In the present study, almost all complete genome sequences of the LSD viruses were retrieved from the NCBI public database. The complete genome sequences were aligned using the MAFFT 7.407_1 alignment program with the parameters of Gap extend penalty-0.123 and Gap opening penalty-1.53 (Katoh and Standley, 2013; Mareuil et al., 2017; Lemoine et al., 2019). Further, the quality of the alignments was checked and curated using the LAST Plot hits utilizing the MAFFT version 7 of the online server tool with a score of 39 (E = 8.4e⁻¹¹)¹ (Katoh and Frith, 2012; Katoh et al., 2019). When these nucleotide sequences were aligned, it was revealed that the three sequences OK422492.1/India/2019/Ranchi-1/P10, OK422493.1//India/2019/Ranchi-1/P3, and ON400507.1/208/PVNRTVU/202 submitted to the NCBI public database from India had the highest nucleotide diversity. Subsequently, LAST hits plot analysis revealed that these three sequences were submitted as reverse complement compared to NCBI reference sequences NC_003027.1_LSDV_NI-249 (Supplementary Figures S1A–D). Therefore, throughout this present study, we have converted these three sequences into reverse complements and subjected them to analysis. Also, we have retrieved LSDV complete genome sequences from SRA run files SRR21590382, SRR21590384, SRR21590385, SRR21590386, and SRR21590383 related to LSDV submitted to NCBI public database from India and subjected to analysis. Briefly, these SRA run files were subjected to quality control using Trimmomatic (Bolger et al., 2014) to filter and remove low-quality reads and potential adopters sequences from the reads (Vilsker et al., 2019). From these quality control passed reads, the virus-specific reads were filtered using the protein-based alignment method DIAMOND (Buchfink et al., 2015), and de novo assembled using metaSPAdes (Bankevich et al., 2012), de novo assembled virus sequences recognized using Blastx and Blastn in the NCBI RefSeq virus database. Further, individual contig was aligned in the advanced genome aligner (AGA) (Deforche, 2017), and consensus variant caller GATK/BcfTools were used in the analyses. Confirm annotated variants and SNPs and mismatches with Raw reads were presented in Supplementary Data 10.

In the present study, the complete genome nucleotide sequences of LSD viruses were aligned in MAFFT 7.407_1, and then phylogenetic analysis was performed in PhyML 3.3_1 (Galaxy Version 3.3_1) (Mareuil et al., 2017). The GTR (evolutionary model), discrete gamma model (categories with the n = 4), empirical (equilibrium frequencies), subtree pruning and regraphing with tree topology search with tree topology, model parameter, and branch length, and then branch support were tested with approximate Bayes branch. Subsequently, the generated phylogenetic tree with the above parameters was visualized in the interactive tree of life (iTOL)-v5 (Letunic and Bork, 2021).

In the present study, the whole genome nucleotide sequences of the LSD viruses were aligned in MAFFT 7.407_1, and then the aligned nucleotide sequences were used for the NBGMD analysis in MEGA7 (Kumar et al., 2016) using the Kimura two-parameter model with the transitions + transversions substitution, gamma distribution (shape parameter = 5), gaps/missing data were deleted by pairwise deletion, and finally, the standard errors for the NBGMD analysis were calculated by the bootstrap of 1,000 replicates. The calculated standard error in the analysis was presented above the diagonal of the result table.

The SimPlot 3.5.1 (Desingu et al., 2022a,b) tool was used to determine the per cent identity/similarities among the different clusters of LSD viruses against reference sequences. In this study, the whole genome nucleotide sequences of the LSD viruses were aligned in MAFFT 7.407_1. Then the aligned nucleotide sequences were exported to SimPlot 3.5.1 tool for the subsequent analysis using the Kimura (two-parameter) method and base pairs of the window of 500 at a base-pair step of 50.

The complete genome sequences of LSDVs were aligned in the MAFFT 7.407_1 and then exported to RDP4 (Martin et al., 2015) for the recombination analysis. The recombination analysis was performed using default parameter values for the RDP, GENECONV, BOOTSCAN, Chimaera, 3seq, SISCAN, and MaxChi methods, and a minimum of four approaches was assessed for possible recombination using a Bonferroni corrected p-value cut-off (0.05).

