Skip to main content
NCBI home page
Genome Biology and Evolution logoLink to Genome Biology and Evolution
. 2024 Mar 19;16(3):evae048. doi: 10.1093/gbe/evae048

Giants among Cnidaria: Large Nuclear Genomes and Rearranged Mitochondrial Genomes in Siphonophores

Namrata Ahuja 1,✉,#, Xuwen Cao 2,#, Darrin T Schultz 3,#, Natasha Picciani 4, Arianna Lord 5, Shengyuan Shao 6, Kejue Jia 7, David R Burdick 8, Steven H D Haddock 9, Yuanning Li 10, Casey W Dunn 11
Editor: Dennis Lavrov
PMCID: PMC10980510  PMID: 38502059

Abstract

Siphonophores (Cnidaria: Hydrozoa) are abundant predators found throughout the ocean and are important constituents of the global zooplankton community. They range in length from a few centimeters to tens of meters. They are gelatinous, fragile, and difficult to collect, so many aspects of the biology of these roughly 200 species remain poorly understood. To survey siphonophore genome diversity, we performed Illumina sequencing of 32 species sampled broadly across the phylogeny. Sequencing depth was sufficient to estimate nuclear genome size from k-mer spectra in six specimens, ranging from 0.7 to 2.3 Gb, with heterozygosity estimates between 0.69% and 2.32%. Incremental k-mer counting indicates k-mer peaks can be absent with nearly 20× read coverage, suggesting minimum genome sizes range from 1.4 to 5.6 Gb in the 25 samples without peaks in the k-mer spectra. This work confirms most siphonophore nuclear genomes are large relative to the genomes of other cnidarians, but also identifies several with reduced size that are tractable targets for future siphonophore nuclear genome assembly projects. We also assembled complete mitochondrial genomes for 33 specimens from these new data, indicating a conserved gene order shared among nonsiphonophore hydrozoans, Cystonectae, and some Physonectae, revealing the ancestral mitochondrial gene order of siphonophores. Our results also suggest extensive rearrangement of mitochondrial genomes within other Physonectae and in Calycophorae. Though siphonophores comprise a small fraction of cnidarian species, this survey greatly expands our understanding of cnidarian genome diversity. This study further illustrates both the importance of deep phylogenetic sampling and the utility of k-mer–based genome skimming in understanding the genomic diversity of a clade.

Keywords: k-mer spectra, genome skimming, genome size, mitochondrial genomes

Significance.

Descriptions of basic genome features, such as nuclear genome size and mitochondrial genome sequences, remain sparse across many clades in the tree of life, leading to over generalizations from very small sample sizes and often limiting selection of optimal species for genome assembly efforts. Here, we use whole-genome shotgun (WGS) skimming to assess a variety of genome features across 35 siphonophores (Cnidaria). This deep dive within a single clade identifies optimal candidates for future genomic work. It further reveals siphonophores have a greater range in nuclear genome size and diversity of mitochondrial genome orders than had been described across all Cnidaria.

Introduction

Siphonophores (Fig. 1) are among the longest (Robison 1995) and most abundant animals in the ocean. They occupy a critical position in the food web of the open ocean (Hetherington et al. 2022) and possess many unique biological features (Mapstone 2014; Munro et al. 2018). Like their coral relatives, siphonophores are colonial animals with many highly integrated, genetically identical bodies that all arise from a single embryo by asexual reproduction. Unlike corals, they are free swimming, and the bodies (zooids) within the colonies are specialized for different tasks, such as feeding, locomotion, and sexual production of new colonies. Studies on siphonophores have been limited, though, because they are fragile and difficult to collect. Improved genomic resources would greatly facilitate work on all aspects of their study and make it possible to learn much more from each of the specimens that are collected.

Fig. 1.

Fig. 1.

Open in a new tab

Photographs of some siphonophore species included in this study by CWD and SHD. Lengths are approximate. a) Physalia (float is about 20 mm across). b) Frillagalma vityazi (about 8 cm long). c) Nanomia bijuga (about 5 cm long). d) Apolemia rubriversa (specimen image about 7 cm across). e) Gymnopraia lapislazula (nectophore is about 2 cm across).

