Skip to main content
Springer logoLink to Springer
. 2026 Oct 16;171(11):300. doi: 10.1007/s00705-026-06683-5

Insights into goatpox virus and sheeppox virus genomes from pangenome graphs

Tim Downing 1,✉
PMCID: PMC13675261  PMID: 42888350

Abstract

The capripoxviruses comprise three species: goatpox virus (GTPV), sheeppox virus (SPPV) and lumpy skin disease virus (LSDV). They are large double-stranded DNA viruses with highly conserved core genomes and variable terminal regions. Previous studies have described variation in Capripoxvirus gene content, their broader population structure and the contribution of non-coding and structural variation remains opaque. This study investigated the genomic diversity and evolutionary history of GTPV and SPPV using phylogenetics, pangenome variation graphs (PVGs), and gene-specific analyses. We found clear differences in population structure between the two viruses. GTPV had three deeply divergent and genetically stable lineages with limited evidence of recent gene flow, whereas SPPV had weaker clade separation consistent with an ancestral bottleneck followed by recent population expansion. PVG-based analyses indicated that GTPV has a comparatively closed pangenome, while SPPV’s was open, particularly at the inverted terminal repeats (ITRs). Structural and haplotype variation was concentrated at these ITRs, which moderate host immunity and specificity. In several lineages, extended putative ORFs spanning adjacent ITR genes were observed, indicating recurrent structural plasticity at these regions. Patterns of gene-specific conservation and divergence highlighted loci under strong constraint and lineage-specific structural changes that may contribute to host specificity. Together, these results demonstrated how graph-based genome models complement gene-based analyses in resolving poxvirus genome evolution and provide a resource for improved comparative and population genomic studies of large DNA viruses.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s00705-026-06683-5.

Significance

The capripoxviruses are economically important livestock pathogens, yet the genomic mechanisms underlying their diversification and host specificity remain poorly resolved. By applying pangenome variation graphs alongside phylogenetic and gene-level analyses, this study reveals fundamental differences in how goatpox and sheeppox viruses have evolved. Goatpox virus had a deeper, more stable lineage structure, whereas sheeppox virus was more recent and diverse. Importantly, structural variation at the inverted terminal repeats emerged as a major driver of genomic diversity, including lineage-specific haplotypes and variable gene structures. These findings demonstrated the value of graph-based genome representations for resolving complex variation in large DNA viruses and provides approaches for improving genomic surveillance, comparative analyses, and future investigations into host range, virulence and tropism.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s00705-026-06683-5.

Introduction

Goatpox virus (GTPV) and sheeppox virus (SPPV) belong to the Capripoxvirus (CaPV) genus in the family Poxviridae and primarily infect goats and sheep as hosts, respectively. The other species in the CaPV genus is the lumpy skin disease virus (LSDV) whose preferential host is cattle. Their International Committee on Taxonomy of Viruses (ICTV) names are Capripoxvirus goatpox, Capripoxvirus sheeppox and Capripoxvirus lumpyskinpox, respectively. Collectively, the CaPV are re-emerging transboundary dsDNA viruses that impair animal health, and consequently cause significant economic harm [1, 2]. They are all notifiable diseases according to the World Organization for Animal Health (WOAH) WOAH [3] because of their wide geographic spread and their economic importance. GTPV and SPPV are endemic across parts of Africa, the Middle East, and Asia, and phylogenetic analyses identify geographically associated sub-clades in these regions. Additionally, SPPV has deep ancestry within Europe dating back several centuries [4], with possible replacement by Clade 3.3 (also called clade A1), which has spread into southern Europe in recent years) [5]. Consequently, the potential for GTPV and SPPV to cause disease outbreaks is a global challenge in relation to animal welfare, economic stability and transboundary spread.

GTPV and SPPV can survive in the environment for long periods of time, and can transmit via contact, unlike LSDV that typically needs an insect vector [6]. GTPV can infect cattle and sheep as well, though less effectively [7, 8]. In the same way, SPPV can experimentally infect goats and cattle, and LSDV goats and sheep. There is vaccine cross-protection between GTPV and SPPV contrasting with the partial protective effect of LSDV vaccination against SPPV, and likewise for SPPV vaccination against LSDV [9]. This has resulted in the wider use of GTPV as a CaPV vaccine component thanks to its broader cross-protective profile, such as reducing the risk of LSDV infection [10].

CaPV possess a linear genome with lengths of about 150 Kb, are AT-rich (75%), and typically have ~ 147–156 orthologous genes [11]. The estimated interspecies genome-wide similarity is about 96%, stemming from the ancient diversification of LSDV, GTPV and SPPV [12]. All share diverse inverted terminal repeats (ITRs) at both genome termini whose genes moderate host range, interactions and virulence [13, 14]. These ITRs form complementary hairpin structures at both ends, which are essential for concatemer resolution and the formation of mature genomes during replication [15]. These ITRs are also structurally dynamic, frequently exhibiting deletions, duplications, and rearrangements that are associated with attenuation, host adaptation, and vaccine derivation. The 5’ (LSDV ORFs 1–23) and 3’ (LSDV ORFs 124–156) are more variable and typically support functions linked to immunomodulation and virulence. The intervening genomic region (LSDV ORFs 24–123) is highly conserved and encodes core replicative and structural machinery [11, 12, 16]. In addition, there are nine genes in LSDV that are pseudogenised or truncated in GTPV and SPPV that moderate virulence and host range [12].

The genome evolution of CaPV remains understudied, particularly for GTPV and SPPV. To address this, this study uses pangenome variation graphs (PVGs) (also known as pangenome graphs or sequence graphs) to illuminate genomic diversity and the evolution of GTPV and SPPV in all available high-quality genome assemblies [17]. By representing multiple genomes as paths through a shared graph structure, PVGs enable the joint analysis of SNP, indel and structural variation [18, 19]. Here, a gene-level analysis of genome-wide diversity in both species was created to probe their population structure and genetic variability, including the ITRs. Additionally, this study linked GTPV and SPPV evolution to that of LSDV using a consistent nomenclature throughout based on the Oman LSDV reference genome [20] to provide a consistent framework for genes that are intact in LSDV but altered in GTPV and SPPV. Lastly, representative PVGs were constructed for both species, with potential to improve mutation detection during read mapping of short read CaPV sequencing libraries. Together, this work provided a graph-based perspective on GTPV and SPPV genome evolution and highlighted how differences in population structure, terminal genome plasticity, and lineage-specific variation may contribute to host specificity and virulence.

Materials and methods

Genome assembly collection, alignment and phylogenetic analysis

All available complete and valid GTPV (n = 14) and SPPV genomes (n = 30) were downloaded from the NCBI nucleotide database [21] (5th August 2025) (Table S1). KT438551.1 and KT438550.1 were also two listed SPPV complete genomes, but were excluded because their genome assembly quality was low.

This initial dataset was aligned with Mafft v7.453 [22] using automatic optimisation and default parameters. Phylogenetic reconstruction of the samples’ evolutionary relationships was conducted using RAxML-NG (Randomised Axelerated Maximum Likelihood) v1.2.0 [23] with a GTR (general time reversible) model and gamma substitution rate heterogeneity selected as the best-fit model by modeltest-ng [24]. Bootstrap support was calculated using both Felsenstein Bootstrap Proportion (FBP) and Transfer Bootstrap Expectation (TBE) methods with 1000 replicates where support values ≥ 90% were displayed. These phylogenies were mid-pointed rooted and visualised using R v4.3.2 [25] packages ape v5.8-1 [26], phangorn v2.12.1 [27], treeio v1.28.0 [28], ggtree v3.12.0 [29] and ggplot2 v2_3.5.2 [30]. Principal components analysis (PCA) was implemented using R based on the genome-wide SNP data. As above, the corresponding sequences for each gene were extracted, aligned with mafft, their phylogenies were reconstructed with RAxML. PCA was implemented to detect population structure patterns not apparent in the phylogenies.