The nucleotide/amino acid mismatch, transition/transversion, and silent/non-silent mutations were measured among the different clusters of LSD viruses against the reference sequence at the complete genome levels, gene levels, and genes that are transcribed in the forward and reverse directions by the Highlighter tool (Keele et al., 2008) with or without similarity sorting of the sequences, with/without treating the gaps as a character, and the reference sequences used in the analysis were displayed in the respective figures. Additionally, the nucleotide/amino acid mismatch was also visualized by the online Variant Visualizer,² and the reference sequences used in the analysis were displayed in the respective figures.

The APOBEC motif mutations in the LSD viruses’ complete genome sequences and genes transcribed in the forward/reverse directions were determined in the Hypermut 2.0 tool with the customized options (Rose and Korber, 2000) against the reference sequences, and the reference sequences used were displayed in the respective figures. Next, the dN/dS ratio in the LSD virus’s genes transcribed in the forward/reverse directions was measured in the SNAP v2.1.1 (Ota and Nei, 1994; Ganeshan et al., 1997; Korber, 2000). The reference sequences used in the analysis were displayed in the respective figures.

The nucleotide composition (A%, T%, G%, and C%) of the LSD viruses’ complete genome sequences and genes transcribed in the forward/reverse directions were determined in MEGA7 (Kumar et al., 2016) and Automated Codon Usage Analysis (ACUA) Software (Vetrivel et al., 2007).

The effective number of codon usage from 61 codons for the 20 amino acids for the LSD viruses’ genes transcribed in the forward/reverse directions was determined in version 6 of the DNA Sequence Polymorphism (DnaSP) software (DnaSP 6) (Rozas et al., 2017).

In ENc-GC3s plot analysis, the ENc values are plotted against the third position of GC3s of codon values for the LSD viruses’ genes transcribed in the forward/reverse directions. The expected curve was determined as recommended in previous publications (Wang et al., 2018; Tian et al., 2020). The ENc values and the GC3s of codon values were measured in version 6 of the DNA Sequence Polymorphism (DnaSP) software (DnaSP 6) (Rozas et al., 2017).

For parity rule 2 (PR2)-bias, the AT bias [A3/(A3 + T3)] is plotted against GC-bias [G3/(G3 + C3)] in the LSD viruses’ genes transcribed in the forward/reverse directions. The A3, T3, G3, and C3 values of nucleotide sequences were obtained by using ACUA Software (Vetrivel et al., 2007).

To understand the phylogenetic relationship of LSD viruses at the whole genome level, we retrieved almost all LSDV complete genome sequences from the NCBI public database and performed phylogenetic analysis. In the phylogenetic analysis, viruses related to wild-type were clustered into cluster-1.2, viruses related to vaccine into cluster-1.1, and recombinant viruses clustered into recombinant-cluster (Figure 1A), as reported by the previous studies (van Schalkwyk et al., 2020; Krotova et al., 2022a,b; Ma et al., 2022). Consistent with this, we also observed the recombinant events in the recombinant viruses by recombination detection program (RDP) analysis (Supplementary Figures S1–S3). Further, with the topology of the phylogenetic tree, LSD viruses, we have divided cluster-1.1 viruses into five sub-clusters, namely cluster-1.1.1 to 1.1.5 (Figure 1A), and similarly cluster-1.2 viruses into three sub-clusters namely cluster-1.2.1 to 1.2.3 (Figure 1A) (Ma et al., 2022; van Schalkwyk et al., 2022; Bhatt et al., 2023). Among these, sub-cluster-1.1.1 and 1.1.3 contain vaccine viruses, sub-cluster-1.1.2 contains vaccine-associated virulence field strains detected in South Africa, and sub-clusters-1.1.4 and 1.1.5 contain vaccine-associated recombinant virulence field strains detected in Russia (Figure 1A). Similarly, cluster-1.2.3 of the Udmurtiya strain is also a recombinant virus (Figure 1A). Furthermore, the recombinant viruses in clusters-1.1.4, 1.1.5, and 1.2.3 are different from the recombinant viruses belonging to the recombinant cluster that were detected in Russia, China, Thailand, Hong Kong, Taiwan, and Vietnam in 2019–2021 (Figure 1A). For better readability, hereafter recombinant cluster viruses are recognized as recombinant, and other recombinant viruses such as Utmurtia, Saratov, and Tyumen are recognized as 1.2.3, 1.1.4, and 1.1.5, respectively. In particular, wild-type viruses in sub-cluster-1.2.1 have been detected only in African countries and India. Finally, wild-type viruses in sub-cluster-1.2.2 are detected in African countries, Russia, Kazakhstan, Turkey, Israel, Greece, Bulgaria, and India (Figure 1A), however, Greece and Bulgaria have eradicated LSDV through mass vaccination.

