A Coding Theoretical Approach to Predict Sequence Changes in
H5N1 Influenza A Virus Hemagglutinin
Keiko Sato, Toshihide Hara and Masanori Ohya
Department of Information Science, Tokyo University of Science, Noda, Japan
Keywords: Receptor Binding Domain, H5N1, Code Structure, Sequence Prediction.
Abstract: The changes in the receptor binding domain of influenza A virus hemagglutinin lead to the appearance of
new viral strains that evade the immune system. To prepare the future emergence of potentially dangerous
outbreaks caused by divergent influenza strains including human-adapted H5N1 strains, it is imperative that
we understand the rule stored in the sequence of the receptor binding domain. Information of life is stored as
a sequence of nucleotides, and the sequence composed of four nucleotides seems to be a code. It is
important to determine the code structure of the sequences. Once we know the code structure, we can make
use of mathematical results concerning coding theory for research in life science. In this study, we applied
various codes in coding theory to sequence analysis of the 220 loop in the receptor binding domain of H1,
H3, H5 and H7 subtype viruses isolated from humans. Sequence diversity in the 220 loop has been observed
even within the same hemagglutinin subtype. However, we found that the code structure of the 220 loop
from the same subtype remains unchanged. Our results indicate that the sequences at the 220 loop have the
structure of subtype-specific codes. In addition, in view of these finding, we predicted possible amino acid
changes in the 220 loop of H5N1 strains that will emerge in the future. Our method will facilitate
understanding of the evolutionary patterns of influenza A viruses, and further help the development of new
antiviral drugs and vaccines.
1 INTRODUCTION
Influenza A viruses have eight pieces of segmented
RNA, which encode 11 proteins (Olsen et al., 2006).
The antigenic properties in the two viral surface
proteins, hemagglutinin and neuraminidase, are used
to classify influenza A viruses into different
subtypes. Currently Influenza A viruses circulating
among humans are the H1N1 and H3N2 subtypes.
Although other subtypes such as H5N1 and H7N9
have not yet gained the ability to spread efficiently
from person to person, these virus subtypes have
occasionally infected humans.
High-pathogenicity avian H5N1 influenza
viruses exhibiting high lethality continue to pose
threats to our lives since their emergence in China in
1996. According to the World Health Organization
(WHO), there have been 826 human cases with
H5N1 influenza infection since 2003, and
approximately 53% of the cases have died (as of
March 31, 2015). Despite the high mortality, H5N1
viruses have not yet gained the ability to spread
efficiently from person to person. However, the
outbreaks of H5N1 have been reported among
domestic poultry and wild birds in many countries
(Durand et al., 2015; Pfeiffer et al., 2011;
Yamamoto et al., 2011). In addition, recent studies
reported that a reassortant influenza virus containing
a hemagglutinin protein from an H5N1 virus with
four mutations can be transmitted between ferrets
(Imai et al., 2012). The viral surface protein,
hemagglutinin mediates binding of the virus to target
cells via the host cell receptor, sialic acid (Jiang et
al., 2012; Rumschlag-Booms and Rong, 2013). The
hemagglutinin of avian influenza viruses
preferentially binds sialic acid receptors (α2,3-SA)
on epithelial cells in the intestinal tract of birds and
in the lower respiratory tract of humans, whereas the
hemagglutinin of human influenza viruses
preferentially binds another type of sialic acid (α2,6-
SA) (Schrauwen and Fouchier, 2014; Yen and
Peiris, 2012). The receptor binding domain (RBD)
of hemagglutinin, situated at the outer surface on top
of the viral spike, is composed of three major
structural elements: a 130-loop (residues 134-138), a
190-helix (residues 188-190), and a 220-loop
Sato K., Hara T. and Ohya M.
A Coding Theoretical Approach to Predict Sequence Changes in H5N1 Influenza A Virus Hemagglutinin.
DOI: 10.5220/0005659701590167
In Proceedings of the 9th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOINFORMATICS 2016), pages 159-167
ISBN: 978-989-758-170-0
Copyright
c
2016 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved
159
(residues 221-228) based on H3 numbering (Das et
al., 2009; Durand et al., 2015; Jiang et al., 2012;
Stevens et al., 2006). It is considered that the
mutations in the RBD could affect the receptor
binding avidity and specificity of hemagglutinin
(Chen et al., 2011; de Vries et al., 2013; de Vries et
al., 2014; Schrauwen and Fouchier, 2014). The RBD
is the primary target of neutralizing antibodies,
which are induced by virus infection or by
vaccination with specific antigen (Bright et al.,
2003; Chen et al., 2011; Jiang et al., 2012; Khurana
et al., 2011; McCullough et al. 2012). However, the
mutations in the RBD lead to change in viral
immunogenicity and antigenicity (Chen et al., 2011;
Xu et al., 2010). Jiang et al. (2012) state that RBD
plays a critical role in the elucidation of antiviral
immune response and protective immunity.