Pangenome variation graph creation, analysis and annotation

Panalyze v1.0 [31] implemented PVG construction, analysis and annotation of these GTPV (n = 14) and SPPV (n = 30) genomes. Panalyze created PVGs in Graphical Fragment Assembly (GFA) format using pangenome graph builder (PGGB) v0.4.0 [32] and wfmash v0.7.0 [33], with Multiqc v1.14 [34] to collate and assess PVG metrics. Panalyze analysed these PVGs using ODGI (optimized dynamic genome/graph implementation) v0.8.3 [35] to extract SNPs, indels and compound mutations using VG from the PVG paths [36]. Subgraphs were selected using gfatools v0.4 [37], evaluated with gfastats v1.3.6 [38] and visualised with VG. Panalyze visualised the PVGs with ODGI and ggplot2 v2_3.4.4 [30]. Panalyze used gfautil v0.3.2 [39] to get mutation rates, coordinates for samples at each path and site, and VCF files. Mutation density and population genetic statistics were computed for the 5’ end (100 − 13,850 bp), core (13,851 − 106,910 bp) and 3’ end (106,910 to 150,000 bp) as per previous work [16].

The PVGs were annotated using Prokka [40] and indexed with SAMtools v1.19 [41]. Using the reference genome annotation for NC_004002 and NC_004003, BUSCO v5.8.0 [42] showed all GTPV and SPPV had all 18 expected BUSCO poxvirus genes. The annotated GTPV, SPPV and LSDV genomes were aligned and compared using NC_004003, NC_004002 and PV877838 [20] to extract gene boundaries for CDSs present in LSDV that were absent in GTPV and SPPV (Table S2). The Prokka GFFs were converted to gene transfer format (GTF) with gffread [43] and parsed. The annotation position information was merged with the ODGI rendering of the PVGs, creating one CSV annotation file per PVG, which was visualised using Bandage v0.8.1 [44].

Pangenome size and growth estimation

Panalyze inferred PVG growth using a sequence-based approach with Panacus v0.2.3 using default coverage thresholds that included all segments [45], and a k-mer-based method in pangrowth [46]. Both estimated the total number of mutations (n) in a sample collection (N) for an estimated parameter gamma as: n = ĸNgamma, where gamma reflected a closed PVG if negative and an open PVG if positive based on Heap’s Law, and ĸ was a fitting parameter. They also estimated the first derivative of this, the rate of new mutations per sample (Δn) as: Δn = kN−alpha, which is based on alpha < 1 denoting an open PVG and alpha > 1 a closed one, and k was a fitting parameter [47]. This modelled the rate of addition of new nodes or k-mers as each sample was added to the PVGs. Here, a new nodes or new k-mer indicated a new mutations Pangrowth also estimated the core PVG size based on k-mer differences. Panalyze also used ODGI heaps to infer PVG growth using 1000 simulations.

Community detection and visualisation

Community detection was performed on PVGs, where these communities reflect the relative similarity of samples based on PVG data. Here, these were identified with Panalyze using wfmash and PGGB with a k-mer of 19 bp, a window size of 67 bp, a threshold of 80% sequence mash-level identity, and ten mappings per node. Panalyze evaluated these communities with PGGB scripts that built a network with the contigs as nodes and mappings as weighted edges, where the latter were proportional to the mapping’s similarity and length. Panalyze visualised these networks with igraph tools [48]. The mappings threshold was the number of detected haplotypes minus one for the given region.

Tests for selective processes

We tested for selection using the direction of selective (DoS) metric expressed as: DoS = DN/(DN+DS) - PN/(PN+PS) [49] where DN and DN stand for the number of ancestral nonsynonymous and synonymous substitutions (respectively) between the GTPV and SPPV species, and PN and PS stand for the number of current population-level (within-species) nonsynonymous and synonymous mutations (respectively). A DoS value near 0 is consistent with neutrality, DoS < 0 with purifying selection, and DoS > 0 with an excess of nonsynonymous fixed differences consistent with adaptive evolution. This analysis included an existing dataset of 121 LSDV genomes for comparison [50].

Results

Contrasting levels of population structure within goatpox virus (GTPV) and sheeppox virus (SPPV)

Pangenome variation graph (PVGs) were created for available GTPV (n = 14) and SPPV (n = 30) genomes to explore their genomic diversity (Table S1). GTPV and SPPV had marked differences in their population structures and evolutionary histories, consistent with previous work [51]. GTPV had three separate clades named 2.1 (n = 6 samples), 2.2 (n = 5) and 2.3 (n = 3) based on phylogenetic, PCA, network-based interpretation of genome-wide SNPs (Fig. 1A). Repeating this for the core region, 5’ end and 3’ end reiterated this overall topology (Figure S1). Here, an existing clade naming system was used [52]. In GTPV, the first PC (56% of variation) split clades 2.1 and 2.2 from 2.3, and the PC2 (30%) separated all three clades (Fig. 1B). This suggested long-term genetic differentiation of these GTPV clades, with no evidence of recent admixture in this dataset (Fig. 1C). Clade 2.3 was the only clade without vaccine-related specimens and contained the sole African sample (MN072624), together with two Middle Eastern isolates. As shown below using LSDV-rooted phylogenies, this clade also had the deepest evolutionary root. A genome-wide comparison of GTPV vaccine-related versus field samples found no SNPs differentiating these groups. With Clade 2.2, comparison of the two vaccine-linked samples against the other three identified 23 differentiating SNPs: 21 at the 5’ ITR near the hairpin sequence, another at LSDV037 (30,904 bp in NC_004003) and one at LSDV066 (56,939 bp).

Fig. 1.

Fig. 1

Genome-wide SNP diversity in the GTPV and SPPV PVGs. (A) A midpoint-rooted GTPV phylogeny showing three clades: 2.1 (grey, n = 6), 2.2 (purple, n = 5) and 2.3 (orange, n = 3). The scale bar shows the numbers of substitutions per site. Nodes with bootstrap support > 90% are shown. (B) A GTPV PCA where PC1 represented 56% of variation, and PC2 30%. (C) A GTPV network where each node represents one sample. (D) A midpoint-rooted SPPV phylogeny showing three clades: 3.1 (green, n = 2), 3.2 (blue, n = 9) and 3.3 (red, n = 18). (E) A SPPV PCA where PC1 had 31% of variation, and PC2 had 25% of it, and MN072628.1 (grey) was likely a rare lineage. (F) A SPPV network where each node represents one sample

SPPV had three clades based on the phylogenetic patterns, PCA and networks of PVG data: 3.1 (n = 2 samples), 3.2 (n = 9) and 3.3 (n = 18 samples) (Fig. 1D). Within all three SPPV clades, there was a high level of genetic variability that was informed by PCA. Repeating this for the core region, 5’ end and 3’ end reiterated this overall pattern (Figure S2). SPPV PC1 (31%) and PC2 (25%) together showed high diversity within Clade 3.3 (Fig. 1E), especially for MW167070, MW020571, MN072630, MN072631, AY077833 and MT137384 that were separate from the main 3.3 cluster. SPPV PC1 and PC2 separated Clade 3.2 into two paraphyletic clusters with three (3.2.1: MW167071, OR239060, MN072629) and six (3.2.2: AY077832, NC_004002, PP886236, PP886237, PP886238, PP886239) samples. Clade 3.1 was genetically distinct and was composed of two vaccine-related samples. Clade 3.3 was spread across Asia and the Middle East. SPPV Clades 3.2 was found in the Middle East, of which subclade 3.2.2 has spread more recently into southern Europe. A genome-wide comparison of SPPV vaccine-related versus field samples found no SNPs differentiating these groups, nor any when this was applied to Clade 3.3 alone.

