The complete genome sequences and the list of annotated gene and protein sequences of TIGR4, D39 and R6 are retrieved from the NCBI ndash; FTP server (ftp://ftp.ncbi.nih.gov/genomes). We used the run-mummer3 program available in the standalone MUMmer 3.20 (http://mummer.sourceforge.net/) and its built-in mummerplot for obtaining the whole genome pairwise alignment of S. pneumoniae strains TIGR4, D39, and R6 in different combinations. MUMmer at Comprehensive Microbial Resource (CMR) is used for the whole genome pairwise alignment of the strains TIGR4, G54 and R6 in different combinations.

The tool in CMR database (http://cmr.tigr.org/tigr-scripts/CMR/ CmrHomePage.cgi), the role category piechart is used for the genome features and functional role category comparison of the whole genomes of TIGR4, G54 and R6. Bacterial Annotation System (BASys - http://wishart.biology.ualberta.ca/basys) - A web server for automated bacterial genome annotation is used to know the role category for three strains TIGR4, D39 and R6, whose whole genomes are already available in it. From the prediction server of the Center for Biological Sequence Analysis (CBS - http://www.cbs.dtu.dk/services), the Genome Atlas is used for the analysis of repeats of S. pneumoniae. The sequences of various virulence factors, which are taken for our study, have been verified by using the virulence factors database (http://www.mgc.ac.cn/VFs). BioEdit (http://www.mbio.ncsu.edu/ BioEdit/bioedit.html) is used to compute sequence composition of the genomes and genes. Further, LALIGN (http://www.ch.embnet.org/software/LALIGN_form.html) is used for the pairwise global alignment of the gene sequences of the strains of S. pneumoniae.

VirulentPred (http://bioinfo.icgeb.res.in/virulent) is a SVM (Support Vector Machine) based method to predict bacterial virulent protein sequences, which can be used to screen virulent proteins in proteomes. In the present study the above tool is used to analyse the hypothetical sequences of the strains TIGR4 and R6 of S. pneumoniae. From the proteome of TIRG4 and R6 of S. pneumoniae, all unannotated hypothetical protein sequences are retrieved using PERL script and those sequences are used as data set for virulence factor prediction.

Table 1 summarizes the general information about the genomes including statistics of genes of these four strains, obtained and compiled from CMR and NCBI web servers. The genome sizes of these four strains range between 2 Mb and 2.16 Mb (c.f. Sl.No.2 of Table1). Among these four strains, D39 is the smallest and TIGR4 is the largest based on genome size. The nucleotide base (A, T, G, C, AT and GC) compositions of four strains show that the strains have low GC (~40%) genomes. The number of genes encoding for proteins of these four strains ranges between 1914 and 2234 (c.f. Sl.No.3 of Table1). Of the total base pairs of four genomes, approximately 85 - 87% of base pairs (bps) are involved in coding and the remaining are non-coding or junk DNA. The number of genes involved in RNA synthesis (structural RNA, tRNA, and rRNA) is more or less similar in all strains. Finally, by comparing the global and local repeats of TIGR4 and R6 using CBS web server, it is evident that both the repeats are high in TIGR4 than in R6 (c.f. Sl.No.4 of Table1) and this may be related to the duplicated regions of the chromosome (Gregory and DeSalle, 2005).

The whole genome pairwise alignments of the strains TIGR4, D39 and R6 of S. pneumoniae (whose sequence data are available at NCBI) are obtained using the standalone version of MUMmer and the results are plotted using its built-in mummerplot. The whole genome pairwise alignments of the strains TIGR4, G54 and R6 are obtained using CMR, where these sequences are available, and the five possible alignments are shown in Figure 1(a) ndash; (e). Generally, the genomes of prokaryotes are very dynamic, with insertions, deletions, inversions, and translocations being commonly observed among related species or even between different strains of the same species (Gregory and DeSalle, 2005; Hughes, 2000). The net result is that the particular complement of genes and their order along the chromosome are not typically conserved over evolutionary time. In some cases, genes that are grouped into operons in one species may be dispersed throughout the genome in others. We find similar results, while we analyzed the genomes of four strains of S. pneumoniae. In particular, we find that there exists a stability of the gene order in the genome pairs TIGR4 vs. D39 and TIGR4 vs. R6 and they are shown by fact that most of the points lie along the diagonal in Figures 1a and 1b. The results (Figures 1a and 1b) indicate that the stability of gene order of D39 vs. R6 must also be relatively high and it is shown in Figure 1c. This also confirms the fact that R6 is the derivative of D39. The whole genome pairwise alignments of TIGR4 vs. G54 and that of R6 vs. G54 do not show such a high degree of the stability of gene order compared to the above results (for D39 strain) and are shown in Figures 1d and 1e, respectively.