McCullough et al. (2012) also state that a better
understanding of mutations in the RBD may be
useful in vaccine and drug design effort. To prepare
the future emergence of potentially dangerous
outbreaks caused by divergent influenza strains
including human-adapted H5N1 strains, it is
imperative that we understand the rule stored in the
sequence of the RBD.
Information of life is stored as a code composed
of four nucleotides: adenine (A), cytosine (C),
guanine (G), and thymine (T). Therefore, we can
consider that the DNA or gene in each organism is a
code showing its inherent structure. In protein
coding region, each group of three consecutive
nucleotides is called a codon, and each codon
corresponds to one amino acid. The total number of
three nucleotide groups is the third power of 4,
which means we have 64 codons. However, only 20
proteinogenic amino acids exist in nature. Moreover,
it is supposed that the third nucleotide for a codon
will not play an essential role in making of an amino
acid. This shows that a gene has redundancy to
correct errors to some extent. In other words, it has a
structure that is similar to one of an error-
correcting/detecting code for the transmission of
information. In life-science research, it is important
to determine the code structure of the target gene.
Once we know the code structure, we can make use
of mathematical results concerning coding theory for
research in life science. How can the RBD
sequences of influenza A viruses be discussed using
coding theory? The present study was conducted to
find out the code structure of the 220 loop of
influenza A viruses, and to predict sequence changes
in the 220 loop of H5N1 virus.
2 METHODS
2.1 Sequence Data
We applied artificial codes in coding theory to
sequence analysis of the 220 loop in the H1, H3, H5
and H7 RBD. All full-length amino acid and
nucleotide sequences of hemaggulutinin from
influenza A H1, H3, H5, and H7 subtypes were
downloaded from the Influenza Research Database
on September 2014. The hemaggulutinin data set
consists of 8,941 human sequences from the H1
subtype between 1918 and 2014, 6,013 human
sequences from H3 subtype between 1968 and 2014,
230 human sequences from the H5 subtype between
1997 and 2013, and 51 human sequences from H7
subtype between 1996 and 2014. The sequences
were aligned using MAFFT (Katoh and Toh, 2008)
which can quickly process a large dataset.
2.2 Sequence Analysis of the 220 Loop
by Coding Theory
We explain how to encode the nucleotide sequence
of the 220 loop to detect the code structure. The
method for applying artificial codes to sequence
analysis has been described in detail previously
(Ohya and Sato, 2000; Sato et al., 2013)
. Since the
Galois Field GF(4) consists of four elements, 0, 1,
and
such that
++1=0 , the four
nucleotides can be expressed in
each of four elements.
There are a total of 24 (= 4!) different possible
combinations to map the four nucleotides to the four
elements in GF(4).
First, an important part of the nucleotide
sequence of the 220 loop from an influenza strain,
namely the nucleotide sequence excluding the third
nucleotide of each codon, is transformed into the
information sequence which consists of the elements
of GF(4). Next, the information sequence is grouped
into blocks and then encoded into code words of an
error-correcting/detecting code C. The total length of
such a code (code word length) is multiples of 3 and
the length of the information symbols (information
block length) is multiples of 2. The check symbols
in each code word are placed into the corresponding
position of the third nucleotide of codon. Then, the
encoded sequence, which consists of the set of the
code words, is written back to nucleotide sequence.
We call it the encoded nucleotide sequence. After
that, the encoded nucleotide sequence is converted
into amino acid sequence. We call it the encoded
amino acid sequence. Finally, the degree of
similarity between the amino acid sequence of the
BIOINFORMATICS 2016 - 7th International Conference on Bioinformatics Models, Methods and Algorithms
160
220 loop from the influenza strain and the encoded
amino acid sequence described above is computed.
We think that if the amino acid sequence of the 220
loop is identical to the encoded amino acid sequence
generated by the code C, i.e. the similarity is 100%,
then the nucleotide sequence of the 220 loop has the
structure of the code C. Therefore, it is possible to
find the code structure of the 220 loop by computing
the degree of similarity for various artificial codes.