The sole SPPV genome from Africa (MN072628) was not assigned to any of the other clades. The sample had a central position in the PCA (Fig. 1E) and had a long external branch (Fig. 1D). MN072628 had similarity to Clade 3.1, four samples from Clade 3.3 (OQ434235, MN072631, MN072630, MW020571.) and four from Clade 3.2 (PP886236, PP886237, PP886239, NC_004002) in the PVG network (Fig. 1F). PC3 (10%) and PC4 (6%) separated Clade 3.3 and MN072628, suggesting that it was a rare lineage, which was also supported by simplot analysis showing no notable variation in genome-wide similarity with NC_004002 (Figure S3). This also pointed to a sampling gap and the need for more SPPV genomes from the African continent.

The rooting of these GTPV and SPPV genomes was explored using LSDV sample OQ555660.1, which recapitulated the same GTPV and SPPV phylogenetic structures as previous work [52] (Figure S4). This was supported by phylogenetic analysis of a larger sample of (n = 116) LSDV genomes combined with GTPV and SPPV ones (Figure S5). This showed that the population structure was weaker and more recent in SPPV, such that the scale of diversity in SPPV was comparable to a GTPV clade. GTPV Clade 2.3 had the most evolutionary deep root, suggesting the differences within GTPV may inform host specificity given that some GTPV isolates infect sheep as well as goats [52].

GTPV and SPPV pangenome variation graph structure, openness and diversity

Although there were fewer GTPV samples, its complete PVG length was almost as long, spanning 160,732 bp compared to SPPV’s 161,135 bp. The GTPV PVG had more nodes than the SPPV PVG (7,801 vs. 3,044) and more edges (10,644 vs. 4,208). The median PVG length was 147,684 bp for GTPV and 148,183 bp for SPPV, reflecting the median genome size (Figure S6) that was marginally smaller than LSDV. The shared PVG for GTPV was smaller than for SPPV (142,411 bp vs. 144,593 bp), implying that if more GTPV genomes were acquired, its shared PVG would shrink, whereas for SPPV it would not.

The distinct GTPV population structure meant it had a closed PVG (alpha = 1.76 from Pangrowth), indicating that if more GTPV genomes were sampled, a small number of new mutations would be discovered (Figure S7). These estimates were confirmed with Panacus for GTPV (alpha = 1.63) (Figure S8). In contrast, the SPPV PVG was open (alpha = 0.69) (Figure S7), suggesting that many new mutations would be found if more SPPV genomes were examined. The Panacus SPPV value was more marginal (alpha = 1.01) (Figure S8), suggesting that moderate SPPV sampling may have been achieved. Nonetheless, the lack of samples from African may be limiting for both species.

Representative PVGs for GTPV and SPPV were created using a genome from each main clade (Table 1) using Panalyze [31]. These allow more resolved inference from mapping of read libraries than a linear reference genome [50]. For GTPV, the representative PVG totalled 152,917 bp, with 6,865 nodes and 9,296 edges. It was made from three genomes with 6,247 mutations between them: field sample MN072621 from Clade 2.1, vaccine specimen AY077836 from Clade 2.2 (149,673 bp with 155 CDSs) [12], and field isolate MN072625 from Clade 2.3 [52] (Table 1). NC_004003 (149,599 bp with 153 CDSs) was not used because it lacked sequence corresponding to LSDV155 and LSDV156 at the 3’ end (LSDV001 and LSDV002 were present at its 5’ end).

Table 1.

The genome used to create representative PVGs for GTPV and SPPV. MN072628.1 was a unique SPPV sample and was not assigned to a clade

Species Accession Clade Genome length (bp) Number of CDSs Isolation
Year Country Type
GTPV MN072621.1 2.1 150,308 156 2005 Vietnam Field
AY077836.1 2.2 149,673 155 2000 Kazakhstan Vaccine
MN072625.1 2.3 150,408 153 1983 Yemen Field
SPPV MN072626.1 3.1 148,534 155 NA Iraq Vaccine
OR239060.1 3.2 150,441 151 2017 United Arab Emirates Field
AY077834.1 3.3 149,612 154 1994 Kazakhstan Vaccine
MN072628.1 NA 149,906 149 1977 Nigeria Field

The representative SPPV PVG spanned 151,016 bp with 1,611 nodes and 2,235 edges, making it shorter and less diverse than the GTPV PVG. It had four genomes with 1,596 mutations among them (Table 1): vaccine-related sample MN072626 from Clade 3.1, field sample OR239060 from Clade 3.2, vaccine-related specimen AY077834. from Clade 3.1, and African sample MN072628 to represent less-sampled diversity. NC_004002 (149,905 bp) was not used because it only had 148 annotated genes, suggesting that some ORFs were missing or mutated.

Genome-wide variability measured based on sequence differences in the PVGs showed higher diversity in GTPV than SPPV, and this was surprisingly higher at CDS regions compared to non-CDS ones (Fig. 2). However, both had a shared pattern of higher diversity at the 5’ (< 14 Kb) and 3’ (> 107 Kb) ends but less variation in the core region, like previous LSDV work [16] (Table 2). The 5’ and 3’ ends had 23% and 22% more mutations/Kb than the core genome for GTPV, and 83% and 75% more for SPPV (Table 2). Watterson and Wu’s theta per Kb [53] was greater than nucleotide diversity (pi) per Kb genome-wide for both species, and for GTPV Clades 2.1 (38.0 vs 29.0) and 2.2 (28.2 vs 23.5) as well as SPPV Clades 3.2 (12.5 vs 7.6) and 3.3 (19.4 vs 7.4). This pattern was consistent across the genomic regions, resulting in negative Tajima’s D [54] values for GTPV (−2.48) and SPPV (−2.87). This excess of low-frequency variants could be caused by numerous processes, included a recent population expansion.

Fig. 2.

Fig. 2

Genome-wide mutation density (mutations/Kb, y-axis) in GTPV (blue) and SPPV (red) highlighting regions associated with CaPV differentiation (LSDV002, LSDV004, LSDV009, LSDV013, LSDV026, LSDV132, LSDV136, LSDV153, LSDV155 in orange) and attenuated virulence (purple) encoding: ankyrin-repeat proteins (LSVD012, LSVD145, LSVD147, LSVD148); kelch-like proteins (LSDV019, LSDV144, LSDV151), a B22R-like protein (LSDV134), or host range factors (LSDV067, LSDV141). The SPPV and GTPV gene annotation across the genomes is represented below the main plot where grey boxes indicated each gene that are read in the forward (lower strip) or reverse (upper strip) directions. The annotation used was from PV877838, NC_004002 and NC_004003

Table 2.

