Biologia 69/4: 407—418, 2014Section Cellular and Molecular BiologyDOI: 10.2478/s11756-014-0341-4

Evolution trends of the 2009 pandemic influenza A (H1N1) virusesin different continents from March 2009 to April 2012

Zhao-Hui Qi*, Jun Feng & Chen-Chen Liu

College of Information Science and Technology, Shijiazhuang Tiedao University, Shijiazhuang, Hebei – 050043, People’sRepublic of China; e-mail: zhqi wy2013@163.com

Abstract: The World Health Organization (WHO) announced that the 2009 pandemic influenza A (H1N1) viruses,A/California/07/2009 (H1N1) – like virus, has gone into the post-pandemic period on August 10, 2010. People still havesome concerns the virus would likely mutate and become a new pandemic virus in the future. Here, we use MUSCLEprogram and graphic mapping method to look into the evolutionary characteristics of the 6219 hemagglutinin and 4860neuraminidase full-length sequences from March 2009 to April 2012. The graphic and statistical analyses showed that thenovel pandemic isolates, A/California/07/2009 (H1N1) – like virus, experienced several different times. During the early-pandemic period (03/2009–08/2009), the viruses have spread globally in several clusters and deviated slightly from therecommended vaccine strain, A/California/07/2009. During the pandemic period (09/2009–08/2010), new clusters beganto emerge from Asia and North America, and further deviated from the recommended vaccine strain. During the post-pandemic period (09/2010–08/2011) and the recent period (09/2011–04/2012), the original cluster with the recommendedvaccine isolate, A/California/07/2009, has nearly disappeared. The deviation degree between the new clusters and thevaccine isolate became larger and larger. However, the deviation degree and the deviation speed were low. The WHO didnot choose a new vaccine isolate instead of the original vaccine isolate, A/California/07/2009. Even so, it is necessary tomonitor continuously the 2009 pandemic influenza A (H1N1) viruses.

Key words: post-pandemic period; pandemic H1N1; graphic representation; sequence analysis; evolution.

Abbreviations: 2D, two-dimensional; 3D, three-dimensional; HA, hemagglutinin; NA, neuraminidase; PCA, PrincipleComponents Analysis; WHO, World Health Organization.

Introduction

The pandemic influenza A (H1N1) virus after its emer-gence in the United States and Mexico in March 2009spread around the world within several months. TheWorld Health Organization (WHO) raised the influenzapandemic alert level to phase 6 on 11th June 2009(Fereidouni et al. 2009). The novel A (H1N1) 2009,A/California/07/2009 (H1N1) – like virus, shows astrong ability to broadcast from human to human andspread worldwide (CDC 2009; WHO 2011a). The WHOannounced the global pandemic alert level to phase 6on 11th June 2009 (Qu et al. 2011). The pandemic in-fluenza A (H1N1) virus is a combination of genes fromswine, avian and human influenza viruses (Fraser et al.2009; Garten et al. 2009). It has a high mean evolu-tionary rate for individual genes and the whole genome,varying from 2.34 × 10−3 to 3.67 × 10−3 substitutionsper site per year (Smith et al. 2009). From March 2009to April 2010 the swine-origin influenza virus has re-sulted in about 18,000 deaths around the world (WHO2010b).

The WHO announced the 2009 pandemic influenzaA (H1N1) viruses have gone into the post-pandemic pe-riod on 10th August 2010 (WHO 2010c). Even so, peo-ple still have some concerns the virus may mutate in thefuture. The mutation can result in more easily transmit-table or more pathogenic viruses. Phylogenetic analy-ses based on the early surveillance data have shown theviruses in two major clusters have spread globally andcirculated over time and space since April 2009 (Fer-eidouni et al. 2009). Nelson et al. (2009) analysed 290H1N1pdm isolates sampled globally between 1st Apriland 9th July 2009. The analysis revealed that at least 7phylogenetically distinct viral clades have spread glob-ally and co-circulated in localities. Then according tothe 7 clades suggested by Nelson et al. (2009), subse-quent study discussed the genetic characterization ofthe influenza A pandemic (H1N1) 2009 virus isolatesfrom India (Potdar et al. 2010). Recently, both phylo-genetic and cluster analyses further confirm that thegene exchange of the influenza A pandemic (H1N1)2009 virus takes place between viruses originated fromdifferent species (Arunachalam et al. 2012).

* Corresponding author

c©2014 Institute of Molecular Biology, Slovak Academy of Sciences

408 Z. H. Qi et al.

Table 1. Summary information for HA and NA about human influenza A H1N1 from March 2009 to April 2012 (the “year/month/*”is “year/month/” or “year/month/day”).

Genes Continents Sequences with full length and Genes Continents Sequences with full length andtime tag “year/month/*” time tag “year/month/*”

HA Africa 84 NA Africa 62Asia 1712 Asia 1227Europe 1665 Europe 1092North America 2244 North America 2125Oceania 178 Oceania 171South America 336 South America 183

Table 2. Binary codes of 20 amino acids (Xiao et al. 2005, 2006).

Amino acid P L Q H R S F Y W C

Binary code 00001 00011 00100 00101 00110 01001 01011 01100 01110 01111

Amino acid T I M K N A V D E G

Binary code 10000 10010 10011 10100 10101 11001 11010 11100 11101 11110

Since the outbreak of the 2009 pandemic in-fluenza A (H1N1) virus, the representative strain,A/California/07/2009 (H1N1) – like virus, has beenrecommended as the vaccine strain (WHO, 2010a,2010d, 2011b,c, 2012). The vaccine based on this strainwill be valid if the existing mutations in the present in-fluenza A (H1N1) viruses do not result in a rapid propa-gation or an unknown pathogenicity. Here, we will lookinto the evolution trends of the 2009 pandemic influenzaA (H1N1) viruses in different continents from March2009 to April 2012. The purpose of the present studywas to find out whether the isolates during the post-pandemic period have some distinct evolution diver-sification from the isolates during the early-pandemicperiod.