Furthermore, Net Between Group Mean Distance (NBGMD) analysis revealed less than 1.25% nucleotide diversity among wild-type viruses (sub-clusters-1.2.1 and 1.2.2) and vaccine and vaccine-derived viruses (sub-clusters-1.1.1, 1.1.2, and 1.1.3) (Figure 1B). Viruses in the recombinant cluster exhibited 0.52%–0.72% nucleotide diversity with viruses in other clusters, including wild-type and vaccine strains (Figure 1B). Since there is only a very low level of nucleotide diversity between different clusters, it is clear that microevolution has occurred between these clusters. After this, we conducted a similarity plot analysis to determine which genomic regions have the highest genetic diversity among these clusters. This analysis revealed less than 0.25% genetic diversity in almost all genomic regions between distinct clusters, and viruses in the recombinant cluster exhibited recombination with wild-type and vaccine viruses in multiple genomic regions (Figure 1C) (Krotova et al., 2022a,b; Ma et al., 2022). Also, similar to similarity plot analysis, nucleotide mismatches analysis revealed that viruses in the recombinant cluster alternately exhibited nucleotide similarity with wild-type and vaccine viruses in multiple genomic regions and that nucleotide differences were primarily due to SNPs (Figure 1D) and majorly, transition mutations (Supplementary Figure S4). Since viruses in the recombinant cluster alternately express SNPs with wild-type and vaccine viruses in multiple genomic regions, LSDV-vaccine strains are attenuated (likely due to disruption of virulence genes to attenuate the virus) mainly by the passage in the unnatural host or unnatural host’s cells (Wallace and Viljoen, 2005; Tuppurainen et al., 2021), and this virus infects cattle, buffaloes, springbok, impala and giraffe (Young et al., 1970; Le Goff et al., 2009; Namazi and Khodakaram Tafti, 2021). A recent study using a short-read next-generation sequencing method explains the possibility of the origin of recombinant LSDVs through the homologous recombination of the Neethling-like LSDV vaccine strain and KSGP-like LSDV vaccine strains in the vaccine (Vandenbussche et al., 2022).

In some viruses recently detected in the LSDV outbreak, LSDV019 and LSDV144 genes have been reported to be two fragments, LSDV019a, LSDV019b, and LSDV144a, LSDV144b, respectively, through frameshift mutations (Kumar et al., 2023), so we are interested in finding out which other genes have frameshift mutations and in-frame nonsense mutations among different clusters. In the present study, a frameshift mutation in LSDV035, LSDV019, LSDV134, and LSDV144 genes and in-frame non-sense mutations in LSDV026, LSDV086, LSDV087, LSDV114, LSDV130, LSDV131, LSDV145, LSDV154, LSDV155, LSDV057, and LSDV081 genes were revealed among different clusters (Figures 2–4; Supplementary Figure S5).

Remarkably, if the translation of the LSDV035 (putative RNA polymerase subunit-402 amino acid length) gene starts at the start codon frame and position of wild-type viruses, this gene is truncated in the vaccine, vaccine-derived, and recombinant viruses (Figure 2A). Whereas, if the LSDV035 gene starts translation in the start codon frame and position of vaccine, vaccine-derived, and recombinant viruses, this gene is truncated in wild-type viruses (Figure 2B). Also, it is significant that 32 amino acids “MFVLKLFNFNIYKNEFLVLLYLDFSINAKMENN” are extra at the N-terminal of the wild-type viruses (Figures 2A,B). In addition, OK422493.1/India/2019/Ranchi-1/P30, OK422494.1/India/2019/Ranchi-1/P50, and MT007951.1/Namibia/2016/10F viruses have LSDV035 gene truncated in both frames. It is also worth noting that the vaccine-derived virus OL542833.1/Russia/Tyumen/2019 follows the frame of wild-type viruses (Figures 2A,B).