Molecular data available for siphonophores include a few genes sequenced across a large number of species (Dunn et al. 2005), transcriptomes for developmental and phylogenetic work (Siebert et al. 2011; Munro et al. 2022, 2018), and mitochondrial genomes for three species (Kayal et al. 2015). Chromosome-scale genomes have been assembled for a broad diversity of cnidarians (sea anemones, corals, and jellyfish) in recent years (Fig. 2), but most of these are for hexacorals and relatively few are from Hydrozoa, the clade that includes Siphonophora (Zapata et al. 2015). While no nuclear siphonophore genomes have been sequenced, two genome size estimates based on flow cytometry are available: Agalma elegans has been estimated to be 3.482 Gb and Physalia physalis at 3.247 Gb (Adachi et al. 2017). These are by far the largest genomes known to exist for cnidarians, similar in size to human genomes. However, with a sample size of 2 out of nearly 200 known species, sampling is insufficient to confidently describe nuclear genome size diversity within Siphonophorae.

Fig. 2.

Fig. 2.

Open in a new tab

Number of nuclear genomes sequenced across Cnidaria, including both scaffold and chromosome-level assemblies. Siphonophores are within Hydroidolina, a subclade within Hydrozoa.

In parallel to nuclear genomics, considerable progress has been made in understanding cnidarian mitochondrial genomics, but this sampling is not uniformly distributed. Currently, mitochondrial genome assemblies are available for over 300 anthozoans and 78 medusozoans; however, only 35 medusozoan mitochondrial genomes are complete (Ling et al. 2023). Almost all sequenced anthozoan mitochondrial genomes are circular. Though medusozoan mitochondrial genomes are thought to be linear (Bridge et al. 1992; Kayal et al. 2015), several medusozoan mitochondrial genomes in public archives are annotated as being circular. However, there has not been enough data on siphonophore mitochondrial genomes to distinguish how they may compare to other Medusozoa, especially given their unresolved position within the clade and the subclade Hydroidolina. Currently, there are three nearly complete mitochondrial siphonophore genomes, including P. physalis, Nanomia bijuga, and Rhizophysa eysenhardtii (Kayal et al. 2015). These have identical gene orders to each other, consistent with few evolutionary changes in mitochondrial genome structure in Siphonophorae. However, the partial nature of these genome sequences and limited taxon sampling is too small to robustly test this.

Genome skimming by shotgun Illumina sequencing has emerged as a powerful tool for gaining insight about multiple genomic features (Heath-Heckman and Nishiguchi 2021; Hogan et al. 2022). Illumina reads are too short to assemble full nuclear genomes, but are sufficient to assemble mitochondrial genomes and to run a variety of analyses on nuclear genome properties based on k-mer spectra. With sufficient depth, Illumina sequencing provides estimates of nuclear genome size, heterozygosity, and repeat fraction. Here, we apply genome skimming to a broad diversity of 35 specimens from 32 siphonophore species across the siphonophore phylogeny, representing every major subclade. Our objectives are primarily biological: we seek to expand sampling to understand basic features of genome diversity across siphonophores and understand them relative to other cnidarians. Our objectives are also technical: we hope to identify the species that would be most amenable to chromosome-scale genome sequencing projects and evaluate the resources and approaches that would be needed for such studies at this time.

Results

Specimen Sampling and Sequencing

A total of 2,978 Gb of sequence data were collected across 35 specimens from 32 species (Table 1, Supplementary table S1). We took an iterative approach to sequencing, with a shallow first pass of most samples and then deeper sequencing in a progressively narrower subset of samples. The amount of data per sample therefore ranged widely, from 29 to 311 Gb (Table 1). We prioritized samples according to two criteria. First, we collected more data for a specimen if we observed peaks in the k-mer spectrum but needed more data for the GenomeScope2 genome size estimate models to fit well. Second, we prioritized several abundant species that are likely to be the focus of future genomic work, including Physalia and A. elegans, even though it was clear from preliminary k-mer skimming results that their genomes are quite large.

Table 1.