Many of the gene and protein sequences among these strains are approximately the same and this is not surprising as all the strains occupy the same niche in the human respiratory system. The small differences might have arisen after the divergence of these strains from other evolutionary lineages for adaptations in their host. This increases greatly in pathogens and appears to be associated with the ability to infect eukaryotes, perhaps reflecting a mechanism for evading host immune defenses and the unique genes may be located in a plasticity zone.

Since G54 genome sequence is not available at NCBI web server and D39 genome is not available at CMR server, we could not get the whole genome alignment for D39 vs. G54. However, we are able to predict the whole genome pairwise alignment of D39 vs. G54, based on the earlier result. As the Figures 1d and 1e are similar, it indicates that the alignment of D39 vs. G54 must also possess similar structure. This prediction may be confirmed if the whole genome sequence of G54 is made available in NCBI or genome sequence of D39 is included in CMR.

We have compared the capsular polysaccharide (cps) synthesizing genes of the strains TIGR4, D39, G54 and R6 of S. pneumoniae and the results are shown in Table 2. There are 15 different cps genes in TIGR4, 7 in D39 and 9 in G54 and only one in R6. Their gene IDs, G+C percentage, protein length, gene length and gene coordinates are shown in Table 2. On comparison, it is estimated that 5 cps genes of TIGR4 (gi|15900275-cps4A, gi|15900276-cps4B, gi|15900278-cps4D, gi|15900046-cps-ptv amp; gi|15901666-cpsptv) are related to that of D39 (gi|116516963-cps2A, gi|116516159-cpsB, gi|116517023-cps2D, gi|116517199-cps and gi|116516120-cps-ptv). All the cps genes of D39 are present in TIGR4 except gi|116516773-cps2E and gi|116516341-cps-ptv.

Between TIGR4 and G54, 6 cps genes are related (gi|15900275-cps4A, gi|15900276-cps4B, gi|15900277-cps4C, gi|15900278-cps4D, gi|15900046-cps-ptv amp; gi|15901666-cps-ptv of TIGR4 with NT05SP0190-cps4A, NT05SP0191-cps4B, NT05SP0192-cps4C, NT05SP0193- cps4D, NT05SP2185-cps9E & NT05SP1650-cps7G of G54). Likewise, between D39 and G54, 5 cps genes are related (gi|116516963-cps2A, gi|116516159-cpsB, gi|116517023-cps2D, gi|116517199-cps and gi|116516120- cps-ptv of D39 with NT05SP0190-cps4A, NT05SP0191- cps4B, NT05SP0192-cps4C, NT05SP2185-cps9E & NT05SP1650-cps7G of G54), but gi|116516773-cps2E and gi|116516341-cps-ptv of D39 are not present in G54. Similarly, it is interesting to note that the only cps gene of R6 (gi|15902136-capD), has 99.8 % identity with the gene gi|15900046-cps-ptv of TIGR4, 100 % identity with the gene gi|116517199-cps of D39 and 99.5 % identity with the gene NT05SP2185 of G54. All the above results are in support of the Avery’s statement (Avery et al., 1979) that the capsule is responsible for pathogenecity.

From similar analysis, we have also noted that the genes, gi|15900279-cps4E, gi|15900280-cps4F, gi|15900281-cps4G, gi|15900282-cps4H, gi|15900286-cps4I, gi|15900287-cps4J, gi|15900288-cps4K, gi|15900289-cps4L and gi|15900788- cps-ptv are unique to TIGR4. Similarly, the genes gi|116516773-cps2E and gi|116516341-cps-ptv are unique to D39 strain. In the same way, the genes NT05SP0198, NT05SP0202 and NT05SP1909 are unique to the strain G54. But in R6, the only cps gene gi|15902136-capD is common to all other strains (Table 2). As the TIGR4 strain has more number of cps genes than other strains it indicates the high virulence nature of TIGR4. Further, the results also explain that the virulence nature is lesser in D39 and G54 strains, and very less in R6 compared to TIGR4.