Interestingly, if the translation of the LSDV019 (putative remodeling and stabilization of the host cytoskeleton and host immune evasion) gene is initiated in the start codon frame and position of wild-type viruses, this gene is truncated in viruses detected elsewhere except those detected in African countries (except AF409137.1/Neethling_Warmbaths/LW, and MW656253.1/280-KZN/RSA/2018 detected in Africa) (Figure 2C). Further, in this frame vaccine virus cluster 1.1.1 is also truncated; it is also worth noting that the SRR21590386 detected in India follows the frame of wild-type viruses (Figure 2C). On the other hand, if the LSDV019 gene starts the translation in the start codon frame and position of recombinant viruses, this gene is truncated in cluster-1.2 viruses detected in other places except the viruses detected in African countries (except AF409137.1/Neethling_Warmbaths/LW, and MW656253.1/280-KZN/RSA/2018 detected in Africa; SRR21590386). Also, it is significant that 129 amino acids “MTLKRYINKEYVEELKYMLLKNDYDRILIFTIGGISQTRKDIFEIS SNYFKKAIKKYSNEVKLPFKYKPFTYVLEYINTGYITLNSKNVVDIFAISNVIEIKFIM DACTDFMINYIDDSNCVDIFRRSY” are extra at the N-terminal of the wild-type viruses (Figures 2C,D). Also, the LSDV019 gene is translated in both frames in wild-type and vaccine-derived viruses found in African countries, but in vaccine cluster-1.1.1 and recombinant viruses, it is translated in only one frame and produces a protein with 129 amino acids less. Significantly, the LSDV019 gene has two fragments, LSDV019a, and LSDV019b, in wild-type cluster-1.2 viruses detected elsewhere, except for viruses detected in African countries (except AF409137.1/Neethling_Warmbaths/LW and MW656253.1/280-KZN/RSA/2018 detected in Africa) (Figures 2C–E).

Similarly, the LSDV134 (variola virus B22R-like protein) gene produces two fragments, LSDV134a and LSDV134b, in vaccine cluster-1.1.1 (except KX764644.1/Neethling-Herbivac/vaccine and MW656252.1/Haden/RSA/1954) and OL542833.1/Russia/Tyumen/2019 viruses (Figures 3A–C). LSDV134b is truncated in the KX683219.1/KSGP/0240 virus belonging to wild-type cluster-1.2.1 (Figures 3A–C). Additionally, the LSDV144 (putative remodeling and stabilization of the host cytoskeleton and host immune evasion) gene produces two fragments, LSDV144a, and LSDV144b, in vaccine cluster-1.1.1, vaccine-derived cluster 1.1.2 and wild-type cluster-1.2.1 viruses detected from India (Figures 3D,E).

Further, our analysis revealed that the LSDV026 gene was truncated by in-frame non-sense mutations in viruses belonging to wild-type cluster-1.2.2, except for viruses such as AF409137.1/Neethling-Warmbaths/LW, MW656253.1/280-KZN/RSA/2018, MW699032.1 /Russia/Dagestan/2015, MH893760.2/Russia/Dagestan/2015, and KY702007.1/SERBIA/Bujanovac/2016 (Figure 4A). The LSDV026 gene was truncated by in-frame non-sense mutations in MW631933.1_LSDV_LSD viruses belonging to wild-type cluster-1.2.1 and MW435866.1_LSDV_SA-Neethling viruses belonging to cluster-1.1.3 (Figure 4A). Similarly, the LSDV086 (similar to vaccinia virus strain Copenhagen D9R) gene was found to be truncated by in-frame non-sense mutations in viruses belonging to vaccine cluster-1.1.1 except MW656252.1_LSDV/Haden/RSA/195 virus (Figure 4B). The LSDV086 gene was truncated by in-frame non-sense mutations in MN636839.1_LSD-103-GP-RSA-1991 virus belonging to vaccine-derived (Figure 4B). Also, LSDV087 (similar to vaccinia virus strain Copenhagen D10R) gene has been truncated by in-frame non-sense mutations in viruses belonging to vaccine cluster-1.1.1 and SRR21590382, SRR21590384, SRR21590385, and SRR2159038 belonging to wild-type cluster-1.2.2 found in India (Figure 4C). Interestingly, the LSDV114 gene is truncated by in-frame non-sense mutations in viruses other than vaccine cluster-1.1.1, cluster-1.1.5, and recombinant clusters (except MW732649.1_LSDV/HongKong) (Figure 4D). Finally, the LSDV145 (ankyrin repeat protein) gene is truncated by in-frame non-sense mutations in viruses other than cluster-1.1 viruses (Figure 4E), and the LSDV131 (superoxide dismutase precursor) gene is truncated in the majority of the vaccine strains in the cluster-1.1.1 (Supplementary Figure S5A).