Sequence diversity metrics for GTPV and SPPV. Pi is the nucleotide diversity (the mean number of pairwise SNPs). Theta was calculated as Watterson and Wu’s theta

Virus Region Mutations/Kb Pi/Kb Theta/Kb
GTPV Genome 162 23.7 52.1
GTPV 5’ 184 27.5 60.5
GTPV core 149 21.7 47.7
GTPV 3’ 182 26.7 58.8
SPPV Genome 100 7.0 25.8
SPPV 5’ 141 10.3 38.2
SPPV core 77 5.2 19.3
SPPV 3’ 135 9.8 36.0

Divergent haplotypes at the ITRs highlight novel variation within SPPV

Poxvirus ITRs are repetitive, structurally complex regions underpinning viral replication and virulence that are also associated with CaPV differentiation [12]. GTPV and SPPV lack functional LSDV002, LSDV004, LSDV153 and LSDV155 ORFs [12]. Here, the 5’ ITR of the GTPV PVG had more nodes than SPPV one (226 vs 138) due to its higher diversity (Figure S9). Using LSDV to obtain homologous GTPV and SPPV regions (Table S2), GTPV Clades 2.1 and 2.2 had a 533 bp haplotype at the LSDV002 region (that is similar to MYXV gene M003.2) that differed from a rarer Clade 2.3 487 bp haplotype (Fig. 3). Two clade 2.3 samples (MN072624 and MN072625) had truncated LSDV004 ORFs, and MN072625 had a truncated LSDV153 as well. Despite this, LSDV005 (BCRF1, 512 bp) adjacent to the 5’ ITR was highly conserved in all samples. At the 3’ ITR, Clades 2.1 and 2.2 had a 622 bp haplotype at the region homologous to LSDV153 differing from a 555 bp Clade 2.3 haplotype (Fig. 3). These divergent 3’ ITR haplotypes’ had a high level of differentiation (similarity of 79%).

Fig. 3.

Fig. 3

Top: The (A) GTPV 5’ ITR, (B) SPPV 5’ ITR, (C) GTPV 3’ ITR, and (D) SPPV 3’ ITR. (A) The GTPV PVG region homologous to LSDV002 had two divergent haplotypes, separating at node 57 and merging at node 104. One haplotype was 533 bp (in Clades 2.1/2.2) and the other was three nodes spanning 487 bp (in Clade 2.3). (B) The SPPV PVG region homologous to LSDV002 had two divergent haplotypes, separating at node 55 and merging at node 109. One haplotype spanned 1,201 bp (Clade 3.1) at five nodes that had a truncated LSDV003. The other was 48 nodes spanning 1,304 bp and encoded a complete LSDV003 (Clades 3.2/3.3). (C) The GTPV LSDV153 region had two divergent haplotypes: one common to Clades 2.1/2.2 was 622 bp across 50 nodes, and a Clade 2.3 one that was 555 bp across 16 nodes. (D) The SPPV LSDV155 region had two divergent haplotypes. One spanned 1,201 bp (Clade 3.1) across five nodes and had a truncated LSDV154. The other spanned 53 nodes totalling 1,306 bp (Clades 3.2/3.3) with a complete LSDV154. Bottom: SequenceTubeMap images of the PVG ITR regions. (E) The GTPV PVG at 759 to 1,230 bp using the NC_004003 coordinates showing the divergent haplotype in the three Clade 2.3 samples (top three rows). (F) The SPPV PVG at 147,907–972 bp using the NC_004002 coordinates showing the divergent haplotype in the Clade 3.1 samples (top two rows)

The SPPV PVG had two divergent haplotypes at the 5’ ITR: one 1,201 bp in Clade 3.1, and the other 1,304 bp (in Clades 3.2 and 3.3) (Fig. 3). These haplotypes were at bases 53–683 of the LSDV003 genes coordinates, and had 86% similarity due to numerous rearrangements, which contrasted with 98–99% similarity for adjacent regions. Vaccine-related Clade 3.1 had a truncated non-functional LSDV003, unlike Clades 3.2 and 3.3. At the 3’ ITR, there were two divergent haplotypes with lengths 1,201 bp (Clade 3.1) and 1,306 bp (Clades 3.2/3.3) in which only the latter encoded a complete LSDV154 (Fig. 3). Within SPPV clades 3.2 and 3.3, intra-clade variation produced two alternative putative ORFs: a shorter LSDV003 in 24 samples and all of GTPV and LSDV, and a longer LSDV003 in SPPV six samples with 124 additional amino acids associated with a 46 bp indel (Figure S10). These had identical homology at the 3’ end, implying short and long LSDV154 ORFs in these samples. The longer LSDV003 and LSDV154 isoforms were in four Clade 3.2 (OR239060, AY077832, NC_004002, MN072629) and two Clade 3.3 (PV167793, PV434148) samples.

Evidence for selective processes was evaluated at each gene by comparing the ancestral DN and DS numbers between these two species to those within each species for GTPV and SPPV (PN, PS). At the ITRs, this showed high conservation at LSDV001 and LSDV156, which had no nonsynonymous mutations in any of the GTPV and SPPV samples (Table S3). LSDV001 and LSDV156 promote host IL-1B-related and TNFa-related signalling pathways that facilitate skin nodule formation [55]. The regions similar to LSDV002 and LSDV155 had ancestral conservation as well as no synonymous changes but numerous nonsynonymous ones in GTPV (PN/PS 3/0 for each gene) and SPPV (PN/PS 13/0). These differences stemmed from the high diversity between the haplotypes at this region. LSDV003 and LSDV154 also had ancestral conservation, and in GTPV there similar levels of nonsynonymous and synonymous change compared to the genome-wide ratio of 0.98. However, in SPPV both had a high rate of nonsynonymous changes (PN/PS 10/0 for both) arising from the divergent haplotypes (Fig. 3). LSDV004 and LSDV153 encode secreted immunomodulatory virulence factors that are expressed early during infection [12]: the regions similar to LSDV004 and LSDV153 had high rates of nonsynonymous changes stemming from pseudogenisation (DN/DS 10/1 and 26/2, respectively). LSDV004 had moderate conservation within GTPV (PN/PS 13/19) and SPPV (PN/PS 1/1), as did LSDV153 in SPPV (PN/PS 0/0) but not GTPV (PN/PS 8/0), which was associated with two divergent haplotypes (Fig. 3).

Genetic variation at genes associated with CaPV species specificity

To explore genetic variation across GTPV and SPPV, diversity at each gene was examined using the LSDV gene nomenclature for consistency, including regions encoding genes in LSDV that are not CDSs in GTPV or SPPV (Figure S11). GTPV genes had more SNPs/Kb than SPPV ones (median 15.3 vs. 5.0) and more mutations/Kb/sample (mean 4.66 vs. 0.91) (Table S4). The SNPs/Kb (r = 0.53) and mutations/Kb (r = 0.46) were positively correlated between GTPV and SPPV, particularly at three genes with high levels of diversity in both species: LSDV130, LSDV131 encoding superoxide dismutase-like (SODC) protein that may be a virulence factor [56], and LSDV152 encoding an ankyrin repeat protein (Figure S12). In SPPV, 18 genes were conserved (0 or 1 SNPs: LSDV023, LSDV031, LSDV040, LSDV046, LSDV052, LSDV055, LSDV060, LSDV067, LSDV072, LSDV078, LSDV092, LSDV097, LSDV104, LSDV106, LSDV111, LSDV113, LSDV120, LSDV126). Only one gene in GTPV had no SNPs (LSDV105), and five others had < 5 SNPs/Kb (LSDV044, LSDV093, LSDV095, LSDV096, LSDV107). These could represent genes involved in essential GTPV- and SPPV-specific processes.

