Tweet
International Journal of Computer Theory and Engineering, Vol. 7, No. 2, April 2015
Examining the Probability of the Critical Mutation of
H5N8 by Comparing with H7N9 and H5N1 Using Apriori Algorithm and Support Vector Machine
Dae Young Kim, Hye-Jun Kim, Junhyeok Bae, and Taeseon Yoon
Abstract?In January, 2014, the outbreak of H5N8 in South
Korea started in one duck farm. The representative outbreak of
H5N8 is to turkey 1983 Ireland and to duck 2010 China.
Obviously, people were concerned about whether this Influenza
A virus is highly pathogenic or human transmissible. In this
research to identify the probability of H5N8?s pathogenic rate,
we will investigate its chance to have Cytokine Storm, a deadly
attribute of Influenza A virus, by seeking similarity in
glycoprotein amino acid sequence with H5N1, which has the
same hemagglutinin subtype, using Support Vector Machine. In
addition, to identify H5N8?s human transmissible possibility, we
will compare the its year-on-year glycoprotein amino acid
mutating trend with H7N9, which was previously known not to
be transmissible to human but mutated to infect human, using
Apriori Algorithm.
Index Terms?Apriori algorithm, H5N8, H5N1, H7N9,
influenza A Virus, support vector machine.
...
V. DISCUSSION
A. Similarity between H5N8 and H5N1
See the amino acid dataset of hemagglutinin and
neuraminidase in H5N8 which was classified by SVM in the
condition of window size 9, where the result is more precise
compared to other window sizes (see Table III). It showed
out that the accuracy of hemagglutinin was higher than
neuraminidase. In other words, the classification of
neuraminidase in H5N8 and H5N1 was better done than
hemagglutinin and this means that the similarity of
hemagglutinin in H5N8 and H5N1 is much higher. The high
similarity of hemagglutinin in H5N1 and H5N8 can be
explained by the same hemagglutinin subtype; H5. But since
the similarity of neuraminidase is low, we cannot conclude
that the H5N8 of 2010 and H5N1 are alike. Therefore, it is
viewed that the H5N1?s feature of cytokine storm will not
take place in H5N8 but this needs further prudent research to
be confirmed.
B. Comparison of the Mutational Attribution of H5N8 and
H7N9
H7N9 was not infectious to humans in 1988 to 2009. But,
as the amino acid sequence of hemagglutinin and
neuraminidase in H7N9 changed dramatically, it became
infectious to humans starting from 2013. Especially, despite
the same host, fowls, the sequence shows a great change.
(See Fig. 3 ~ Fig. 8). As seen in the case of H7N9, the
sufficient condition for the influenza to be infectious to
human is to have a considerable change in the amino acid
sequence even for the same host. However, compared to the
mutation in H7N9, there was only a little change in the
sequence between the 1983 H5N8 and 2010 H5N8. Thus, we
concluded that the possibility of H5N8 to be infectious to
humans is comparatively low.
Examining the Probability of the Critical Mutation of
H5N8 by Comparing with H7N9 and H5N1 Using Apriori Algorithm and Support Vector Machine
Dae Young Kim, Hye-Jun Kim, Junhyeok Bae, and Taeseon Yoon
Abstract?In January, 2014, the outbreak of H5N8 in South
Korea started in one duck farm. The representative outbreak of
H5N8 is to turkey 1983 Ireland and to duck 2010 China.
Obviously, people were concerned about whether this Influenza
A virus is highly pathogenic or human transmissible. In this
research to identify the probability of H5N8?s pathogenic rate,
we will investigate its chance to have Cytokine Storm, a deadly
attribute of Influenza A virus, by seeking similarity in
glycoprotein amino acid sequence with H5N1, which has the
same hemagglutinin subtype, using Support Vector Machine. In
addition, to identify H5N8?s human transmissible possibility, we
will compare the its year-on-year glycoprotein amino acid
mutating trend with H7N9, which was previously known not to
be transmissible to human but mutated to infect human, using
Apriori Algorithm.
Index Terms?Apriori algorithm, H5N8, H5N1, H7N9,
influenza A Virus, support vector machine.
...
V. DISCUSSION
A. Similarity between H5N8 and H5N1
See the amino acid dataset of hemagglutinin and
neuraminidase in H5N8 which was classified by SVM in the
condition of window size 9, where the result is more precise
compared to other window sizes (see Table III). It showed
out that the accuracy of hemagglutinin was higher than
neuraminidase. In other words, the classification of
neuraminidase in H5N8 and H5N1 was better done than
hemagglutinin and this means that the similarity of
hemagglutinin in H5N8 and H5N1 is much higher. The high
similarity of hemagglutinin in H5N1 and H5N8 can be
explained by the same hemagglutinin subtype; H5. But since
the similarity of neuraminidase is low, we cannot conclude
that the H5N8 of 2010 and H5N1 are alike. Therefore, it is
viewed that the H5N1?s feature of cytokine storm will not
take place in H5N8 but this needs further prudent research to
be confirmed.
B. Comparison of the Mutational Attribution of H5N8 and
H7N9
H7N9 was not infectious to humans in 1988 to 2009. But,
as the amino acid sequence of hemagglutinin and
neuraminidase in H7N9 changed dramatically, it became
infectious to humans starting from 2013. Especially, despite
the same host, fowls, the sequence shows a great change.
(See Fig. 3 ~ Fig. 8). As seen in the case of H7N9, the
sufficient condition for the influenza to be infectious to
human is to have a considerable change in the amino acid
sequence even for the same host. However, compared to the
mutation in H7N9, there was only a little change in the
sequence between the 1983 H5N8 and 2010 H5N8. Thus, we
concluded that the possibility of H5N8 to be infectious to
humans is comparatively low.