In addition, LSDV130 gene was truncated by in-frame non-sense mutations only in OK422492.1/Cattle/India/2019/Ranchi-1/P10, OK422493.1/India/2019/Ranchi-1/P30, OK422494.1/India/2019/Ranchi-1/P50 and OK318001.1/isolate-V28 viruses (Supplementary Figure S5B). Similarly, LSDV057, LSDV081, LSDV154, and LSDV155 genes were found to be truncated by in-frame non-sense mutations only in viruses MT007951.1/Namibia/2016/10F, MH893760.2/Russia/Dagestan/2015, OM984486.1/FJ2019, and OM793603.1/Russia/Khabarovsk/2020& MW732649.1/HongKong/2020 viruses, respectively (Supplementary Figures S5C–F).

It appears that none of the detected frameshift mutations and in-frame nonsense mutations (except LSDV114) are common to all viruses in a particular wild-type cluster-1.2.1, whereas these mutations are common among some of the viruses in different clusters of vaccine, vaccine-derived, and recombinant viruses (Figure 4F). A similar trend was observed in wild-type cluster-1.2.2 genes except for LSDV114, LSDV035, and LSDV019 (Figure 4F). Also, it is noteworthy that there were no detected frameshift mutations and in-frame nonsense mutations present only in all the viruses in the vaccine cluster-1.1.1 (Figure 4F). From these, since these frameshift and in-frame nonsense mutations are common among viruses in different clusters detected at various geographical locations at different times, these frameshift and in-frame nonsense mutations are likely caused by some common factors such as host adaptation, immune evasion, and recombination.

In the earlier sections, we analyzed differences in genes associated with frameshift mutation and in-frame nonsense mutations between different clusters of LSD viruses, so here we aimed to find out what differences exist in genes other than those described above. For this, we included genes that were not analyzed in the earlier section and also did not have open reading frames (ORF) overlaps; further, the genes that are transcribed in forward and reverse directions were analyzed separately in this study. First, we analyzed the nucleotide composition in the coding regions, and this analysis revealed that “AT” occupies around 75% of the nucleotides in genes transcribed in both forward and reverse directions, similar to the whole genome level (Figures 5A–C). From these, it is revealed that there is a bias in “AT” and “GC” in LSD viruses at the complete genome levels and the coding regions. Therefore, we were interested in whether this “AT” bias in coding regions could mediate codon usage bias among different clusters of LSD viruses. For this, we first performed an effective number of codons (ENc) analysis (Zhao et al., 2016; Wang et al., 2018; Desingu and Nagarajan, 2022; Desingu et al., 2022a,b), and this analysis revealed that ENc was around 39 of the genes transcribed in forward and reverse directions of all LSD viruses (Figures 5D,E). From these results, LSD viruses effectively use only 39 codons out of 61 codons in hosts to produce 20 amino acids, so it is clear that LSD viruses have a bias in using host codons. Further, in the analysis of ENc-GC3s plot (Wang et al., 2018; Tian et al., 2020; Desingu and Nagarajan, 2022; Desingu et al., 2022a,b), the genes transcribed in forward and reverse directions of LSD viruses fall slightly below the expected curve (Figures 5F,G), so it could be realized that selection pressure has a little more influence than mutation pressure in the evolution of these genes. Next, we performed PR2-bias analysis to detect AT bias and GC-bias (Tian et al., 2020; Desingu and Nagarajan, 2022; Desingu et al., 2022a,b), in which AT bias [A3/(A3 + T3)] was plotted against GC-bias [G3/(G3 + C3)]. In this PR2-bias analysis, since genes transcribed in the forward and reverse directions of LSD viruses are around 0.5 (Figures 5H,I), A-to-T and G-to-C bias do not significantly account for the bias in codon usage. Furthermore, CT bias [C3/(C3 + T3)] was plotted against GA-bias [G3/(G3 + A3)], and in the analysis conducted to detect CT bias and GA-bias, genes transcribed in forward and reverse directions of LSD viruses were around 0.25 (Figures 5J,K), suggests that C (25%) to T (75%) bias and G (25%) to A (75%) bias largely account for the bias in codon usage. Collectively, from the host codons usage bias, selection pressure, codon third position CT bias, and GA-bias in the coding regions of LSD viruses, it could be inferred that LSD viruses are more likely to mutate to adapt to the new host’s synonymous codon usage and achieve evolutionary development if they undergo host-jump or (or) pass through unnatural hosts for virus attenuation.