Overview of all samples

Sample Result category Haploid genome size estimate (Gb) Haploid genome size minimum (Gb) Read pair number (millions) Total nucleotides (Gb) Heterozygosity (%) Repeat (%)
Abylopsis tetragona No peak … 1.56 106.342 31.30 … …
Agalma clausi No peak … 1.53 103.939 30.67 … …
Agalma elegans Atlantic No peak … 1.44 96.987 28.77 … …
Agalma elegans Pacific Good fit 2.32 … 1034.742 308.00 2.80 57.47
Agalma okeni Low peak … … 255.282 72.97 … …
Apolemia rubriversa No peak … 3.35 249.821 67.03 … …
Apolemia sp3 No peak … 2.01 145.572 40.16 … …
Bargmannia elongata No peak … 4.33 299.470 86.61 … …
Bargmannia lata No peak … 2.12 150.626 42.49 … …
Chelophyes appendiculata Good fit 1.20 … 353.189 105.00 5.16 56.91
Chuniphyes multidentata No peak … 3.11 221.280 62.27 … …
Cordagalma ordinatum Good fit 0.69 … 150.794 43.83 5.31 50.76
Craseoa lathetica No peak … 2.45 170.722 48.99 … …
Diphyes dispar No peak … 2.90 198.254 57.98 … …
Erenna sirena No peak … 4.60 331.332 92.00 … …
Forskalia No peak … 3.71 267.534 74.23 … …
Frillagalma vityazi No peak … 3.86 284.049 77.16 … …
Gymnopraia lapislazula Poor fit 2.33 … 878.095 256.00 6.11 83.98
Halistemma rubrum Atlantic No peak … 2.50 170.905 50.04 … …
Halistemma rubrum Mediterranean No peak … 1.54 103.786 30.77 … …
Hippopodius hippopus No peak … 3.65 266.036 73.09 … …
Lilyopsis fluoracantha No peak … 4.02 282.743 80.43 … …
Nanomia bijuga North Carolina Good fit 0.71 … 312.832 92.50 4.04 44.55
Nanomia California Good fit 1.45 … 290.026 86.47 0.67 63.30
Physalia Guam Good fit 1.65 … 663.882 185.00 3.01 64.85
Physonect No peak … 3.30 229.185 65.97 … …
Praya No peak … 3.34 235.932 66.86 … …
Praya dubia No peak … 5.55 383.256 111.00 … …
Resomia ornicephala T845-D3 Low peak … … 167.278 46.81 … …
Resomia ornicephala T898-D2 Poor fit 1.01 … 566.160 157.00 3.59 60.48
Rhizophysa eysenhardtii No peak … 3.44 238.482 68.85 … …
Rhizophysa filiformis No peak … 2.36 159.073 47.29 … …
Rosacea flaccida No peak … 1.49 101.192 29.81 … …
Stephalia dilata No peak … 3.14 219.048 62.75 … …
Stephanomia amphytridis No peak … 2.40 169.178 48.09 … …
Open in a new tab

Read pair number is the number of paired reads after trimming. Total length is the sum of the number of nucleotides in paired reads after trimming. Genome size estimates, heterozygosity, and repeat fraction are shown for the poor fit and good fit samples (i.e. samples that had peaks in GenomeScope). Given that the incremental analyses with as much as 16× coverage did not have a peak (supplementary fig. S1, Supplementary Material online), the genome size minimum for those samples without a peak is calculated conservatively as 1/20 the total of read length.

Trimmomatic (Bolger et al. 2014) removed from 1.3% to 12.4% of nucleotides per data set resulting in very few nucleotides being trimmed for the majority of specimens sampled (supplementary table S2, fraction_trimmed, and fig. S1, Supplementary Material online). Then, Kraken (Wood and Salzberg 2014) identified from 0.1% to 3.8% of reads as bacterial (supplementary table S2 and fig. S1, Supplementary Material online) and from 1.0% to 36.33% of reads as human (supplementary table S2 and fig. S1, Supplementary Material online). Because Kraken has been shown to have very high false-positive rates (Garrido-Sanz et al. 2022), we further assessed the potential for human contamination by mapping reads to the human genome. When we mapped a million reads from each sample to the human genome, most samples had zero paired mapped reads (supplementary table S2, Supplementary Material online, human_mapped_count). Of the samples that were nonzero, the maximum number of reads that mapped to the human genome was 1,042 (Stephanomia amphytridis), well below what could impact results presented here. These results are consistent with the high false classification rate by Kraken. We assessed the effects of putative prokaryotic read contamination on the siphonophore genome size estimates by also estimating genome sizes with the subset of reads that were labeled “unclassified” by Kraken (i.e. not flagged as possible contaminants; supplementary fig. S1, Supplementary Material online). This exclusion of classified, putatively prokaryotic reads had very little impact on genome size estimates, indicating that our results are robust to the amount of environmental or metagenomic bacterial contamination present in these siphonophore samples. Given the robustness of the genome size estimates, regardless of filtering from Kraken, our remaining analyses focused on the data sets with reads that were only quality trimmed with Trimmomatic.