Material and methods

MaterialsThe hemagglutinin (HA) and neuraminidase (NA) gene se-quences of the 2009 pandemic influenza A (H1N1) virus fromMarch 2009 to April 2012 were downloaded from the NCBIInfluenza Virus Resource (http://www.ncbi.nlm.nih.gov/genomes/FLU/SwineFlu.html). Two datasets – one for theHA gene and the second one for the NA gene – were thuscreated. Then each dataset was further divided into 6 sub-datasets according to different continents: Africa, Asia, Eu-rope, North America, Oceania and South America. Thesesub-datasets can reflect the virus evolution related to differ-ent continents from March 2009 to April 2012. For the down-loaded protein sequences of HA and NA, many of them werenot full-length sequences. To avoid the biases from the se-quences without full-length, the incomplete sequences wereremoved from the datasets. Finally 12 datasets with full-length sequences were obtained. In addition, all sequencesin datasets have time record, such as year//, year/month/and year/month/day. The time record “year/month/” or“year/month/day” marks the exact sequencing time of thesequence. These detailed time records can be helpful to knowmore details of evolution history of influenza viruses for the

months studied. To avoid the biases from the rough timerecord “year//”, all sequences with time record “year//”were removed, too. Summary details of HA and NA abouthuman influenza A (H1N1) are shown in Table 1.

For all protein sequences of HA and NA in Table 1, wekeep the most of the available sequences. From the overallview of all sequences, our choices about the sequences withfull-length and exact time record have stochastic character-istics. The stochastic characteristics can avoid the biases inthe analyzed results of evolution history of influenza for theyears.

Binary codes of amino acids in protein sequence and thegraphic mapping of biological dataA protein sequence is a linear sequence of n symbols from afinite alphabet set (A, C, D, E, F, G, H, I, K, L, M, N, P, Q,R, S, T, V, W, Y ), where the parameter n means the lengthof protein sequence. It is difficult to view the character-istics of this linear sequence, especially for a long proteinsequence. In Xiao et al. (2005, 2006), the authors gave somebinary codes of the 20 amino acids (Table 2). The codescan better reflect the chemical and physical properties ofamino acids. By the codes shown in Table 2, a proteinsequence is transformed to a serial of binary characters.For example, the sequence ‘MKNYWQH’ is transformed to‘10011101001010101100011100010000101’.

Graphic mapping of biological data, a powerful tool,has been employed to help people gain a useful insights byan intuitive manner (Randic et al. 2003, 2008; Bielinska-Waz et al., 2007; Qi & Fan 2007; Novic & Randic 2008;Chou & Shen 2009; Bielinska-Waz 2010; Chou, 2010; Yaoet al. 2010;). Here we first used MUSCLE program (Edgar2004) to do a multi-alignment of all analyzed sequences toget the identical residue position of each sequences. Thenall aligned sequences of the same gene were ordered ac-cording to their time records. Based on the binary codes ofamino acids (Table 2), we can map all aligned sequences into0-1 vectors. However, it is impossible to represent graphi-cally all sequences in a high dimensional space. To representgraphically the biological data, we used dimension reductionmethod to map the high dimensional data into low dimen-sional data. This is an important approach to study anti-

Evolution trends of the 2009 pandemic influenza A (H1N1) 409

Fig. 1. The 3D graphic representation models of HA and NA. For HA gene, the first and second principal axes account for 65.40% and5.69% of the total inertia of the 2905-D space, respectively. For NA gene, the first and second principal axes account for 80.36% and3.41% of the total inertia of the 2345-D space, respectively.

genic evolution of influenza besides of phylogenetic recon-struction method. The dimension reduction methods werepresented in Lapedes & Farber (2001) and Smith et al.(2004) and used to study antigenic evolution of influenza.He & Deem (2010) presented another low dimensional clus-tering method to detect the cluster containing an incipientdominant strain for an upcoming influenza season before thestrain becoming dominant.

Based on the binary codes of amino acids (Table 2),we first mapped all aligned sequences into 0-1 vectors. Then,the Principle Components Analysis (PCA; Jackson & Wiley1991) method was used to reduce the multi-dimension vec-tor set to a two-dimensional (2D) space. We can get a new2D vector set. Based on the reduced 2D data, we further de-tected the evolution trends of the 2009 pandemic influenzaA (H1N1) virus in different continents from the outbreak to2012.

Muscle program and PCA methodMUSCLE is a program for creating multiple alignmentsof amino acid or nucleotide sequences (Edgar 2004, 2010).Some default parameters are those that gave the best aver-age benchmark accuracy in tests by author.

PCA (Jackson &Wiley 1991) is a projection method toanalyze dataset and reduce it from high-dimensional spaceto a few hidden variables, while keeping information on itsvariability. Now, PCA and its many expanded methods havebeen successfully applied to resolve many problems (Costaet al. 2009; Xiao et al. 2009). Here, we give a simple descrip-tion of PCA. Assume the mean of sample X = {xi}n

i=1 of

space RD is x̄ = (1/n)n∑

i=1

xi. Write the singular value de-

composition of covariance matrix Σ as Σ = U ∧ UT, whereΣ = E(x − x̄)(x − x̄)T. Matrix U is orthogonal matrix. Di-agonal matrix ∧ is made up of the eigenvalues of Σ, where∧ = diag(λ1, · · · , λD) and λ1 ≥ · · · ≥ λD. The principlecomponent transformation is UT (X − X̄). Then a new dataset Y = {yi}n

i=1 is obtained by Y = UT(X − X̄). The meanand covariance matrix of Y are 0 and diagonal matrix ∧,respectively. Now we ignore the components of lesser signif-icance and leave out some important components. Then thefinal data set will have fewer dimensions than the original.