Artificial codes used for our study are the so-called
linear codes, cyclic codes, Bose-Chaudhuri-
Hocquenghem (BCH) codes, self-orthogonal codes
and Iwadare codes. Practically, we used 95 types of
codes including differences in generator polynomial.
Let
( = 1, 2, ⋯ , 230) be 230 amino acid
sequences of the 220 loop from the H5 subtype. As
described above, we encode the 230 nucleotide
sequences of the 220 loop in a code C, and then get
the encoded amino acid sequences
( =
1, 2, ⋯ , 230). Because the 220 loop is composed of
8 amino acid residues, a degree to measure the
similarity between
and
is denoted by rate of
coincidence (RC) as follows:
RC(
,
) = 1 − 8
(
0≤RC(
,
)≤1
)⁄
,
where is the numbers of sites for which two amino
acid sequences differ from each other. RC(
,
)=1
means that the similarity between
and
is 100%.
If all of the 230 amino acid sequences of the 220
loop from H5 subtype are identical to the encoded
amino acid sequences generated by the code C, i.e.
∑
RC(
,
)
230
⁄
=1, then 100% of the 220
loop nucleotide sequences have the structure of the
code C.
(A) (24, 16) cyclic code with
(
)
=
+
+
+
+
+
+
+
+
(B) (12, 8) cyclic code with
(
)
=
+
subtype subtype
(C) (24, 16) cyclic code with G(x)=x^8+1
D) Self-orthogonal code of R=2/3 with
G_3^1 (D)=D^2+1,G_3^2 (D)=D+1
subtype subtype
Figure 1: Percentage of the 220 loop nucleotide sequences with the structure of the indicated codes for H1, H3, H5, and H7
subtypes. The percentage was calculated to the second decimal place. is the generator polynomial of each code. For
figures (A) and (B), the correspondence between the four nucleotides and the elements in GF(4) was given as A → 0, C
→1, T → and G →
. For figures (C) and (D), C → 0, A →1, G → and T →
.
A Coding Theoretical Approach to Predict Sequence Changes in H5N1 Influenza A Virus Hemagglutinin
161
By using 95 types of codes for each case of the
24 representations of the four nucleotides in the
elements of GF(4), we tried to find the code
structure of the 220 loop in each of the H1, H3, H5
and H7 subtype viruses in this way.
Once we found the code structure for the 200
loop of influenza A virus by using various artificial
codes, we can apply this results to the prediction of
amino acid residues in the 220 loop of influenza
strains that will emerge in the future. The 220 loop is
composed of 8 amino acid residues (24 nucleotides).
The 220 loop of the H5N1 viruses isolated since
1997 showed nucleotide changes in 5 positions (the
first at codons 221, 222 and 223, and the second at
codons 226 and 227) out of 16 positions excluding
the third nucleotide position from each of the 8
codons. Therefore we consider the 5 positions as
variable positions, while the remaining 11 positions
as no variable positions. Given the possibility of any
one of four nucleotides at each of the 5 positions, the
information sequence is composed of 16 nucleotides
as follows, where N stands for any one of four
nucleotides: NCNANTAAGGCNANGG. In other
words, as for the information sequence of length 16,
1,024 (= 4
5
× 1
11
) patterns are made through
combination of these 16 positions. To predict
possible amino acid changes in the 220 loop of
H5N1 influenza hemagglutinin, each of these
information sequences was encoded using the
encoding scheme of the code characterizing the 220
loop sequences from H5N1 viruses.
3 RESULTS
3.1 The Code Structure of the 220
Loop of Influenza A Viruses
Figure 1 shows the percentage of the 220 loop
nucleotide sequences with the structure of the
indicated codes for their respective subtypes.
Interestingly, more than 95% (8,504/8,941) of the
220 loop nucleotide sequences of the H1 subtype
that infected humans between 1918 and 2014 had
the structure of the (24, 16) cyclic code with the
generator polynomial
(
)
=
+
+
+
+
+
+
+
+
(Figure 1(A)).
Almost all of the 220 loop nucleotide sequences
from other subtypes (H3, H5 and H7) did not have
that structure. For the H3 subtype that infected
humans between 1968 and 2014, more than 99%
(5,956/6,013) of the 220 loop nucleotide sequences
had the structure of the (12, 8) cyclic code with the
generator polynomial
(
)
=
+1 (Figure 1(B)).
Table 1: Possible amino acid changes in the 220 loop of
H5N1 influenza A strains that will emerge in the future.