k-mer Analysis

The k-mer results for 35 sequenced samples fall into 4 result categories (Table 1), those with the poorest sampling relative to genome size to the best sampling. First is the “no peak” category. These 25 samples had no peak in the k-mer spectra (supplementary table S2, Supplementary Material online, peak_location). Second is the “low peak” category. These two samples, Agalma okeni and Resomia ornicephala T845-D3, had peaks at very low counts; they were so low that they were out of range in GenomeScope plots, and these samples were not considered further in k-mer analyses. Third is the “poor fit” category. These two samples, Gymnopraia lapislazula and R. ornicephala T898-D2, had peaks that were in the range of GenomeScope plots (Fig. 3), but did not have consistently good model fit across all k-mer analyses (Fig. 3; supplementary fig. S2, Supplementary Material online). Fourth is the “good fit” category. These six samples (Fig. 3) resulted in well-defined k-mer spectra with good model fit and confident genome size estimates. GenomeScope2 also provides estimates of heterozygosity and repeat percentage for these specimens (Table 1).

Fig. 3.

Fig. 3.

Open in a new tab

Plots of k-mer spectra for samples with peaks. Model fit is assessed by whether the full model curve (outer curved line) has peaks congruent with the underlying k-mer spectrum. The reported genome sizes for the specimens with poor model fit, indicated by asterisks, are from the Jellyfish counting.

Incremental k-mer Counting

We wrote a new k-mer counter, sharkmer, that counts k-mers incrementally. Rather than providing a single snapshot of the k-mer spectrum when all data are considered, it provides a progressive understanding of a k-mer spectrum as sequencing depth is increased. We ran sharkmer on the six good fit and two poor fit samples (i.e. all samples with peaks in GenomeScope) to assess k-mer spectra with subsets of the data from 1% to 100% in 1% increments. To maintain a tractable memory footprint, we considered only the first billion reads if the data sets were larger than this, capping the largest data sets at 151 Gb (supplementary table S3, Supplementary Material online) since each read was 151 bp.

We first validated sharkmer performance by comparing k-mer counts from the final sharkmer sample to two widely used k-mer counters, Jellyfish and KMC (Deorowicz et al. 2013, 2015; Kokot et al. 2017). The k-mer counts were identical across the three tools (supplementary fig. S2, Supplementary Material online).

We used the incremental k-mer counting results to assess the stability of genome size estimation methods. Ideally, we would see a method converge on a genome size as data are added. However, if a method does not converge, results are sensitive to sampling effort. The parametric method implemented in GenomeScope2 does converge well for the six samples with good fit, but not the two with poor fit. This can be seen in plots of genome size and fit versus the percent of reads sampled (Fig. 4). It can also be seen in animations of GenomeScope plots as sampling is increased (see Github repository). At low sampling, both genome sizes and model fits are low and unstable. This is expected given that much of this range is before there is sufficient coverage to produce a peak. Peaks arise at coverages ranging from 8.33× to 15.65× coverage (supplementary table S4, Supplementary Material online). The point at which the maximum model fit reaches 95% corresponds well with the model arriving at the final plateau it converges on, though for most species, estimated genome size still tends to decline slightly as reads are added even on this plateau. This sampling depth that arrives at good model fit is about three times the coverage that was needed for the first peak to arise in the histogram (supplementary table S4, Supplementary Material online). Most of the sequencing effort needed to obtain a credible result is therefore after the first peak has arisen.

Fig. 4.

Fig. 4.

Open in a new tab

GenomeScope convergence for eight siphonophore species shown in Fig. 3. Vertical lines indicate where the models reached 95% model fit, in the six specimens with good model fit.

We found that a popular manual method of genome size estimation does not converge (supplementary fig. S3, Supplementary Material online; see Materials and Methods for more information). Genome size estimates decrease as additional sequence data are added. This indicates that the manual method is not reliable, and parametric tools such as GenomeScope should be used.