Results

Graphic representation of isolate clusters from March2009 to April 2012As shown in Table 1, we got 6 HA datasets and 6 NAdatasets from March 2009 to April 2012. We first usedMUSCLE program to do a multi-alignment of all an-alyzed sequences to get the identical residue positionof each sequences. Then all aligned sequences were or-dered according to their time records. According to themapping in Table 2, we converted every sequence to0-1 sequence. We obtained a high-dimensional 0-1 vec-tor set. Then, the PCA method was used to reduce thehigh-dimensional set to a 2D space. Thus a new 2D vec-tor set was created. Then based on the time records ofall aligned sequences, the 2D vector set was expandedalong the third dimension in a three-dimensional (3D)space. Then we got a 3D graphic representation of theanalyzed sequences in the 3D space.For the HA gene, we got a 2905-D 0-1 vector set.

By the PCA method we projected the 2905-D vectorset onto 2D planes spanned by the first and the secondprincipal axes. The first and the second principal axesaccount for 65.40% and 5.69% of the total inertia of the2905-D space, respectively.For the NA gene, we got a 2345-D 0-1 vector set.

Similarly, by the PCA method we projected the 2345-Dvector set onto 2D planes spanned by the first and thesecond principal axes. The first and the second principalaxes account for 80.36% and 3.41% of the total inertiaof the 2345-D space, respectively.Then based on the time records of all sequences,

we got the 3D graphic representation models of HAgene and NA gene, respectively (Fig. 1). It showsthat all isolates are separated into two distinct clus-ters by the first principal axis. This indicates theisolates in the clusters have different evolution char-acteristics. Here, one cluster was called as new iso-late branch, A/California/07/2009 (H1N1) – like virus.

410 Z. H. Qi et al.

Fig. 2. Comparison for the graphic representation of the 7 clades (Nelson et al. 2009) in the background of the global isolates fromMarch 2009 to April 2012.

The other one was named as old isolate branch,A/Brisbane/59/2007 (H1N1) – like virus. In addi-tion, in Figure 1 we marked the typical strains,two recommended vaccine isolates of different peri-ods by WHO, A/Brisbane/59/2007 (WHO, 2009), andA/California/07/2009 (WHO, 2010a).According to the positions of two recommended

vaccine isolates in Figure 1, A/California/07/2009 andA/Brisbane/59/2007, we easily separated the new iso-lates from the old isolates. On inspection, it was foundthe old isolates represented by A/Brisbane/59/2007died off little by little. To avoid the possible biases fromthe old virus population, based on Figure 1 we removedall the old isolates from the 6 HA datasets and 6 NAdatasets. Then we got 5984 HA sequences and 4643 NAsequences about the new viruses.The new isolates by A/California/07/2009 show

more and more scattered characteristics from March2009 to April 2012 (Fig. 1). The scattered characteris-tics mean the genetic drift in virus population becomes

larger little by little. Based on the earlier whole genomeanalysis, Nelson et al. (2009) found out there were atleast 7 phylogenetically distinct clades about the 2009pandemic influenza A (H1N1) viruses. Since then manyresearchers referred to the naming scheme as “globalclades”, e.g. in Potdar et al. (2010). The graphic rep-resentation of the 7 clades (Nelson et al. 2009; Potdaret al. 2010) in the background of the global isolatesfrom March 2009 to April 2012 is given in Figure 2. Itshows that there exist new clusters different from the7 clades. Only the clade 7 is closer to the new clusters.In addition, for HA gene, the clades 2, 4 and 5 are clus-tered into one group. For NA gene, the clade 1 and 2are clustered into one group. The clade 4, 6 and 7 arealso clustered into one big group. One of the possiblereasons could be the Figure 2 includes 6219 HA genesequences and 4860 NA gene sequences from differentcontinents from March 2009 to April 2012. The 7 cladesnamed by Nelson et al. (2009) were only based on thewhole-genome sequences of H1N1pdm isolates collected

Evolution trends of the 2009 pandemic influenza A (H1N1) 411

Fig. 3a. For HA gene, the 2D graphic representation of 20 subsets of 6 continents in the first two stages. There are no virus sampleswith full-length testing data in the 4th stage of Africa, Asia, Oceania, and South America.

from 1st April 2009 to 9th July 2009. At the same time,the typical strain, the recommended vaccine isolate byWHO, A/California/07/2009, was marked by the redcircle for comparison (Fig. 2).

Continent-difference evolution trends by Graphic repre-sentation from March 2009 to April 2012The WHO announced that the 2009 pandemic influenzaA (H1N1) viruses have gone into the post-pandemicperiod on 10th August 2010 (WHO 2010c). People stillhave some concerns that the virus would likely mutateand become a new pandemic virus in the future. It has

also been of interest whether or not there are differentevolutionary tendency among different continents.To discover the different evolution characteris-

tics, we divided the period of time from March2009 to April 2012 into four stages. The first stagewas named as the early-pandemic period (03/2009-08/2009). The second stage was called as the pandemicperiod (09/2009-08/2010). The third and forth stageswere regarded as the post-pandemic period (09/2010-08/2011) and the recent period (09/2011-04/2012), re-spectively. Then we further analyzed the evolution-ary trends of the 2009 pandemic influenza A (H1N1)

412 Z. H. Qi et al.

Fig. 3b. For HA gene, the 2D graphic representation of 20 subsets of 6 continents in the last two stages. There are no virus sampleswith full-length testing data in the 4th stage of Africa, Asia, Oceania, and South America.