A genome-wide scan for selective processes identified additional patterns, including a comparatively low correlation between the GTPV and SPPV DoS metrics (rho = 0.23, p = 0.0087) and evidence of long-term conservation across genes (median DN/DS =0.63). Although the median DoS values were similar (GTPV 0.031 vs. SPPV − 0.038), the low correlation was driven by a smaller standard deviation (SD) in the DoS values for GTPV compared to SPPV (0.26 vs. 0.39). In addition, there was a higher PN/PS ratio within SPPV compared to GTPV (0.98 vs. 0.56), which could be associated with the relatively more ancient origin of GTPV allowing more time for selection to remove deleterious nonsynonymous alleles. Four genes had DoS > 0.291 in GTPV and DoS > 0.36 in SPPV (corresponding to the respective median + SD values): LSDV004, LSDV023, LSDV068 and LSDV129. In addition to these, five GTPV and 12 SPPV genes met the species DoS criteria (Supplementary Text). Among the latter 12, LSDV087 and LSDV141 were previously associated with a SPPV Clade 3.3 outbreak in Russia [57]: LSDV087 encodes a mutF motif protein that decaps mRNAs, and LSDV141 encodes an extracellular enveloped virion (EEV) membrane protein that may support viral formation and spread.

Genetic variation at non-ITR genes associated with CaPV species specificity

Our PVG-based approach permitted insights into haplotype-level patterns at key genes of interest. We focused on five LSDV non-ITR ORFs that are inactivated in GTPV and SPPV, which have also been associated with moderating host immune responses (LSDV009, LSDV013, LSDV026, LSDV132, LSDV136) [12, 58]. In GTPV, LSDV009, LSDV013 and LSDV026 has SNP densities in excess of the genes’ median rate + 2*SD (along with LSDV003 and LSDV153). In SPPV, LSDV009, LSDV130 and LSDV132 also had comparably high SNPs/Kb. There were divergent haplotypes at the regions homologous to these five genes differentiating the three GTPV Clades, but none in SPPV. An example was LSDV026: an LSDV-specific gene potentially similar to VACV gene F11L, which affects host cell machinery manipulation and spread [58]. The GTPV LSDV026 region (at 17,410 − 18,341) had three haplotypes that extended into LSDV025, which encodes a putative Ser-Thr protein kinase based on homology with VACV VPK2 (Fig. 4). One haplotype was 565 bp, another was 563 bp and the third was 552 bp, differentiating Clades 2.1, 2.2 and 2.3, respectively (Fig. 4). LSDV026 had a high PN/PS ratio within GTPV (25/3) and an elevated ancestral DN/DS of 5/0, indicating pseudogenisation. In SPPV, the LSDV026 region had two main isoforms that had a large number of stop codons but a normal PN/PS ratio (4/3), raising questions about the role of this sequence (Figure S13).

Fig. 4.

Fig. 4

The GTPV PVG region homologous to LSDV026. (A) A sequenceTubeMap image shows Clade 2.3 at the top (MN072623, MN072624, MN072625 in grey, blue and orange), seven Clade 2.1/2.2 samples in the middle, and four samples from Clade 2.2. (B) Bandage graph of three LSDV026 haplotypes of lengths 522 bp, 563 bp and 565 bp with LSDV027 at the top and LSDV025 at the bottom. The blue box indicates the region shown in (A). A full 1,344 bp LSDV025 ORF was present in all genomes. (C) A mid-rooted phylogeny with three clades denoted 2.1 (grey, n = 6), 2.2 (purple, n = 5) and 2.3 (orange, n = 3). Nodes with bootstrap support > 90 are shown. (D) PCA based on SNP data. (E) A network where each node represents one sample. Grey lines indicate inter-community connections, and black lines indicate intra-community ones. The NC_004003.1:17,410 − 18,341 coordinates for LSDV026 were used

Possible signals of convergent evolution in GTPV and SPPV

Recent ancient and pre-modern SPPV genomes [51] further illuminate genetic variation during CaPV evolution and SPPV differentiation through comparison of the shared and unique alleles in the core genomes of LSDV, GTPV and SPPV representative samples. This showed that modern SPPV were the most divergent based on the number of unique alleles, following by GTPV (1,204 vs. 1,054 vs. 641). In pre-modern SPPV, the same pattern was present, though the number of core alleles shared by all species increased from 95 to 158. The latter metric increased again to 522 when LSDV, GTPV and ancient SPPV were examined. Ancient SPPV (MYRZ180) shared more alleles with LSDV (1,008) than GTPV (544), relative to LSDV and GTPV’s overlap (697). This also showed the relative number of shared SPPV-GTPV alleles increasing from ancient to modern (544 to 564), whereas the level for SPPV-LSDV fell (1,008 to 973), hinting at the potential for the higher genetic similarity of sheep and goat host immune systems to shape CaPV evolution convergently and differently from that of cattle.

In the coding SPPV data [51], there were 627 mutations where the ancient SPPV allele matched LSDV and GTPV, but differed with pre-modern and modern SPPV, and also 245 CDS mutations where the ancient and pre-modern SPPV allele matched LSDV and GTPV, but differed from modern SPPV (Table S5). 327 amino acid alleles where both pre-modern and modern SPPV were identical and differed from ancient SPPV (excluding pseudogenes) were more often identical to GTPV than to LSDV (16 vs. 7). Similarly, of the 110 amino acid alleles in modern SPPV differing from ancient and pre-modern SPPV (excluding pseudogenes), more were like GTPV than LSDV (9 vs. 1). This implicated 25 amino acid substitutions in possible convergent adaptations (Table S6). Notably, two genes had changes in both GTPV-related subsets: both had two amino acid alterations in the ancient to pre-modern transition, followed by another in the pre-modern to modern period. These genes were: LSDV032 encoding a poly(A) polymerase large subunit, and LSDV042 encoding a serine/threonine kinase. Collectively, these results underpinned the high differentiation and putative convergent evolution of SPPV and GTPV.

Discussion

This study explored the evolution and genomic diversity of GTPV and SPPV using phylogenetics, PVG-based and gene-specific approaches. GTPV and SPPV cause extensive disease outbreaks, along with LSDV WOAH [3]. Understanding better the genetic basis for these differences is pivotal. This study revealed clear contrasts in population structure, evolutionary history, and genome evolution between GTPV and SPPV. GTPV had a deeper and more stable population structure, comprising three genetically distinct clades with little evidence of recent admixture, whereas SPPV was a mosaic of more recent variation with weaker clade separation. For both viruses, genome-wide diversity patterns were consistent with an excess of rare mutations.

This work builds on earlier gene-centric CaPV pangenome analyses [56] by using PVGs that did not rely on gene annotation, capture non-coding and structural variation, and quantified PVG openness using multiple complementary approaches. Importantly, pangenome openness here reflected both biological diversity and uneven sampling, particularly the limited representation of samples from Africa, which was a constraint for both species. Across all genomes, the shared number of genomic bases was smaller for GTPV than for SPPV (142.4 vs 144.6 kb), reflecting GTPV’s stronger geographic structure. However, diversity at the 5′ and 3′ genome ends was higher relative to the core genome in SPPV than GTPV, consistent with the 5’ and 3’ end genes’ roles in encoding non-essential virulence and host immunomodulatory factors [12]. The open PVG inferred for SPPV indicated that broader geographic and temporal sampling will continue to reveal novel variation, whereas the closed GTPV PVG suggested limited returns from additional sampling in Eurasia. Lastly, the representative PVGs constructed here can be a practical resource for improved short-read mapping and variant detection in future CaPV work [50].