Though all the cps genes of TIGR4 are not present in D39, G54 and R6 strains, they are also pathogenic. Therefore, to know the other virulence factors in addition to cps genes, we consider the other genes of the strains from the gene role category aspect.

Role category of genes of the different strains are compared by using the two different tools, namely, i. CMR – role category pie chart for TIGR4, G54 and R6 (Table 3) and ii. Bacterial Annotation System (BASys) for the strains TIGR4, D39 and R6, based on the availability of genome sequences. The genes responsible for biosynthesis of various proteins (Sl. Nos. 1-9 of Table 3) of TIGR4 are nearly same as in G54 and R6, which suggests the basic complement of proteins required for certain cellular processes. But the genes responsible for the biosynthesis of some other proteins (Sl.Nos.10-23 of Table 3) of TIGR4 are notably different from that of G54 and R6. This suggests that, these proteins are important for strain uniqueness and they may be involved in variations in pathogenesis among the strains of S. pneumoniae. The percentage values given for a particular role category in Table 3 is specific to the gene involved in that category only and does not represent the overall gene percentage. For example, autolysin (SP1937) of TIGR4 is categorized into two role categories such as cell envelope and cellular processes (Sl.Nos.11 and 12 of Table 3) and the percentage given is specific to the respective categories.

The number of genes which are responsible for pathogenesis in the strains TIGR4, G54 and R6 are manually counted from CMR gene role category (sub role categories pathogenesis, toxin production and resistance) and found to be 101 (4.52 %), 47 (2.30 %) and 42 (1.89 %) respectively (Sl.No.19 of Table 3). TIGR4 has many pathogenic factors and it is highly virulent and G54 and R6 strains have approximately 50% of the pathogenic factors of TIGR4. Mobile and extra chromosomal elements comprise a significant fraction of the genome as with the 134 genes (5.99 %) in TIGR4, 71 (3.46 %) in G54 and 86 genes (3.87 %) in R6 (Sl.No.18 of Table 3). Generally transposons encode genes for antibiotic resistance (Gregory and DeSalle, 2005); therefore from our results, it is evident that the antibiotic resistance may be relatively higher in TIGR4 than the strains G54 and R6.

From the results of the comparative study on TIGR4, D39 and R6, using BASys server, we find that most of the values are more or less similar. But, there is a higher percentage for unknown functions in the strains TIGR4, D39 and G54, which indicates that the reason for the differences may also be hidden in the unknown genes or proteins (data not shown).

From Table 3, the number of hypothetical, conserved hypothetical, unclassified and unknown genes of whole genomes of the strains TIGR4, G54 and R6 are noted and is shown in Table 4. Nearly 37 - 42 % of genes are of unknown type and it shows that these sequences have to be annotated and assigned functions of which some of them may be responsible for the virulence nature. Using the multigenome homology comparison tool, which is available at CMR, the numbers of unique genes in TIGR4, G54 and R6 are found to be 288, 104 and 78, respectively (Table 4).

The unique genes of the strains TIGR4, G54 and R6 themselves have many hypothetical, conserved hypothetical, unknown and unclassified sequences and their percentage ranges from 65 to 74, thus the other possible differences among the strains may be known by studying the above said gene sequences. As far as the virulence factors are concerned, in the unique genes of the strain TIGR4, 3 capsular polysaccharide biosynthesis proteins (Sp_0351 (cps4F), Sp_0352 (cps4G) and Sp_0359 (cps4K)), 4 cell wall surface anchor family proteins (Sp_0462, Sp_0463, Sp_0464 and Sp_1772), a PspC protein (Sp_1417), a NanA protein (SP_1693) and a IgA1 protease (SP_2155) are there. In the case of R6, it has three proteins of type 2 capsule locus (Spr0315, Spr0317 and Spr0319) in its unique genes. But the strain G54 does not have such virulence factors in its unique genes (Table 4). The above result shows the high virulence nature of TIGR4 and it also suggests that those virulence factors are specific to TIGR4 and R6. The above differences might have arisen because of the species-specific adaptation to their host particularly in the sake of defense mechanism.

In S. pneumoniae, the surface and cytoplasmic proteins such as pneumococcal surface protein A (PspA), autolysin (LytA), hyaluronate lyase (Hyl), pneumolysin (Ply), two neuraminidases (NanA and NanB), choline binding protein A (CbpA), pneumococcal surface antigen A (PsaA) and immunoglobulin A1 (IgA1) protease are already stated as the virulence factors (Jedrzejas, 2001; Rigden et al., 2003). The comparative results of the above mentioned sequences obtained from CMR, are given in Table 5. It provides more insight into the virulence factors of the strains TIGR4, D39, G54 and R6 of S. pneumoniae.

