TABLE 5.

Conditional probabilities for “virulent in mammals” sequences for combinations of sites^a

Nodeb			P(virulence|state)c	Virulence prediction (P ≥ 0.5)
1	2	3
0	0	0	0.09	0
0	0	1	0.70	1
0	1	0	0.25	0
0	1	1	0.94*	1
1	0	0	1*	1
1	0	1	1*	1
1	1	0	1	1
1	1	1	1*	1

^aA sequence with the combination of mutations [“virulent” (1)/“nonvirulent” (0) amino acids] as indicated in columns 1 to 3 has a probability of virulence in mammals as indicated in column 4. If the probability of virulence in mammals is >0.5, then the sequence is classified as ‘virulent’.

^bNode 1, PB1-317/PB2-355; node 2, NS1-92/NS1-228; node 3, HA-102/NS1-195.

^cThe values marked with an asterisk indicate that this particular combination of mutations was not observed in the training data sets (or 69 sequence data set), so the probability of virulence in mammals was estimated from the individual values in Table 4 (see Materials and Methods). All probabilities were calculated by averaging over the 10 training data examples.