viruses in different continents from March 2009 to April2012.For HA gene, we first divided the analyzed se-

quences into 20 subsets according to 6 continents and 4stages. There are no virus samples with full-length test-ing data in the 4th stage of Africa, Asia, Oceania andSouth America. To distinguish the evolutionary trendsfrom different continents, for every stage we used themapping method of Table 2 to convert the influenzasamples of different continent to the high-dimensional0-1 vector sets. The PCAmethod was used to reduce ev-ery subset to 2D space. Then we graphically representedthe reduced 2D subsets, as shown in Figure 3a (the firsttwo stages) and Figure 3b (the last two stages). Sim-ilar to the 7 clades according to Nelson et al. (2009),based on Figures 2 and 3, we found that in the early-pandemic period (03/2009–08/2009) the viruses havespread globally in several clusters. During the pandemicperiod (09/2009–08/2010), the viruses further scatteredglobally to several major clusters. Some new clusters inthis stage began to emerge from Asia and North Amer-ica. These newly emerging viruses scattered globally inthe post-pandemic period (09/2010–08/2011). At the

same time, in the third stage, the original cluster withthe vaccine isolate, A/California/07/2009, has nearlydisappeared. In the recent period (09/2011–04/2012),there are no new changes comparing with the virusesin the post-pandemic period (09/2010–08/2011).For NA gene, we also divided the analyzed se-

quences into 20 subsets according to 6 continents and4 stages. There were no virus samples with full-lengthtesting data in the 4th stage of Africa, Asia, Oceaniaand South America. Like the analysis of HA gene, bythe similar method we graphically represented the re-duced 2D subsets, as shown in Figure 4a (the first twostages) and Figure 4b (the last two stages). Based onFigures 2 and 4, for NA gene we can discover somesimilar evolutionary trends for different continents.In the early-pandemic period (03/2009–08/2009), theviruses in several clusters have spread globally. In thepandemic period (09/2009–08/2010) some new clus-ters began to emerge from Asia and North Amer-ica. The newly emerging viruses scattered globally inthe post-pandemic period (09/2010–08/2011). In thethird stage, the original cluster with the vaccine iso-late, A/California/07/2009, has nearly disappeared.

Evolution trends of the 2009 pandemic influenza A (H1N1) 413

Fig. 4a. For NA gene, the 2D graphic representation of 20 subsets of 6 continents in the first two stages. There are no virus sampleswith full-length testing data in the 4th stage of Africa, Asia, Oceania, and South America.

In the recent period (09/2011–04/2012), there are nonew changes comparing with the viruses in the post-pandemic period (09/2010–08/2011).

Statistical results of continent-difference evolutiontrends by quantitative methodThe graphic representations of HA gene and NA genefrom March 2009 to April 2012, respectively, are givenin Figures 3 and 4. They show that the early clus-ter with the vaccine isolate, A/California/07/2009, hasnearly disappeared in recent times (09/2009–04/2012).Some new clusters have deviated from the vaccine iso-

late little by little although this deviation did notmake WHO choose a new vaccine isolate instead ofthe old vaccine isolate A/California/07/2009. Here, todiscover the quantitative results of continent-differencegenetic drifts, we defined two relative parameters Dand R. The parameter D means the relative drift de-gree per site between the virus samples and the vac-cine isolate, A/California/07/2009. The parameter Rmeans the relative drift speed per site between thevirus samples and the vaccine isolate. Next we willgive the detailed definition of the two relative parame-ters.

414 Z. H. Qi et al.

Fig. 4b. For NA gene, the 2D graphic representation of 20 subsets of 6 continents in the last two stages. There are no virus sampleswith full-length testing data in the 4th stage of Africa, Asia, Oceania, and South America.

We first used MUSCLE program to do multi-alignment of all analyzed HA and NA sequences toget identical sites for each dataset. Then we orderedall aligned HA and NA sequences according to theirtime records. We used such representation N247D todescribe the amino acid substitutions between vaccinestrain and virus samples. The representation N247Dmeans the asparagine in position 247 of the vaccinestrain mutates into the aspartic acid in that positionof the compared sequence. Based on the amino acidsubstitutions, we defined two parameters D and Rto describe quantitatively the evolutionary trends ofviruses in different continents from March 2009 to April2012.Now let L be the length of the aligned sequences.

The Nseq means the number of the analyzed sequences.The Si is the number of the amino acid substitutionsin the i-th sequence Seqi. The Ai is the number ofthe newly added amino acid substitutions in the se-quence Seqi. The newly added amino acid substitu-tions Ai in the Seqi means the substitutions whichdo not appear in the i − 1 time-ordered sequencesSeqi−1, Seqi−2, . . . , Seq1. Then the parameters D and

R are defined as follows:

D = (1/Nseq)Nseq∑

i=1

(Si/L)

R = (1/Nseq)Nseq∑

i=1

(Ai/L)

The parameter D describes the relative deviation de-gree which the viruses deviate from the vaccine strainin a period. The parameter R means the relative devi-ation speed which the viruses deviate from the vaccinestrain in a period.Table 3 lists the analyzed results of the rela-

tive deviation degree D and the relative deviationspeed R of the 2009 pandemic influenza A (H1N1)viruses for the HA gene. For stage 1 of Africa andstage 4 of Europe, there is a slight inaccuracy aboutthe statistical results because of the too small sam-ple data. According to Table 3 the relative devia-tion degree D from different continents shows that theviruses are deviating from the recommended vaccine

Evolution trends of the 2009 pandemic influenza A (H1N1) 415

Table 3. The analyzed results of the relative deviation degree D (× 10−3) and the relative deviation speed R (× 10−3) of HA gene ofthe 2009 pandemic influenza A (H1N1) viruses.a

Continents Early-pandemic period Pandemic period Post-pandemic period Recent period(03/2009–08/2009) (09/2009–08/2010) (09/2010–08/2011) (09/2011–04/2012)

D R D R D R D R

Africa 5.2356 0 9.8366 0.1904 14.4602 0.8726 — —Asia 6.4847 0.3601 9.0420 0.5043 16.2054 0.3853 — —Europe 6.0653 0.2975 7.6468 0.2008 13.4609 0.0572 13.9616 0North America 5.4231 0.2726 8.2469 0.1864 14.9719 0.1837 19.6732 0.1587Oceania 7.4889 0.2651 9.5622 0.2182 13.0719 0.1027 — —South America 6.3967 0.3348 9.5332 0.1745 15.2082 0.9973 — —

a The gray colour highlights the imprecise results.