Residues 221-228
TEMNGQNG SEVKGLNG PKLNGQNG
TEMNGQSG SEVKGLTG PKLNGQSG
TEMNGRTG SEVNGRNG PKLKGLIG
TEMKGPIG SEVNGRSG PKLKGLTG
TQINGQNG SEVKGPSG PKLNGRNG
TQINGQSG SEVKGPTG PKLNGRSG
TQIKGLIG SELNGQNG PKLKGPIG
TQIKGLTG SELNGQSG PKLKGPTG
TQIKGPNG SELKGLTG PKVNGQNG
TQIKGPIG SELNGRNG PKVNGQSG
TQIKGPTG SELNGRSG PKVKGLIG
TQVKGLIG SELKGPTG PKVKGLTG
TQVKGLTG SQINGHIG PKVNGRNG
TQVKGPIG AKINGQNG PKVNGRSG
TQVKGPTG AKINGQSG PKVKGPIG
TQLKGLIG AEMNGHIG PKVKGPTG
TQLKGLTG AEMNGHTG PEMNGQNG
TQLKGPIG AEMKGLNG PEMNGHIG
TQLKGPTG AEMKGLIG PEMNGQSG
SKINGQNG AEMKGLSG PEMNGHTG
SKINGQSG AEMNGRIG PEMKGLNG
SKIKGLTG AEMNGRTG PEMKGLSG
SKINGRNG AEMKGPIG PEMNGRNG
SKINGRSG AEMKGPTG PEMNGRSG
SKIKGPTG AQVKGLIG PEMKGPNG
SKVNGQNG AQVKGLTG PEMKGPSG
SKVNGQSG AQVKGPIG PEVNGQNG
SKVKGLSG AQVKGPTG PEVNGQSG
SKVKGLTG AQLKGLIG PEVKGLIG
SKVNGRNG AQLKGLTG PEVKGLTG
SKVNGRSG AQLKGPIG PEVNGRNG
SKVKGPTG AQLKGPTG PEVNGRSG
SKLNGQNG PKINGQNG PEVKGPIG
SKLNGQSG PKINGHIG PEVKGPTG
SKLKGLTG PKINGQSG PELNGQNG
SKLNGRNG PKIKGLNG PELNGQSG
SKLNGRSG PKIKGLIG PELKGLIG
SKLKGPTG PKIKGLSG PELKGLTG
SEMNGQNG PKIKGLTG PELNGRNG
SEMNGQSG PKINGRNG PELNGRSG
SEMKGPNG PKINGRSG PELKGPIG
SEVNGQNG PKIKGPIG PELKGPTG
SEVNGQSG PKIKGPTG
Those from other subtypes (H1, H5 and H7) did
not have that structure. In addition, we found the
BIOINFORMATICS 2016 - 7th International Conference on Bioinformatics Models, Methods and Algorithms
162
code structure characterizing the 220 loop sequences
from the H5 and H7 subtypes, respectively. All
(230/230) of the nucleotide sequences of the H5
subtype that infected humans between 1997 and
2013 had the structure of the (24, 16) cyclic code
with the generator polynomial
(
)
=
+1
(Figure 1(C)). For the H7 subtype that infected
humans between 1996 and 2014, approximately 98%
(50/51) of the nucleotide sequences had the structure
of the self-orthogonal code of information rate
R=2/3 with the generator polynomial
(
)
=
+
1,
(
)
=+1 (Figure 1(D)). The amino acid
sequences of the 220 loop are diverse even within
the same subtype (Tables S1-S4). However,
surprisingly, the code structure of the 220 loop from
the same subtype remains unchanged.
3.2 Future Sequence Changes in H5N1
220 Loop
We found the mutation rules for the 200 loop of
influenza A virus hemagglutinin by using various
artificial codes in information transmission. As
became clear above, the 220loop human sequences
from H5N1 strains have preserved the structure of a
specific code since the emergence of H5N1 in
humans in 1997. In this study of predicting
sequences, we used 95 types of codes including
differences in generator polynomials on the
condition that C, A, G and T of nucleotides
correspond to 0, 1, and
of Galois Field GF(4),
respectively. Every 220 loop amino acid sequence
belonging to the H5 subtype was identical to the
encoded amino acid sequence generated by the (24,
16) cyclic code with generator polynomial
(
)
=
+1 (the similarity is 100%) and was not identical
to that generated by any of different 65 types of
codes (the similarity is 0%).