After this, we were interested in finding out whether LSDV-vaccine strains attenuated by the passage in the unnatural host or the unnatural host’s cell culture have evolved to adapt to the usage of synonymous codons of the unnatural host. For this purpose, we performed a dN/dS analysis comparing LSDV wild-type NCBI reference strain NC_003027.1_LSDV_NI-2490 with viruses from other clusters. This dN/dS analysis revealed negative/purifying selection in genes transcribed in forward and reverse directions in clusters of vaccine, vaccine-derived, and recombinant viruses compared to wild-type (NC_003027.1) (Figures 6A,B). When comparing vaccine, vaccine-derived, and recombinant viruses with wild-type (NC_003027.1) dN/dS is around 0.1 (Figures 6A,B), it is clear that most of the total mutations in these viruses are synonymous codon mutations. Further, synonymous and nonsynonymous mutations in each cluster were subjected to in-depth analysis. In this analysis, it was revealed that the presence of synonymous and nonsynonymous mutations in genes transcribed in forward and reverse directions in LSD viruses was almost equal, and specifically, synonymous mutations were more abundant than nonsynonymous mutations in all clusters compared to wild-type (NC_003027.1) (Figures 6C–T). Also, it was revealed that synonymous and nonsynonymous mutations in vaccine and vaccine-derived viruses are around 500 and 175, respectively, whereas synonymous and nonsynonymous mutations in recombinant viruses are around 300 and 100, respectively (Figures 6C–T). In addition, these synonymous and nonsynonymous mutations can be seen to increase from central to terminal genes in genes that are transcribed in forward and reverse directions in LSD viruses (Figures 6C–T). Notably, genes in the terminal part of the genome of poxviruses generally have high genetic diversity, and these genes play an essential role in host range, host adaptation, evasion of the host immunity, pathogenicity, and virulence (Biswas et al., 2020; Senkevich et al., 2020, 2021; Brennan et al., 2022).

Next, we were interested in identifying nonsynonymous mutations in the genes transcribed in the forward and reverse directions of viruses in the vaccine, vaccine-derived, and recombinant clusters compared to the wild-type virus. In this analysis, nonsynonymous mutations increased from central to terminal genes in genes transcribed in forward and reverse directions of viruses in the vaccine, vaccine-derived, and recombinant clusters compared to wild-type viruses (Supplementary Figures S6, S7). Further, it was observed that the nonsynonymous mutations unique to the viruses in the recombinant cluster were not at a significant level and were a mixture of wild-type cluster-1.2.1 and vaccine cluster-1.1.1 (Supplementary Figures S6, S7). Also, since the nonsynonymous mutations found in the viruses in the wild-type cluster-1.2.2 are mainly absent in the viruses in the vaccine-derived and recombinant clusters (Supplementary Figures S6, S7), it could be realized that the viruses in the wild-type cluster-1.2.2 are evolving in a different direction from the viruses in the vaccine, vaccine-derived, and recombinant clusters.

Overall, it could be felt that the attenuated vaccine strains have evolved possibly through purifying selection for host adaptation by attaining the majority of mutations in synonymous codons as adapted to the codon usage of unnatural hosts. Also, viruses in wild-type cluster-1.2.2 have a purifying selection compared to wild-type cluster-1.2.1, and LSD viruses affect animals such as cattle, buffaloes, springbok, impala and giraffe (Young et al., 1970; Le Goff et al., 2009; Namazi and Khodakaram Tafti, 2021); this purifying selection suggests that possibly host adaptation has resulted in the majority of mutations in synonymous codons adapted to the codon usage of these or other animal hosts.