Table 4. The analyzed results of the relative deviation degree D (× 10−3) and the relative deviation speed R (× 10−3) of NA gene ofthe 2009 pandemic influenza A (H1N1) viruses.a

Continents Early-pandemic period Pandemic period Post-pandemic period Recent period(03/2009–08/2009) (09/2009–08/2010) (09/2010–08/2011) (09/2011–04/2012)

D R D R D R D R

Africa 6.3829 0 6.7435 0.2524 19.1489 4.2553 — —Asia 4.8958 0.3899 6.4625 0.4202 12.6889 0.4236 — —Europe 4.2419 0.1953 5.7993 0.2070 10.2082 0.3395 10.6383 0

North America 4.1588 0.2761 5.6440 0.2621 9.9482 0.3450 14.8936 0Oceania 4.7799 0.1749 7.2252 0.1330 10.8511 0.2128 — —South America 4.5390 0.1986 7.1900 0.2201 4.2553 0 — —

a The gray colour highlights the imprecise results.

strain, A/California/07/2009, little by little. However,the deviation is so small that WHO has not chosen anew vaccine isolate instead of the old vaccine isolateA/California/07/2009. According to the report fromWHO (2012), it is recommended that vaccines for usein the 2012–2013 influenza season (northern hemispherewinter) contain the following: an A/California/7/2009(H1N1) pdm09 – like virus. In addition, the relative de-viation speed R of Table 3 shows that in recent monthsthe viruses have not taken higher mutation speed com-paring with the viruses in the early-pandemic period.Table 4 lists the analyzed results of the relative de-

viation degree D and the relative deviation speed R ofthe 2009 pandemic influenza A (H1N1) viruses for theNA gene. There is a slight inaccuracy about the statisti-cal results because of too small sample data, such as instage 1 and stage 3 of Africa. According to Table 4 therelative deviation degree D from different continentsshows that the viruses are deviating from the recom-mended vaccine strain, A/California/07/2009, little bylittle. The deviation degree of the NA gene is little lowerthan that of the HA gene. Similarly, the relative devia-tion speedR of Table 4 also shows that in recent monthsthe viruses have not taken higher mutation speed com-paring with the viruses in the early-pandemic period.

Discussion

Fereidouni et al. (2009) and Shiino et al. (2009) showedthat the 2009 pandemic influenza A (H1N1) viruses di-verged into two clusters in the early-pandemic period.

One cluster originated from Mexico, Texas and Califor-nia. The other cluster originated from New York. ThenNelson et al. (2009) analyzed 290 H1N1pdm isolatessampled globally from 1st April to 9th July 2009. Theyrevealed that at least 7 phylogenetically distinct viralclades have spread globally and co-circulated in locali-ties. Another large scale phylogenetic analysis (Christ-man et al. 2011) used all available complete genomesequences from March 2009 to August 2010. They gota tree topology similar to the previous studies. Butthe bootstrap support for the two clusters was low.From a statistical view, these results for the early epi-demic viruses are acceptable because of different dataand different clustering level. Recently, Arunachalamet al. (2012) analyzed the novel influenza A/H1N12009 viruses by phylogenetic, comparative and statisti-cal analyses. Their analyses further confirmed that thegene exchange took place between viruses originatedfrom different species. Khandaker et al. (2013) selected75 isolates of Sendai City in Japan to analyze the ge-netic changes in the HA1 (nucleotides 52–1,029) andNA (nucleotides 1–1,395) genes, during 2009–2011. Theresults showed there was a steady rate for the mainte-nance of genetic diversity, followed by a slight increasein the later part of the 2010–2011 influenza seasons.In the present study we analyzed all full-length

6219 HA and 4860 NA sequences from March 2009 toApril 2012 to make their graphic representations. Wedelivered graphic representations of HA and NA se-quences of different continents (Figs 3 and 4) so thatit is possible to draw some conclusions. The viruses in

416 Z. H. Qi et al.

Table 5. The analyzed results of the relative deviation degree D (× 10−3) and the relative deviation speed R (× 10−3) of HA geneand NA gene about the influenza A (H1N1) viruses from September 2006 to March 2009.a

HA (09/2006–06/2007) HA (07/2007–03/2009)Continents

Sequences D R Sequences D R

Africa — — — 3 6.8493 1.7123Asia 17 18.0298 2.2160 127 9.4690 1.4782Europe 3 15.4110 1.1416 35 2.2831 1.3046North America 347 23.1831 0.6119 498 7.1947 0.7423Oceania — — — 12 7.6276 2.6463South America — — — 10 5.4795 0.3424

NA (09/2006–06/2007) NA (07/2007–03/2009)Continents

Sequences D R Sequences D R

Africa — — — 21 7.3248 0.9554Asia 18 26.1854 1.8872 184 14.1713 1.2406Europe — — — 25 4.2463 0.8493North America 365 27.0773 0.6689 488 7.7990 0.7201Oceania — — — 8 6.0661 0.9099South America — — — 13 11.5956 0.4899

aThe symbol ‘—’ means the absence of the full-length sequences in the continents.

