov | ||
Hyper vs. non-hyper | 8 (4.1%) | 16 (8.2%) | 8 (4.1%) |
Meso vs. non-meso | 14 (7.2%) | 59 (30.3%) | 21 (10.8%) |