PVG-based analyses were particularly informative at the ITRs, revealing substantial structural and haplotypic variation centred on regions homologous to LSDV003 and LSDV154. These genes are annotated as encode ER-localised apoptosis regulators that interfere with host mitochondrial function and apoptosis [11]. LSDV003 is predicted to have a signal peptide, a transmembrane domain, and insert into the endoplasmic reticulum [59]. The protein it encodes is similar to MYXV M004, an essential virulence factor Enow [63]. The LSDV003 and LSDV154 truncations in vaccine-related SPPV Clade 3.1 were consistent with attenuated virulence following serial passage in non-natural hosts. Variation at LSDV154 was previously associated with SPPV outbreaks in south-western Europe [5]. The longer putative LSDV003 ORFs in six SPPV samples across Clades 3.2 and 3.3 may be associated with expression regulation of this protein via protease cleavage. This potential structural ITR plasticity could mean that these extended ORFs encode mRNAs with different expression levels or post-translational functions. In addition, the retention of the non-CDS regions suggests a continued functional relevance, perhaps to facilitate recombination for those at the ITRs.

Beyond the ITRs, patterns of gene-specific diversity highlighted candidate loci associated with host specificity. 18 SPPV genes had either zero or one SNP, implying strong functional constraint. In contrast, GTPV only had one gene with high conservation, suggesting selective processes maybe operating on each clade differently, perhaps linked to their higher geographic population structure. There were signals consistent with adaptive evolution at three genes, including LSDV068 encoding a polyA polymerase small subunit. The second was: LSDV023, which had sequence-based and structural homology on viro3D [60] with MYXV m018L gene that encodes non-essential cytoplasmic protein [61]. The third was LSDV129, which had similarity to MYXV M130R, whose product has substantive immunomodulation and virulence effects in rabbit hosts [62]. In addition, the three major GTPV lineages were consistently differentiated by divergent haplotypes at regions corresponding to LSDV002, LSDV009, LSDV013, LSDV026 and LSDV153. In addition, this study found possible evidence of convergent evolution in SPPV and GTPV using published ancient and pre-modern SPPV genomes [51]. This implied that the higher similarity of the goat and sheep immune systems relative to that of cattle since the Bronze Age may elicit shared changes at specific proteins, their genes and even at the nucleotide level. Here, 16 and nine such SPPV amino acid substitutions became GTPV-like in the Bronze Age to pre-modern and pre-modern to modern eras, respectively, whereas comparable numbers for LSDV were seven and one, respectively. One of the proteins implicated (at LSDV032) was a polyA polymerase component, like LSDV068’s gene product above.

In summary, GTPV evolution is structured by deep, clade-specific divergence, whereas SPPV has a more diffuse ongoing diversification from a more recent origin. Several broader questions emerge from this study. First, the retention of long non-coding regions at loci that are genes in LSDV only (LSDV002, LSDV004, LSDV009, LSDV013, LSDV026, LSDV132, LSDV153, LSDV155) suggested that these regions may be subject to functional or structural constraints beyond protein coding. ITRs are constrained by replication mechanisms, whereas the preservation of internal non-CDS loci across divergent lineages argues for targeted functional testing. Second, many of the complex rearrangements described here, especially at the ITRs, are likely under-resolved in short-read assemblies. Long-read sequencing and haplotype-resolved assemblies will be essential for improving CaPV genome annotation, resolving ORF structure, and linking genomic variation to viral phenotype. Third, there were vaccine-linked specimens with unique ITR mutations: these recombination-prone regions may be as hotspots for new adaptive variation, underscoring the need for genomic surveillance of vaccine-linked strains using PVGs [50]. Fourth, GTPV isolates infect both sheep and goats [52] and yet SPPV may be the most basal among the CaPV [51], highlighting continued evolutionary questions that may impact perspectives on host tropism.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1 (589.3KB, xlsx)
Supplementary Material 2 (4.6MB, docx)

Acknowledgements

This study acknowledges the Pirbright Institute’s Bioinformatics STP resources, Tiffany Tobi, Mai Prihartini, Jonas Albarnaz, Miguel Hernandez and Caroline Wright.

Author contributions

T.D. completed analyses & wrote the paper.

Funding

This study had funding support through from UK Research and Innovation (UKRI) Biotechnology and Biological Sciences Research Council (BBSRC) grants BBS/E/PI/230002A, BBS/E/PI/230002B, BBS/E/PI/230002 C and BBS/E/PI/23NB0003.

Data availability

All supporting code and protocols have been provided within the article, through supplementary data files is available at https://github.com/downingtim/GTPV-SPPV-PVG.