Table 1 shows possible amino acid changes in
the 220 loop of H5N1 influenza strains that will
emerge in the future. These are composed of 128
sequences out of the 1,024 encoded amino acid
sequences generated by the (24, 16) cyclic code with
generator polynomial
(
)
=
+1, the rest of
which were removed because of overlap with the
encoded amino acid sequences generated by the 65
types of codes. The possible changes we predicted
are based on the assumption that although sequence
diversity in the 220 loop of H5N1 hemagglutinin
will be observed even from now on, the code
structure will probably not change.
4 CONCLUSIONS
Influenza A H1 and H3 subtypes, which have
circulated among humans for nearly 100 years since
the pandemic of 1918 and for nearly 50 years since
the pandemic of 1968 respectively, continue to
change by accumulation of mutations in the
hemagglutinin. Similarly, other subtypes such as H5
and H7, which have occasionally caused human
infections, change by mutations in the
hemagglutinin. These changes, particularly the
changes in the RBD of the hemagglutinin, lead to
the appearance of new viral strains that evade the
immune system. Therefore, it is imperative for us to
understand the mutational patterns in the RBD.
Sequence diversity in the 220 loop of the RBD, has
been observed among different hemagglutinin
subtypes, or even within the same subtype.
However, the code structure of the 220 loop from
the same subtype remains unchanged. Our results
indicate that the sequences at the 220 loop have the
structure of subtype-specific codes. The first goal of
this study was to find out the code structure of the
220 loop of influenza A viruses. We fortunately
found the rules of mutations for the loop by using
various codes in information transmission. These
findings may be very helpful in predicting sequence
changes in the 220 loop and may provide clues to
the decision of vaccine strain and the development
of new antiviral drugs. The 220 loop of the RBD is
definitely an attractive target for developing antiviral
drugs.
The second goal of this study was to predict
sequence changes in the 220 loop of H5N1 virus.
Based on the assumption that the code structure of
the 220 loop from the same subtype will probably
not change even from now on, we predicted possible
amino acid changes in the 220 loop of H5N1
influenza strains that will emerge in the future. We
cannot deny the possibility that a pandemic H5N1
strain transmissible between humans may not
possess the amino acid changes predicted here.
Monitoring the molecular changes in hemagglutinin
is important for the accurate sequence prediction.
However, our method, which determines the code
structure of the 220 loop of influenza A virus
hemagglutinin, will facilitate understanding of the
evolutionary patterns of influenza A viruses, and
further help the development of new antiviral drugs
and vaccines. Through the generation of mutant
viruses possessing hemagglutinin gene with
mutations of the 220 loop predicted in our method
and the examination of the growth and
transmissibility of the mutant viruses in animal
A Coding Theoretical Approach to Predict Sequence Changes in H5N1 Influenza A Virus Hemagglutinin
163
models, suitable vaccine candidates will be selected.
It is expected that the 220 loop-based influenza
vaccines would be effective against divergent
influenza strains, including those that may cause
pandemics in the future.
REFERENCES
Bright, R.-A. et al., 2003, Impact of glycosylation on the
immunogenicity of a DNA-based influenza H5 HA
vaccine. Virology, 308, 270-278.
Chen, M.-W. et al., 2011, Broadly neutralizing DNA
vaccine with specific mutation alters the antigenicity
and sugar-binding activities of influenza
hemagglutinin, Proceedings of the National Academy
of Sciences, 108, 3510–3515.
Das, P. et al., 2009, Free energy simulations reveal a
double mutant avian H5N1 virus hemagglutinin with
altered receptor binding specificity, J. Comput. Chem.,
30, 1654–1663.
de Vries, R.-P. et al., 2013, Evolution of the
hemagglutinin protein of the new pandemic H1N1
influenza virus: maintaining optimal receptor binding
by compensatory substitutions. J. Virol., 87, 13868-
13877.
de Vries, R.-P. et al., 2014, Hemagglutinin receptor
specificity and Structural Analyses of Respiratory
Droplet-transmissible H5N1 Viruses. J. Virol., 88,
768-773.
Durand, L.-O. et al., 2015, Timing of Influenza A(H5N1)
in Poultry and Humans and Seasonal Influenza
Activity Worldwide, 2004-2013, Emerg. Infect. Dis.,
21, 202-208.
Imai, M. et al., 2012, Experimental adaptation of an
influenza H5 HA confers respiratory droplet
transmission to a reassortant H5 HA/H1N1 virus in
ferrets, Nature, 486, 420-428.
Jiang, S. et al., 2012, Receptor-binding domains of spike
proteins of emerging or re-emerging viruses as targets
for development of antiviral vaccines, Emerging
Microbes & Infections, 1, e13.