several major clusters have spread globally in stage 1and stage 2. From a statistical view, the results are sim-ilar to conclusions reached previously (Fereidouni et al.2009; Nelson et al. 2009; Shiino et al. 2009; Christmanet al. 2011). Besides of the found clusters, the graphicrepresentations (Figs 3a, 4a) show a new major cluster(x < −1 and y > 1.5 in Fig. 3a; y > 1.5 in Fig. 4a). Thenew cluster began to emerge from Asia in the secondstage (09/2009–08/2010). Based on Figure 3b, it wasfound that the newly emerging viruses spread globallyin the post-pandemic period (09/2010–08/2011). It isnot difficult to predict the new major cluster cluster (x< −1 and y > 1.5 in Fig. 3a; y > 1.5 in Fig. 4a) shouldbecome the dominant isolates although the number ofthe available sequences is so small.The graphic representations of HA and NA se-

quences of different continents indicate that the virusesare deviating from the recommended vaccine strain,A/California/07/2009, little by little. However, theWHO has not chosen a new vaccine isolate instead ofthe old vaccine isolate, A/California/07/2009. It hasbeen recommended that vaccines for use in the 2012–2013 influenza seasons (northern hemisphere winter) in-clude the A/California/07/2009 – like viruses (WHO2012). Then it is reasonable to ask to what degree arethe viruses deviating from the recommended vaccinestrain? For comparison we calculated the values of therelative deviation degree D (Table 3) and the relativedeviation speed R (Table 4). It was found that the rela-tive deviation degree D of the viruses during the recentperiod is two-times more than in the early-pandemicperiod (03/2009–08/2009). However, during the recentperiod the relative deviation speed R is almost as muchas in the early-pandemic period although the deviatespeed is also increasing, little by little. These analyzedresults show that currently there has been no great mu-tation in the viruses.For further comparison, we analyzed more H1N1

viruses from different times. It is well-known thatin March 2009 the H1N1 viruses mutated into thenovel A (H1N1) 2009, A/California/07/2009 (H1N1,2009/04/09) – like virus. The pandemic influenza A(H1N1) virus is a combination virus from swine, avianand human influenza viruses (Fraser et al. 2009; Gartenet al. 2009). It shows a strong ability to broadcast fromhuman to human and spread worldwide. We wantedto know what happened to the H1N1 viruses beforeMarch 2009. During the several years before March2009, the WHO adjusted two-times about the rec-ommended vaccine isolates. One time was in October2008 when the WHO recommended the vaccine for usein the 2009 influenza season, an A/Brisbane/59/2007(H1N1, 2007/07/01) – like virus (WHO 2008). Theother one was in October 2007 when the WHO recom-mended the vaccine for use the 2008 influenza season,an A/Solomon Islands/3/2006 (H1N1, 2006/08/21) –like virus (WHO 2007). According to the different vac-cine isolates we considered two stages before March2009: (i) the period (09/2006–06/2007); and (ii) the pe-riod (07/2007–03/2009).Table 5 lists the analyzed results of the rela-

tive degree D and the relative deviation speed R ofthe influenza A (H1N1) viruses from September 2006to March 2009 for HA and NA genes. It is clearthat HA and NA genes have seriously deviated fromthe vaccine isolate, A/Solomon Islands/3/2006 (H1N1,2006/08/21) – like virus. Then WHO chose a new vac-cine isolate, A/Brisbane/59/2007 (H1N1, 2007/07/01)– like virus, instead of the old vaccine. The statistic re-sults of the next period (07/2007–03/2009) show thatthe change about the vaccine isolate reduces the devia-tion to acceptable level. Then based on the data shownin Tables 3 and 4, we found that the 2009 pandemic in-fluenza A (H1N1) viruses were deviating from the vac-cine isolate, A/California/07/2009 (H1N1, 2009/04/09)– like virus. The deviation degree is lower than in the

Evolution trends of the 2009 pandemic influenza A (H1N1) 417

period (09/2006–06/2007). But the trend of deviationin the recent period (09/2011–04/2012) was increasing,little by little. So suggest to continue the monitoringthe evolutionary trends of the 2009 pandemic influenzaA (H1N1) viruses for the next influenza seasons.In addition, according to the relative deviation

speed R (Table 5) we can see some important infor-mation. We found that the relative deviation speed Rhas been increasing little by little from September 2006to March 2009. The relative deviation degree D was re-duced to acceptable degree when WHO used new vac-cine isolate to replace the old one. However, the changedid not reduce the deviation speed R. Then in March2009 the viruses mutated into the novel A (H1N1) 2009,A/California/07/2009 (H1N1) – like virus. The pan-demic influenza A (H1N1) virus is a combination fromswine, avian and human influenza viruses. Based on theanalyzed results, the high deviation speed means thatthe viruses are likely experiencing great selection pres-sure. The virus genome would become no longer stable.The high deviation speed of the viruses likely becomesthe possible evidence of gene recombination.In conclusion, our graphic and statistical analyses

of the 6219 HA genes and 4860 NA genes show sev-eral main views. The old isolates, A/Brisbane/59/2007(H1N1) – like virus, died off little by little after the novelpandemic isolates, A/California/07/2009 (H1N1) – likevirus, exploded in Mexico in March 2009. The novelpandemic influenza A (H1N1) viruses experienced fourstages from March 2009 to April 2012. During the firststage, the early-pandemic period (03/2009–08/2009),the viruses have spread globally in several clusters.They began to deviate from the recommended vaccinestrain, A/California/07/2009, little by little. The de-viation speed was low. During the second stage, thepandemic period (09/2009–08/2010), the viruses fur-ther spread globally to several major clusters. New clus-ters in this stage began to emerge from Asia and NorthAmerica. They further deviated from the recommendedvaccine strain. The deviation speed was also low. Dur-ing the third stage (09/2010–08/2011) and the fourthstage (09/2011–04/2012), the newly emerging virusesfurther spread globally. The deviation degree betweenthe viruses and the vaccine isolate became larger andlarger. The original cluster with the recommended vac-cine isolate, A/California/07/2009, has nearly disap-peared. However, the deviation degree and the devi-ation speed were acceptable, although the viruses de-viated from the vaccine isolate continuously. Thus theWHO did not choose a new vaccine isolate instead ofthe original vaccine isolate A/California/07/2009. Allthese results emphasize the importance of continuousmonitoring of the 2009 pandemic influenza A (H1N1)viruses. It is also crucial to carefully monitor the un-derlying evolutionary changes in the viruses.