Declarations

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

  • 1.Casal J, Allepuz A, Miteva A, Pite L, Tabakovsky B, Terzievski D, Alexandrov T, Beltrán-Alcrudo D (2018) Economic cost of lumpy skin disease outbreaks in three Balkan countries: Albania, Bulgaria and the Former Yugoslav Republic of Macedonia (2016–2017). Transbound Emerg Dis 65(6):1680–1688. 10.1111/tbed.12926 [DOI] [PubMed] [Google Scholar]
  • 2.Vinitchaikul P, Punyapornwithaya V, Seesupa S, Phuykhamsingha S, Arjkumpa O, Sansamur C, Jarassaeng C (2023) The first study on the impact of lumpy skin disease outbreaks on monthly milk production on dairy farms in Khon Kaen, Thailand. Vet World 16(4):687–692. 10.14202/vetworld.2023.687-692 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.WOAH. Version adopted by the World Assembly of Delegates of the OIE in 2010, OIE,Paris.: Terrestrial Manual of Lumpy Skin Disease (2010) [Available from: https://www.woah.org/app/uploads/2021/03/a-reso-2016-public.pdf
  • 4.Binois-Roman A, L’Hôte L, Ferguson R, Daly KG Evidence of sheeppox in 17th-19th century France: a multi-disciplinary investigation of a sheep mass mortality assemblage. BioRxiv 2025 doi: 10.1101/2025.11.24.690131 [DOI] [PMC free article] [PubMed]
  • 5.Breman FC, Haegeman A, Philips W, Krešić N, Hoffman S, De Keersmaecker SCJ, Roosens NHC, Agüero M, Villalba R, Miteva A, Ivanova E, Tasioudi KE, Chaintoutis SC, Kirtzalidou A, De Regge N (2024) Sheeppox virus genome sequences from the European outbreaks in Spain, Bulgaria, and Greece in 2022–2024. Arch Virol 169(11):234. 10.1007/s00705-024-06165-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Sprygin A, Krotova A, Jun M, Byadovskaya O, Kirpichenko V, Chen J, Sainnokhoi T, Chvala I (2025) Whole Genome Sequencing of Lumpy Skin Disease Virus from 2021–2023 in Eastern Eurasia Reveals No More Recombination Signals in the Circulating Pool of Strains. Viruses 17(4):468. 10.3390/v17040468 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Bhanuprakash V, Moorthy AR, Krishnappa G, Singh RK (2006) Sheep pox and goat pox: An update. Vet Res Commun 30(8):825–838. 10.1007/s11259-006-3440-2 [DOI] [Google Scholar]
  • 8.Ramakrishnan MA, Santhamani R, Pandey AB (2017) Capripox outbreak in a mixed flock of sheep and goats in India. Transbound Emerg Dis 64(1):27–30. 10.1111/tbed.12604 [DOI] [PubMed] [Google Scholar]
  • 9.Hamdi J, Bamouh Z, Jazouli M, Boumart Z, Tadlaoui KO, Fihri OF, El Harrak M (2020) Experimental evaluation of the cross-protection between Sheeppox and bovine Lumpy skin vaccines. Sci Rep 10(1):8888. 10.1038/s41598-020-65856-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Haegeman A, Philips W, Mostin L, De Leeuw I, Van Campe W, Saegerman C, De Clercq K, De Regge N (2025) A goatpox but not sheeppox heterologous live attenuated vaccines provide complete protection against lumpy skin disease in cattle under experimental conditions. Sci Rep 15(1):26078. 10.1038/s41598-025-11440-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Tulman ER, Afonso CL, Lu Z, Zsak L, Kutish GF, Rock DL (2001) Genome of lumpy skin disease virus. J Virol 75(15):7122–7130. 10.1128/JVI.75.15.7122-7130.2001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Tulman ER, Afonso CL, Lu Z, Zsak L, Sur JH, Sandybaev NT, Kerembekova UZ, Zaitsev VL, Kutish GF, Rock DL (2002) The genomes of sheeppox and goatpox viruses. J Virol 76(12):6054–6061. 10.1128/jvi.76.12.6054-6061.2002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Kara PD, Afonso CL, Wallace DB, Kutish GF, Abolnik C, Lu Z, Vreede FT, Taljaard LC, Zsak A, Viljoen GJ, Rock DL (2003) Comparative sequence analysis of the South African vaccine strain and two virulent field isolates of Lumpy skin disease virus. Arch Virol 148(7):1335–1356. 10.1007/s00705-003-0102-0 [DOI] [PubMed] [Google Scholar]
  • 14.Zhang M, Shi Y, Lu X, Zhang Q, Zhao Y, Li S, Wen Z, Ge J, Wang X, Li J, Bu Z, Yin X (2025) Lumpy skin disease virus 001/156 protein is a virulence factor that suppresses interferon production through impairing IRF3 dimerization. PLoS Pathog 21(7):e1013362. 10.1371/journal.ppat.1013362 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Shenouda MM, Noyce RS, Lee SZ, Wang JL, Lin YC, Favis NA, Desaulniers MA, Evans DH (2022) The mismatched nucleotides encoded in vaccinia virus flip-and-flop hairpin telomeres serve an essential role in virion maturation. PLoS Pathog 18(3):e1010392. 10.1371/journal.ppat.1010392 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Haga IR, Shih BB, Tore G, Polo N, Ribeca P, Gombo-Ochir D, Shura G, Tserenchimed T, Enkhbold B, Purevtseren D, Ulziibat G, Damdinjav B, Yimer L, Bari FD, Gizaw D, Adedeji AJ, Atai RB, Adole JA, Dogonyaro BB, Kumarawadu PL, Batten C, Corla A, Freimanis GL, Tennakoon C, Law A, Lycett S, Downing T, Beard PM (2024) Sequencing and Analysis of Lumpy Skin Disease Virus Whole Genomes Reveals a New Viral Subgroup in West and Central Africa. Viruses 16(4):557. 10.3390/v16040557 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Downing T (2026) Approaches to Studying Virus Pangenome Variation Graphs. Genomics Proteom Bioinf. 10.1093/gpbjnl/qzag003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Eizenga JM, Novak AM, Sibbesen JA, Heumos S, Ghaffaari A, Hickey G, Chang X, Seaman JD, Rounthwaite R, Ebler J, Rautiainen M, Garg S, Paten B, Marschall T, Sirén J, Garrison E (2020) Pangenome Graphs. Annu Rev Genomics Hum Genet 21:139–162. 10.1146/annurev-genom-120219-080406 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Garrison E, Guarracino A (2023) Unbiased pangenome graphs. Bioinformatics 39(1):btac743. 10.1093/bioinformatics/btac743 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Wright C, Polo N, Freimanis G, Azam S, Sharma GT, Charleston B, Batten C, Downing T, Albarnaz J (2026) A telomere-to-telomere hybrid lumpy skin disease virus genome assembly. BioRxiv 10.64898/2026.03.24.714000 [DOI] [PMC free article] [PubMed]
  • 21.Sayers EW, Cavanaugh M, Frisse L, Pruitt KD, Schneider VA, Underwood BA, Yankie L, Karsch-Mizrachi I (2025) GenBank 2025 update. Nucleic Acids Res 53(D1):D56–D61. 10.1093/nar/gkae1114 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Katoh K, Standley DM (2013) MAFFT multiple sequence alignment software version 7: improvements in performance and usability. Mol Biol Evol 30(4):772–780 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Kozlov AM, Darriba D, Flouri T, Morel B, Stamatakis A (2019) RAxML-NG: a fast, scalable and user-friendly tool for maximum likelihood phylogenetic inference. Bioinformatics 35(21):4453–4455 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Darriba D, Posada D, Kozlov AM, Stamatakis A, Morel B (2020) ModelTest-NG: A new and scalable tool for the selection of DNA and protein evolutionary models. Mol Biol Evol 37(1):291–294 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.R Core Team (2024) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria [Google Scholar]
  • 26.Paradis E, Schliep K (2019) ape 5.0: an environment for modern phylogenetics and evolutionary analyses in R. Bioinformatics 35(3):526–528. 10.1093/bioinformatics/bty633 [DOI] [PubMed] [Google Scholar]
  • 27.Schliep KP (2011) phangorn: phylogenetic analysis in R. Bioinformatics 27(4):592–593 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Wang LG, Lam TTY, Xu S, Dai Z, Zhou L, Feng T, Guo P, Dunn CW, Jones BR, Bradley T, Zhu H, Guan Y, Jiang Y (2020) Yu. treeio: an R package for phylogenetic tree input and output with richly annotated and associated data. Mol Biol Evol 37(2):599–603. 10.1093/molbev/msz240 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Yu G, Smith DK, Zhu H, Guan Y, Lam TT (2017) ggtree: an R package for visualization and annotation of phylogenetic trees with their covariates and other associated data. Methods Ecol Evol 8:28–36 [Google Scholar]
  • 30.Wickham H (2016) ggplot2: Elegant Graphics for Data Analysis. Springer-, New York. https://ggplot2.tidyverse.org [Google Scholar]