Katoh, K. and Toh, H., 2008. Recent developments in the
MAFFT multiple sequence alignment program, Brief.
Bioinform., 9, 286–298.
Khurana, S. et al., 2011, Bacterial HA1 vaccine against
pandemic H5N1 influenza virus: evidence of
oligomerization, hemagglutination, and cross-
protective immunity in ferrets, J. Virol., 85, 1246–
1256.
McCullough, C. et al., 2012, Characterization of influenza
hemagglutinin interactions with receptor by NMR,
PloS one, 7, e33958.
Ohya, M. and Sato, K., 2000, Use of information theory to
study genome sequences. Reports on Mathematical
Physics, 46, 419–428.
Olsen, B. et al., 2006, Global patterns of influenza A virus
in wild birds. Science, 312, 384–388.
Pfeiffer, D.-U. et al., 2011, Implications of global and
regional patterns of highly pathogenic avian influenza
virus H5N1 clades for risk management, Vet. J., 190,
309-316.
Rumschlag-Booms, E. and Rong, L., 2013, Influenza a
virus entry: implications in virulence and future
therapeutics, Advances in virology, 2013, 121924.
Sato, K. et al., 2013, The code structure of the p53 DNA-
binding domain and the prognosis of breast cancer
patients. Bioinformatics, 29, 2822–2825.
Schrauwen, E.-J., 2014, Host adaptation and transmission
of influenza A viruses in mammals. Emerg. Microbes
Infect., 3, e9.
Stevens, J. et al., 2006, Structure and receptor specificity
of the hemagglutinin from an H5N1 influenza.
Science, 312, 404-410.
Xu, Q. et al., 2010, Influenza H1N1 A/Solomon
Island/3/06 virus receptor binding specificity
correlates with virus pathogenicity, antigenicity, and
immunogenicity in ferrets. J. Virol., 84, 4936-4945.
Yamamoto, N. et al., 2011, Characterization of a non-
pathogenic H5N1 influenza virus isolated from a
migratory duck flying from Siberia in Hokkaido,
Japan, in October 2009. Virol. J., 8, 65.
Yen, H.-L. and Peiris, J.-S., 2012, Virology: bird flu in
mammals, Nature, 486, 332-333.
BIOINFORMATICS 2016 - 7th International Conference on Bioinformatics Models, Methods and Algorithms
164
APPENDIX
Table S1: Amino acid sequence diversity in the 220 loop of human H1 hemagglutinin.
H1 strain Residues 221-228
AF117241|A/South_Carolina/1/18|H1N1 PKVRDQAG
CY010788|A/WSN/1933_TS61|H1N1 PKVKDQHG
U08904|A/WS/1933|H1N1 PKVRDQPG
DQ508905|A/Wilson_Smith/1933|H1N1 PKVRDQHG
U08903|A/NWS/1933|H1N1 PKVRNQPG
CY040170|A/Puerto_Rico/8_SV14/1934|H1N1 PKVKGQAG
CY146857|A/Puerto_Rico/8_SV40/1934|H1N1 PKVKDQAG
CY147326|A/BH/JY2/1935|H1N1 PKVRDQTG
CY020445|A/Henry/1936|H1N1 PEVRDQAG
CY013271|A/Hickox/1940|H1N1 PKVRGQAG
CY045772|A/Melbourne/1/1946|H1N1 PEVKDQAG
CY077768|A/Netherlands/002P1/1951|H1N1 PKVRNQAG
CY009340|A/Malaysia/54|H1N1 PKVRGQPG
CY008988|A/Denver/1/1957|H1N1 PKVRDQSG
CY125862|A/Kw/1/1957|H1N1 PKVRGQSG
CY021717|A/California/10/1978|H1N1 PKVRGQEG
CY028724|A/California/45/1978|H1N1 PKVRDQEG
CY020173|A/Lackland/7/1978|H1N1 PKVRDQKG