The virulence factors of TIGR4 are taken as reference and are compared with all other related sequences of the strains such as D39, G54 and R6, likewise the virulence factors of D39 are taken as reference and are compared with all the related sequences of the strains G54 and R6. Similarly the virulence factors of G54 are taken as reference and are compared with all the related sequences of the remaining strain R6 using the pairwise sequence alignment tool LALIGN, with default parameters (Alignment: Global; Scoring matrix: BLOSUM50, Gap opening penalty: -14 and extension penalty: -4), and all the results are comparatively shown in Table 5.

PspA is located in the cell wall of pneumococci and present in all S. pneumoniae strains (Jedrzejas, 2001). PspA of TIGR4 has ~53-63% identities with D39, G54 and R6 (Table 5). When we compare PspA in D39 vs. G54 and G54 vs. R6, the identities between those strains are nearly 63%. The above results indicate that nearly 50-60% virulence nature of PspA of TIGR4 exist in other strains D39, G54 and R6. But it is interesting to note that there is 100% identity between the PspA sequences of D39 and R6, thus the virulence nature of PspA is exactly the same.

Regarding LytA, Hyl, Ply, NanB and PsaA, all the four strains of S. pneumoniae have above 90% identities, thus the effect of the above mentioned five virulence factors is also similar and it also reflects on G+C percentage, protein length and gene length, but the location in their genomes varies and the similarities and differences can be noticed from the Table 5.

All strains have different neuraminidase sequences except G54 and R6 (~90% identity). In the case of CbpA and IgA1 of the strain TIGR4, high percent identities (~73 and 87%) exist with D39 and R6 respectively, exactly identical (100%) between D39 and R6. But very less identities (~40 and 35%) exist with G54 combinations. It seems that the virulence nature based on cbpA and IgaA are similar among the strains TIGR4, D39 and R6 and differs in G54.

From Table 5, it is interesting to note that all the virulence factors of D39 are very similar to R6 (above 99% identities except NanA), and it confirms the fact that the avirulent strain R6 is the derivative of the strain D39 (Lanie et al., 2007). Based on the role category, all TIGR4 virulence factors come under pathogenesis related functions and it also says that TIGR4 has high virulence nature.

Prediction of virulence factors from the hypothetical sequences of S. pneumoniae has implications on the identification and characterization of the virulence mechanism. The present study predicted using VirulentPred (Garg and Gupta, 2008) that 4 hypothetical sequences of TIGR4 and 22 of R6, respectively, are virulence factors. All these sequences are listed in Table 6. The prediction is based on protein features, such as, amino acid composition, di-peptide composition, similarity search, higher order di-peptide composition, PSSM and cascaded SVM module of the tool VirulentPred. However, similar predictions are not possible at present with D39 and G54 as the sequence information of the latter is not fully available.

Among the 4 predicted virulence factors of TIGR4, only one sequence (gi|15901572) is predicted in R6 as a hypothetical protein (gi|15903627) and the functional region is predicted as Plasmid_Txe (PF06769). This family contains many hypothetical proteins and there is no homolog with other mentioned virulence factors. But in R6, it is interesting to note that among the 22 predicted virulence factors of hypothetical protein sequences, 8 different sequences (gi|15902372, gi|15903388, gi|15903446, gi|15902652, gi|15902781, gi|15903694, gi|15903627 and gi|15903771) with 7 different functional regions which are related to the already mentioned virulence factors of the strains R6 and TIGR4. Those virulence factors are hyaluronidase, Immunoglobulin A1 protease, capsular polysaccharide synthesis, pneumolysin, neuraminidase and choline binding protein. The above mentioned related sequences of TIGR4 and R6 except gi|15903771 are compared in Table 7.

The hypothetical protein sequence, gi|15903771 of R6 has 71 amino acids and its functional region is predicted as putative cell wall binding repeat (42-60) using Interproscan (ID - PF01473). It is also found that the same functional region is repeatedly present in the known virulence factors such as pneumococcal surface protein A, autolysin and choline binding proteins of the strains TIGR4 and R6. Since many domain regions have been identified in the above mentioned known virulence factors of TIGR4 and R6, the regions are not explicitly given. But one can easily obtain those regions using the tool Interproscan.