Acknowledgements

We thank the anonymous reviewers for their valuable com-

ments to improve this paper. This work supported by theNational Natural Science Foundation of China (Grant No.61272254), by the Natural Science Foundation of HebeiProvince, China (Project No. F2012210017), and by the Hu-manities and Social Sciences Research of Ministry of Educa-tion of China (Project name, The Origin, Propagation andMigration of Human Influenza Epidemic (1918–2010) fromSpace-time Perspective; Project No. 11YJCZH132).

References

Arunachalam R., Paulkumar K. & Annadurai G. 2012. Phyloge-netic analysis of pandemic influenza A/H1N1 virus. Biologia67: 14–31.

Bielinska-Waz D. 2010. Four-component spectral representationof DNA sequences. J Math. Chem. 47: 41–51.

Bielinska-Waz D., Nowak W., Waz P., Nandy A. & Clark T.2007. Distribution moments of 2D-graphs as descriptors ofDNA sequences. Chem. Phys. Lett. 443: 408–413.

Centers for Disease Control and Prevention (CDC). 2009. Out-break of swine-origin influenza A (H1N1) virus infection –Mexico, March-April 2009. Morbidity and Mortality WeeklyReport 58: 467–470.

Chou K.C. 2010. Graphic rule for drug metabolism systems. Curr.Drug Metab. 11: 369–378.

Chou K.C. & Shen H.B. 2009. FoldRate: a web-server for pre-dicting protein folding rates from primary sequence. OpenBioinform. J. 3: 31–50.

Christman M.C., Kedwaii A., Xu J., Donis R.O. & Lu G. 2011.Pandemic (H1N1) 2009 virus revisited: an evolutionary ret-rospective. Infect. Genet. Evol. 11: 803–811.

Costa J.C., Alves M.M. & Ferreira E.C. 2009. Principal compo-nent analysis and quantitative image analysis to predict ef-fects of toxics in anaerobic granular sludge. Biores. Technol.100: 1180–1185.

Edgar R.C. 2004. MUSCLE: multiple sequence alignment withhigh accuracy and high throughput. Nucleic Acids Res. 32:1792–1797.

Edgar R.C. 2010. Quality measures for protein alignment bench-marks. Nucleic Acids Res. 38: 2145–2153.

Fereidouni S.R., Beer M., Vahlenkamp T. & Starick E. 2009. Dif-ferentiation of two distinct clusters among currently circu-lating influenza A(H1N1)v viruses, March-September 2009.Euro Surveill. 14: pii=19409.

Fraser C., Donnelly C.A., Cauchemez S., Hanage W.P., VanKerkhove M.D., Hollingsworth T. D., Griffin J., Bagga-ley R.F., Jenkins H.E., Lyons E.J., Jombart T., HinsleyW.R., Grassly N.C., Balloux F., Ghani A.C., Ferguson N.M.,Rambaut A., Pybus O.G., Lopez-Gatell H., Alpuche-ArandaC.M., Chapela I.B., Zavala E.P., Espejo Guevara D.M., Chec-chi F., Garcia E., Hugonnet S., Roth C. & WHO Rapid Pan-demic Assessment Collaboration. 2009. Pandemic potential ofa strain of influenza A (H1N1): early findings. Science 324:1557–1561.

Garten R.J., Davis C.T., Russell C.A., Shu B., Lindstrom S.,Balish A., Sessions W.M., Xu X.Y., Skepner E., Deyde V.,Okomo-Adhiambo M., Gubareva L., Barnes J., Smith C.B.,Emery S.L., Hillman M.J., Rivailler P., Smagala J., GraafM., Burke D.F., Fouchier A.M., Pappas C., Alpuche-ArandaC.M., López-Gatell H., Olivera H., López I., Myers C.A.,Faix D., Blair P.J., Yu C., Keene K.M., Dotson Jr. P.D.,Boxrud D., Sambol A.R., Abid S.H., George K.S., T. Ban-nerman, Moore A.L., Stringer D.J., Blevins P., Demmler-Harrison G.J., Ginsberg M., Kriner P., Waterman S., SmoleS., Guevara H.F., Belongia E.A., Clark P.A., Beatrice S.T.,Donis R., Katz J., Finelli L., Bridges C.B., M. Shaw, Jerni-gan D.B., Uyeki T.M., Smith D.J., Klimov A.I. & Cox N.J.2009. Antigenic and genetic characteristics of swine-oOrigin2009 A(H1N1) influenza viruses circulating in humans. Sci-ence 325: 197–201.

He J. & Deem M.W. 2010. Low-dimensional clustering detects in-cipient dominant influenza strain clusters. Protein Eng. Des.Sel. 23: 935–946.

418 Z. H. Qi et al.

Jackson J.E. & Wiley J.W. 1991. User’s Guide to Principle Com-ponents. Wiley-Interscience, New York.

Khandaker I., Suzuki A., Kamigaki T., Tohma K., Odagiri T.,Okada T., Ohno A., Otani K., Sawayama R., Kawamura K.,Okamoto M. & Oshitani H. 2013. Molecular evolution of thehemagglutinin and neuraminidase genes of pandemic (H1N1)2009 influenza viruses in Sendai, Japan, during 2009–2011.Virus Genes 47: 456–466.