In the previous sections, in the attenuation and evolution of LSD viruses, mostly synonymous codons are evolved, and CT-bias and GA-bias are present in the third nucleotide position of codons, so here we are interested in finding out the mechanism by which such mutations are acquired. We investigated the nucleotide mismatches in genes transcribed in forward and reverse directions of LSD viruses in different clusters compared to wild-type NCBI reference strain NC_003027.1_LSDV_NI-2490. In this analysis, it can be realized that there are more nucleotide mismatches in the vaccine and vaccine-derived clusters, and these nucleotide mismatches are increasing from the central part toward the terminal part of the genes that are transcribed in the forward and reverse directions of LSD viruses (Figures 7A,B). Further, our analysis revealed that most of these nucleotide mismatches are silent mutations (Supplementary Figures S8A,B) and are caused by transition mutations (Figures 7C,D). Also, it is of increasing importance that around 80% of the mutations in genes transcribed in forward and reverse directions of LSD viruses of all clusters compared to wild-type (NC_003027.1) are transition mutations (Figures 8A,B). Furthermore, our results show that the G → A & C → T transition mutations fraction is around three times higher than the A → G & T → C transition mutations fraction in genes transcribed in forward and reverse directions of LSD viruses with all clusters compared to wild-type (NC_003027.1) (Figures 8C,D). Interestingly, G → A (or) C → T mutations can be generated by the host’s APOBEC enzymes. It is noteworthy that these APOBEC enzymes play an essential role in the restriction of retroviruses (HIV), DNA viruses such as monkeypox virus (Mpoxv), hepatitis B virus (HBV), and human papillomavirus (HPV) (Bonvin et al., 2006; Bulliard et al., 2011; Warren et al., 2015; Gigante et al., 2022; Isidro et al., 2022; Pecori et al., 2022). Since DNA editing by APOBEC enzymes is based on TC > TT (or) GA > AA, GG > AG (or) CC > CT, and AC > AA (or) GT > AT motifs (Gigante et al., 2022; Isidro et al., 2022; Pecori et al., 2022), we were interested in finding out what motif mutations are present in genes transcribed in forward and reverse directions of viruses in all clusters compared to wild-type (NC_003027.1). Genes transcribed in the forward directions of viruses in the vaccine, vaccine-derived, and recombinant clusters revealed a higher abundance of AC > AA & GT > AT motif mutations compared to wild-type (NC_003027.1) (Figure 8E; Supplementary Figure S9A). On the other hand, AC > AA & GT > AT and TC > TT & GA > AA motif mutations were found to be almost higher abundance in genes transcribed in reverse directions of viruses in the vaccine, vaccine-derived, and recombinant clusters compared to wild-type (NC_003027.1) (Figure 8F; Supplementary Figure S9B). After this, we were interested in finding out how the mutations created by these APOBEC enzymes compared to the wild-type (NC_003027.1) at the complete genome level of the viruses in the vaccine, vaccine-derived, and recombinant clusters. This analysis revealed that almost 80% of the mutations in the viruses in the vaccine, vaccine-derived, and recombinant clusters were transition mutations compared to the wild-type (NC_003027.1) at the whole genome level (Figure 8G), as in the coding regions (Figures 8A,B). Also, as in the coding regions, it was revealed that the G → A & C → T transition mutations fraction was almost three times higher than the A → G & T → C transition mutations fraction in the viruses in the vaccine, vaccine-derived, and recombinant clusters when compared to the wild-type (NC_003027.1) at the complete genome level (Figure 8H), and AC > AA & GT > AT and TC > TT & GA > AA motif mutations were also found to be more prevalent (Figure 8I; Supplementary Figure S9C). Interestingly, it is noteworthy that AC > AA & GT > AT motif mutations are edited by APOBEC1 enzyme present in tetrapod to humans, whereas TC > TT & GA > AA, GG > AG & CC > CT motif mutations are edited by APOBEC3 enzyme present in placental mammals (Gigante et al., 2022; Pecori et al., 2022). Overall, the viruses in the vaccine clusters have a large number of transition mutations at the complete genome and coding region level compared to the wild-type virus, and the G → A & C → T transition mutations fraction is higher in these transition mutations, and the G → A & C → T mutations are in motif mutations that are genome editing by host APOBEC enzymes. Also, since transition mutations in coding regions are mostly silent mutations, it is revealed that APOBEC enzymes are the dominant driver in the evolution of host codon usage adaptation of these vaccine viruses.