CY017203|A/Memphis/23/1983|H1N1 PKVRNQEG
CY104862|A/TayNguyen/TN182/2006|H1N1 PKVRDQGG
EU100724|A/Solomon_Islands/03/2006|H1N1 PKVRDREG
EU199338|A/Texas/06/2007|H1N1 PKVRBQEG
CY027779|A/Kentucky/UR06_0339/2007|H1N1 PKVREQEG
CY118091|A/Malaysia/1794173/2007|H1N1 LKVRDQEG
EU516017|A/Hawaii/31/2007|H1N1 PKIRDQEG
CY073960|A/Mexico/UASLP_009/2008|H1N1 PKLRDQDG
GU367325|A/Novgorod/01/2009|H1N1 PKVREREG
CY049076|A/Singapore/ON141/2009|H1N1 PKVGDQEG
CY051455|A/Wisconsin/629_S0339/2009|H1N1 TKVRDQEG
CY095906|A/Zhejiang/8/2009|H1N1 PKVRDQER
CY054606|A/Thailand/THB0405/2009|H1N1 PRVRDQEG
CY122835|A/Singapore/GP2242/2009|H1N1 PQVRDQEG
CY075897|A/Blore/NIV1196/2009|H1N1 PKMRGKEG
KC781609|A/California/33/2009|H1N1 PKMRDQEG
CY083399|A/Great_Lakes/WRAIR1664P/2009|H1N1 PKVKEQEG
KC782207|A/South_Carolina/18/2009|H1N1 PKVKDQEG
KC781375|A/Oregon/35/2009|H1N1 HKVRDQEG
CY095955|A/Zhejiang/86/2009|H1N1 PKVRDQEA
CY057254|A/New_York/5186/2009|H1N1 PKVMDQEG
CY069114|A/Madrid/INS296/2009|H1N1 PKVRAQEG
HM581919|A/Iran/15583/2009|H1N1 PKVRDRQG
KF411180|A/Qingdao/FF85/2009|H1N1 PKVRDSEG
CY067632|A/Qingdao/66/2010|H1N1 PKVRDQEW
CY092952|A/Chile/15/2010|H1N1 PKLRDQEG
CY079544|A/Switzerland/5165/2010|H1N1 PKVREQAG
JQ796827|A/Zhejiang/HZ19/2011|H1N1 PIVRDQEG
JQ396238|A/Kenya/145/2011|H1N1 PKGRDQEG
KF451900|A/Kenya/262/2013|H1N1 PKVKEQDG
KM013710|A/Shiraz/87/2013|H1N1 PKVRDHEG
KJ645782|A/Gainesville/03/2014|H1N1 PKVRSQEG
All groups of identical sequences in the 220 loop sequences from H1 subtype that infected humans between 1918 and 2014 were
represented by the oldest sequence in the group.
A Coding Theoretical Approach to Predict Sequence Changes in H5N1 Influenza A Virus Hemagglutinin
165
Table S2: Amino acid sequence diversity in the 220 loop of human H3 hemagglutinin.
H3 strain Residues 221-228
CY011120|A/Northern_Territory/60/1968|H3N2 PWVRGLSS
V01103|A/NT/60/68/29c|H3N2 PWVRGQSS
AB284320|A/Aichi/2/1968|H3N2 PWVGGLSS
CY033529|A/Hong_Kong/1_9_MA21_3/1968|H3N2 PWIRGLSS
CY112249|A/Hong_Kong/1/1968|H3N2 PWVRGMSS
CY112297|A/Bilthoven/6022/1972|H3N2 PWVRGPSS
CY113957|A/Akita/4/1993|H3N2 PWVRGQPS
CY113981|A/Lyon/672/1993|H3N2 PWVRGLPS
CY114149|A/Hong_Kong/56/1994|H3N2 PWVRGISS
CY118426|A/Malaysia/07831/1995|H3N2 PWVRGVSS
CY009676|A/New_York/576/1997|H3N2 PWIRGVSS
CY121424|A/California/32/1999|H3N2 HWVRGVSS
CY001397|A/New_York/156/2000|H3N2 PWERGVSS
EU856922|A/Hong_Kong/CUHK22072/2000|H3N2 PWVRDVSS
EU856918|A/Hong_Kong/CUHK21932/2001|H3N2 PWIRDVSS
EU856946|A/Hong_Kong/CUHK24749/2001|H3N2 PRVRDVSS
DQ415319|A/TW/872/02|H3N2 HRVRDVSS
CY112933|A/Fujian/411/2002|H3N2 PRVRGVSS
CY003096|A/New_York/403/2002|H3N2 PWGRGVSS
CY007843|A/Canterbury/14/2002|H3N2 PWARGVSS
EU857019|A/Hong_Kong/CUHK50200/2002|H3N2 PRIRDVSS
EU103747|A/Denmark/87/2003|H3N2 PRVRDVPS
EF568926|A/Thailand/Siriraj_02/2003|H3N2 PRVRDIPS
AY531033|A/Wyoming/3/03|H3N2 PRVRDISS