  • 31.Tennakoon C, Freville T, Downing T (2026) Panalyze: automated virus pangenome variation graph construction, analysis and annotation. Bioinform Adv 6(1):vbag071. 10.1093/bioadv/vbag071 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Garrison E, Guarracino A, Heumos S, Villani F, Bao Z, Tattini L, Hagmann J, Vorbrugg S, Marco-Sola S, Kubica C, Ashbrook DG, Thorell K, Rusholme-Pilcher RL, Liti G, Rudbeck E, Golicz AA, Nahnsen S, Yang Z, Mwaniki MN, Nobrega FL, Wu Y, Chen H, de Ligt J, Sudmant PH, Huang S, Weigel D, Soranzo N, Colonna V, Williams RW, Prins P (2024) Building pangenome graphs. Nat Methods 21(11):2008–2012. 10.1038/s41592-024-02430-3 [DOI] [PubMed] [Google Scholar]
  • 33.Guarracino A, Mwaniki N, Marco-Sola S, Garrison E (2021) wfmash: whole-chromosome pairwise alignment using the hierarchical wavefront algorithm. https://github.com/ekg/wfmash
  • 34.Ewels P, Magnusson M, Lundin S, Käller M (2016) MultiQC: summarize analysis results for multiple tools and samples in a single report. Bioinformatics 32:3047–3048 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Guarracino A, Heumos S, Nahnsen S, Prins P, Garrison E (2022) ODGI: understanding pangenome graphs. Bioinformatics 38(13):3319–3326. 10.1093/bioinformatics/btac308 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Garrison E, Sirén J, Novak AM, Hickey G, Eizenga JM, Dawson ET, Jones W, Garg S, Markello C, Lin MF, Paten B, Durbin R (2018) Variation graph toolkit improves read mapping by representing genetic variation in the reference. Nat Biotechnol 36(9):875–879. 10.1038/nbt.4227 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Li H, GitHub (2024) https://github.com/lh3/gfatools
  • 38.Formenti G (2025) https://github.com/vgl-hub/gfastats
  • 39.Kubica N gfautil https://crates.io/crates/gfautil
  • 40.Seemann T (2014) Prokka: rapid prokaryotic genome annotation. Bioinformatics 30(14):2068–2069 [DOI] [PubMed] [Google Scholar]
  • 41.Danecek P et al (2021) SAMtools v1.19. Bioinformatics 37:123–125 [Google Scholar]
  • 42.Seppey M, Manni M, Zdobnov EM (2019) BUSCO: Assessing Genome Assembly and Annotation Completeness. Methods Mol Biol 1962:227–245. 10.1007/978-1-4939-9173-0_14 [DOI] [PubMed] [Google Scholar]
  • 43.Pertea G, Pertea MGFF, Utilities GffRead and GffCompare. F1000Res. 2020 9:ISCB Comm J-304. 10.12688/f1000research.23297.2 [DOI] [PMC free article] [PubMed]
  • 44.Wick RR, Schultz MB, Zobel J, Holt KE (2015) Bandage: interactive visualization of de novo genome assemblies. Bioinformatics 31(20):3350–3352. 10.1093/bioinformatics/btv383 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Parmigiani L, Garrison E, Stoye J, Marschall T, Doerr D Panacus: fast and exact pangenome growth and core size estimation. Bioinf 2024a 40(12):btae720. 10.1093/bioinformatics/btae720 [DOI] [PMC free article] [PubMed]
  • 46.Parmigiani L, Wittler R, Stoye J (2024b) Revisiting pangenome openness with k-mers. Peer Commun J 4:e47. 10.24072/pcjournal.415 [DOI] [Google Scholar]
  • 47.Decano AG, Downing T (2019) An Escherichia coli ST131 pangenome atlas reveals population structure and evolution across 4,071 isolates. Sci Rep 9(1):17394. 10.1038/s41598-019-54004-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Csardi G, Nepusz T (2006) 1695 The Igraph Software Package for Complex Network Research. InterJournal, Complex Systems, 1–9
  • 49.Stoletzki N, Eyre-Walker A (2011) Estimation of the neutrality index. Mol Biol Evol 28(1):63–70. 10.1093/molbev/msq249 [DOI] [PubMed] [Google Scholar]
  • 50.Downing T, et al (2026) Using pangenome variation graphs to improve mutation detection in a large DNA virus. Microbial Genomics 10.1099/mgen.0.001698 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.L’Hôte L, Sacristán L, Ferguson R, Siekmann A, Rogers L, Richter B, Weissenböck H, Lorke J, Artemis L, Lévêque É, Hark R, Engel P, Webber MTJ, Bennett M, Rose-Beers K, Nichols E, Munoz Alegre M, Ramsøe M, Fiddyment S, Viñas-Caron LC, Papin DV, Light-Maka IC, Key FM, Albarnaz J, Downing T, Pekar J, Lemey P, Bradley D, Vnouček J, Makarewicz CA, Story J, Collins M, Teasdale MD, Binois-Roman A, Calvignac-Spencer S, Daly KG (2026) 3,500 years of sheeppox virus evolution inferred from archaeological and codicological genomes. BioRxiv 10.64898/2026.02.23.707469 [DOI]
  • 52.Biswas S, Noyce RS, Babiuk LA, Lung O, Bulach DM, Bowden TR, Boyle DB, Babiuk S, Evans DH (2020) Extended sequencing of vaccine and wild-type capripoxvirus isolates provides insights into genes modulating virulence and host range. Transbound Emerg Dis 67(1):80–97. 10.1111/tbed.13322 [DOI] [PubMed] [Google Scholar]
  • 53.Watterson GA (1975) On the number of segregating sites in genetical models without recombination. Theor Popul Biol 7(2):256–276. 10.1016/0040-5809(75)90020-9 [DOI] [PubMed] [Google Scholar]
  • 54.Tajima F (1989) Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123(3):585–595. 10.1093/genetics/123.3.585 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Yang Y-L, Jia H-J, Sun M-Y, Guo A-M, Xia T, Shu H-B, Cao L-B (2025) Lumpy skin disease virus LSDV001 protein positively regulates inflammatory response by promoting assembly of the TAK1-Table 2/3 complex. mBio 16(10):e0167725. 10.1128/mbio.01677-25 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Xie S, Cui L, Liao Z, Zhu J, Ren S, Niu K, Li H, Jiang F, Wu J, Wang J, Wu J, Song B, Wu W, Peng C (2024) Genomic analysis of lumpy skin disease virus asian variants and evaluation of its cellular tropism. NPJ Vaccines 9(1):65. 10.1038/s41541-024-00846-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Krotova A, Shalina K, Mazloum A, Kwon D, Van Schalkwyk A, Byadovskaya O, Sprygin A (2022) Genetic characterization of sheep pox virus strains from outbreaks in Central Russia in 2018–2019. Transbound Emerg Dis 69(6):e3430–e3435. 10.1111/tbed.14727 [DOI] [PubMed] [Google Scholar]
  • 58.Wang Y, Zhang S, Li H, Feng Y, Zhang Y (2025) Orf132: a critical gene for LSDV replication and its role in Er stress-related apoptosis. Virol J 22(1):172. 10.1186/s12985-025-02813-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Liaci AM, Förster F (2021) Take Me Home, Protein Roads: Structural Insights into Signal Peptide Interactions during ER Translocation. Int J Mol Sci 22(21):11871. 10.3390/ijms222111871 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Litvin U, Lytras S, Jack A, Robertson DL, Hughes J, Grove J (2025) Viro3D: a comprehensive database of virus protein structure predictions. Mol Syst Biol 21(11):1599–1617. 10.1038/s44320-025-00147-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61.Higley S, Way M (1997) Characterization of the vaccinia virus F8L protein. J Gen Virol 78(Pt 10):2633–2637. 10.1099/0022-1317-78-10-2633 [DOI] [PubMed] [Google Scholar]
  • 62.Barrett JW, Werden SJ, Wang F, McKillop WM, Jimenez J, Villeneuve D, McFadden G, Dekaban GA (2009) Myxoma virus M130R is a novel virulence factor required for lethal myxomatosis in rabbits. Virus Res 144(1–2):258–265. 10.1016/j.virusres.2009.05.009 [DOI] [PubMed] [Google Scholar]
  • 63.Enow JA, Lopes AM, Abrantes J, Esteves PJ, Rahman MM (2025) The Role of Myxoma Virus Immune Modulators and Host Range Factors in Pathogenesis and Species Leaping. Viruses 17(8):1145. 10.3390/v17081145 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64.Guarracino A, Buonaiuto S, de Lima LG, Potapova T, Rhie A, Koren S, Rubinstein B, Fischer C, Human Pangenome Reference Consortium, Gerton JL, Phillippy AM, Colonna V, Garrison E (2023) Recombination between heterologous human acrocentric chromosomes. Nature 617(7960):335–343. 10.1038/s41586-023-05976-y [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1 (589.3KB, xlsx)
Supplementary Material 2 (4.6MB, docx)

Data Availability Statement

All supporting code and protocols have been provided within the article, through supplementary data files is available at https://github.com/downingtim/GTPV-SPPV-PVG.


Articles from Archives of Virology are provided here courtesy of Springer

RESOURCES