Lapedes A. & Farber R. 2001. The geometry of shape space:application to influenza. J. Theor. Biol. 212: 57–69.

Nelson M., Spiro D., Wentworth D., Beck E., Fan J., Ghedin E.,Halpin R., Bera J., Hine E., Proudfoot K., Stockwell T., LinX.D., Griesemer S., Kumar S., Bose M., Viboud C., Holmes E.& Henrickson K. 2009. The early diversification of influenzaA/H1N1pdm. PLoS Curr. 1: RRN1126.

Novic M. & Randic M. 2008. Representation of proteins as walksin 20-D space. SAR QSAR Environ. Res. 19: 317–337.

Potdar V., Chadha M.S., Jadhav S.M., Mullick J., Cherian S.S. &Mishra A.C. 2010. Genetic characterization of the influenzaA pandemic (H1N1) 2009 virus isolates from India. PLoS One5: e9693.

Qi Z.H. & Fan T.R. 2007. PN-curve: a 3D graphical representa-tion of DNA sequences and their numerical characterization.Chem. Phys. Lett. 442: 434–440.

Qu Y., Zhang R., Cui P., Song G., Duan Z. & Lei F. 2011.Evolutionary genomics of the pandemic 2009 H1N1 influenzaviruses (pH1N1v). Virol. J. 8: 250.

Randic M., Vračkoa M., Leršb N. & Plavšic D. 2003. Novel 2-Dgraphical representation of DNA sequences and their numer-ical characterization. Chem. Phys. Lett. 368: 1–6.

Randic M., Zupan J. & Pisanski T. 2008. On representation ofDNA by line distance matrix. J. Math. Chem. 43: 674–692.

Shiino T., Okabe N., Yasui Y., Sunagawa T., Ujike M., ObuchiM., Kishida N., Xu H., Takashita E., Anraku A., Ito R., DoiT., Ejima M., Sugawara H., Horikawa H., Yamazaki S., KatoY., Oguchi A., Fujita N., Odagiri T., Tashiro M. & Watan-abe H. 2009. Molecular evolutionary analysis of the influenzaA (H1N1) pdm, May-September, 2009: temporal and spatialspreading profile of the viruses in Japan. PLoS One 5: e11057.

Smith D.J., Lapedes A.S., de Jong J.C., Bestebroer T.M., Rim-melzwaan G.F., Osterhaus A.D. & Fouchier R.A. 2004. Map-ping the antigenic and genetic evolution of influenza virus.Science 305: 371–376.

Smith G.J., Vijaykrishna D., Bahl J., Lycett S.J., Worobey M.,Pybus O.G., Ma S.K., Cheung C.L., Raghwani J., Bhatt S.,Peiris J.S.M., Guan Y. & Rambaut A. 2009. Origins and evo-lutionary genomics of the 2009 swine-origin H1N1 influenzaA epidemic. Nature 459: 1122–1125.

World Health Organization (WHO). 2007. Weekly Epidemiologi-cal Record (WER). WHO [5 October 2007] 40(82): 345–356.

World Health Organization (WHO). 2008. Weekly Epidemiolog-ical Record (WER). WHO [10 October 2008] 41(83): 365–372.

World Health Organization (WHO). 2009. Weekly Epidemiolog-ical Record (WER). WHO [27 February 2009] 84(9): 65–76.

World Health Organization (WHO). 2010a. Weekly Epidemiolog-ical Record (WER). WHO [5 March 2010] 85(10): 81–92.

World Health Organization (WHO). 2010b. Pandemic (H1N1)2009 – update 97. WHO. [Accessed 23 April 2010]. Availablefrom: http://www.who.int/c-sr/don/2010 04 23a/en/index.html.

World Health Organization (WHO). 2010c. H1N1 in post-pandemic period. WHO. [Accessed 10 August 2010]. Availablefrom: www.who.int/mediacentre/news/statements/2010/h1n1 vpc 20100810/en/index.html.

World Health Organization (WHO). 2010d. Weekly Epidemiolog-ical Record (WER). WHO [8 October 2010] 85(41): 401–412.

World Health Organization (WHO). 2011a. Pandemic (H1N1) –update 64. 04 September 2009. WHO; 2011. Available from:http://www.who.int/csr/don/2009 09 04/en/index.html.

World Health Organization (WHO). 2011b. Weekly Epidemiolog-ical Record (WER). WHO [4 March 2011] 86(10): 81–92.

World Health Organization (WHO). 2011c. Weekly Epidemiolog-ical Record (WER). WHO [14 October 2011] 86(42): 457–468.

World Health Organization (WHO). 2012. Weekly Epidemiolog-ical Record (WER), WHO [9 March 2012] 87(10): 81–96.

Xiao F., Wang S.W., Xu X.H. & Ge G.M. 2009. An isolationenhanced PCA method with expert-based multivariate de-coupling for sensor FDD in air-conditioning systems. Appl.Therm. Eng. 29: 712–722.

Xiao X., Shao S., Ding Y., Huang Z., Chen X. & Chou K.C. 2005.Using cellular automata to generate image representation forbiological sequences. Amino Acids 28: 29–35.

Xiao X., Shao S., Ding Y., Huang Z. & Chou K.C. 2006. Usingcellular automata images and pseudo amino acid compositionto predict protein subcellular location. Amino Acids 30: 49–54.

Yao Y.H., Dai Q., Li L., Nan X.Y., He P.A. & Zhang Y.Z. 2010.Similarity/dissimilarity studies of protein sequences based ona new 2D graphical representation. J. Comput. Chem. 31:1045–1052.

Received February 22, 2013Accepted January 8, 2014