EU857094|A/Hong_Kong/CUHK83422/2003|H3N2 LRVRDVPS
DQ249261|A/Taiwan/30005/2004|H3N2 PRVRHIPS
CY105310|A/HaNoi/HN30147/2004|H3N2 TRVRDVPS
CY013517|A/Wellington/58/2004|H3N2 SRVRDIPS
CY002064|A/New_York/392/2004|H3N2 PRIRDVPS
CY163648|A/Wisconsin/67/2005|H3N2 PRIRNIPS
EU283414|A/Hiroshima/52/2005|H3N2 PRVRNIPS
CY016595|A/South_Australia/18/2005|H3N2 LRVRDIPS
CY016028|A/Western_Australia/74/2005|H3N2 PRIRDIPS
KJ855363|A/Mexico/DIF29/2006|H3N2 LRVRNIPS
CY020357|A/New_York/923/2006|H3N2 PRVRBIPS
EU716471|A/Texas/03/2008|H3N2 HRVRNIPS
CY037543|A/California/UR07_0053/2008|H3N2 PRIKNIPS
FJ179354|A/Minnesota/14/2008|H3N2 PKVRNIPS
GQ385889|A/New_Hampshire/01/2009|H3N2 PRVREIPS
CY050125|A/Qingdao/1329/2009|H3N2 PRVGNIPS
CY091837|A/Guangdong/322/2010|H3N2 TRVRNIPS
JX946754|A/Qingdao/FF184/2010|H3N2 PRLRNIPS
KC882891|A/District_Of_Columbia/02/2010|H3N2 ARVRNIPS
KC882953|A/Minnesota/04/2011|H3N2 SRVRNIPS
CY162984|A/Peru/PER345/2011|H3N2 PRVRNVPS
KC892741|A/New_Jersey/08/2011|H3N2 LRIRNIPS
KC892638|A/California/34/2011|H3N2 PRIRBIPS
KJ942608|A/Hawaii/22/2012|H3N2 PRIRNSPS
KF598718|A/British_Columbia/004/2012|H3N2 HRIRNIPS
KC892959|A/Hawaii/02/2012|H3N2 TRIRNIPS
CY134996|A/Texas/JMM_37/2012|H3N2 PRIRNVPS
KF789696|A/Maine/05/2012|H3N2 PRIRNNPS
KF790228|A/Hawaii/30/2012|H3N2 SRIRNIPS
CY141264|A/Texas/3249/2013|H3N2 PRIRSIPS
KF789872|A/Hawaii/02/2013|H3N2 LRIRDIPS
KM064043|A/Texas/14/2014|H3N2 HRIRDIPS
All groups of identical sequences in the 220 loop sequences from H3 subtype that infected humans between 1968 and 2014 were
represented by the oldest sequence in the group.
BIOINFORMATICS 2016 - 7th International Conference on Bioinformatics Models, Methods and Algorithms
166
Table S3: Amino acid sequence diversity in the 220 loop of human H5 hemagglutinin.
H5 strain Residues 221-228
GU052142|A/Hong_Kong/485/1997|H5N1 PKVNGQSG
GU052089|A/Hong_Kong/378.1/2001|H5N1 SKVNGQSG
AB212054|A/Hong_Kong/213/2003|H5N1 SKVNGQNG
EF107522|A/Thailand/1_KAN_1A_/2004|H5N1 SEVNGQSG
EF456802|A/Viet_Nam/JPHN30321/2005|H5N1 SKINGQSG
DQ371929|A/Anhui/2/2005|H5N1 SKVNGRSG
KF918470|A/Cambodia/X0810301/2013|H5N1 SKVKGLSG
All groups of identical sequences in the 220 loop sequences from H5 subtype that infected humans between 1997 and 2013 were
represented by the oldest sequence in the group.
Table S4: Amino acid sequence diversity in the 220 loop of human H7 hemagglutinin.
H7 strain Residues 221-228
GU053110|A/England/AV877/1996|H7N7 PQVNGQSG
CY181569|A/Anhui/DEWH72_08/2013|H7N9 PQVNGLSG
KF018039|A/Taiwan/1/2013|H7N9 PQVNGPSG
KC853766|A/Hangzhou/1/2013|H7N9 PQVNGISG
KF609511|A/Shanghai/JS01/2013|H7N9 TQVNGQSG
All groups of identical sequences in the 220 loop sequences from H7 subtype that infected humans between 1996 and 2014 were
represented by the oldest sequence in the group.
A Coding Theoretical Approach to Predict Sequence Changes in H5N1 Influenza A Virus Hemagglutinin
167