We calculated minimum genome sizes for all specimens with no peak in the k-mer spectra (Table 1) by dividing the number of base pairs sequenced by 20. This was motivated by the observation that peaks arise in the k-mer spectra by 15.65 coverage in all samples with good model fit (supplementary table S3, Supplementary Material online). Rounding up slightly gives a conservative estimate. Genome sizes could exceed these minimal sizes by a factor of 3, given that we saw peaks with as little as 8.33× coverage (supplementary table S3, Supplementary Material online).

Mitochondrial Genome Structure and Diversity

Mitochondrial genomes were assembled for all specimens, except for Hippopodius hippopus and Lilyopsis fluoracantha, which were partially assembled (Fig. 5; supplementary fig. S4, Supplementary Material online). The assembler indicated that all mitochondrial genomes were linear, except that of Chelophyes appendiculata, which was annotated as circular. Closer inspection with read mapping (see Materials and Methods) revealed that the mitochondrial genome of C. appendiculata was linear, indicating that the circular annotation by the assembler was incorrect (supplementary fig. S5, Supplementary Material online). Nine of the siphonophore species’ mitochondrial genomes were tested using this read mapping method (supplementary fig. S5, Supplementary Material online), and they were all linear. These nine species cover the main lineages of siphonophores as resolved in Munro et al. (2018) and include two cystonects, three calycophorans, three members of Euphysonectae Clade A physonects, and one member of Euphysonectae Clade B (clade designations can be found in Munro et al. 2018). These results confirm that siphonophores, like other medusozoans, have linear mitochondrial genomes.

Fig. 5.

Fig. 5.

Open in a new tab

Diversity of siphonophore mitochondrial gene order. Squares indicate nodes constrained to be congruent with Munro et al. (2018). None of the constrained nodes had strong conflict between the mitochondrial and nuclear data; they were well supported in the nuclear data and had poor support in the mitochondrial data (see supplementary fig. S6, Supplementary Material online, for the unconstrained mitochondrial phylogeny). Bootstrap values less than 100 are indicated at the nodes; where no numerical value is indicated at unconstrained nodes, the bootstrap value is equal to 100. Mitochondrial rRNA and protein-coding gene arrangements of siphonophores are depicted on the right. cox1_c is an incomplete copy of cox1 and lacks the 5′ end of the gene.

Phylogenetic analysis of the mitochondrial protein-coding genes had generally poor support for relationships within siphonophores and for the relationship of siphonophores to other hydrozoans (supplementary fig. S6, Supplementary Material online). This is consistent with low support seen for phylogenetic analyses of mitochondrial sequences in other groups (Kayal et al. 2015). Some nodes of the mitochondrial phylogenetic tree were constrained (Fig. 5) to be consistent with major clades in previous phylogenies (Munro et al. 2018). These constraints did not violate any nodes with 100% support in the unconstrained tree (supplementary fig. S6, Supplementary Material online), but given the poor support of the underlying tree and the use of constraints, this should not be considered an independent examination of the siphonophore phylogeny.

Mitochondrial gene order is conserved between multiple nonsiphonophore hydrozoans, Cystonectae, Apolemiidae (Fig. 5), Bargmannia elongata, and multiple members of Euphysonectae Clade B. We refer to this conserved gene order as Order 1, which was also found in the two previously published cystonect mitochondrial genomes of P. physalis and Rhizophysa (Kayal et al. 2015). Physonect sp. (undescribed) and both specimens of R. ornicephala showed relatively minor variations on Order 1, with cox1 shifted from one end to the other and reversed. The gene order of Nanomia, Halistemma, and Agalma was also similar to Order 1, except that the location of atp8 was changed. However, Forskalia sp., Cordagalma ordinatum, and all members of Calycophorae showed radically rearranged gene orders (Fig. 5). We mapped reads back to all the rearranged mitochondrial genomes and found strong support for the assemblies. This is evident in both visual assessment of the mappings (supplementary fig. S7, Supplementary Material online) and through computational confirmation that reads span all adjacent nucleotides in these genomes, apart from the very ends in a few species where there were several unspanned nucleotides. These few locations were all within 10 bp of the termini and had no impact on gene order.

The atp8 gene has considerable sequence variation in some siphonophores and was initially missing from some mitochondrial genome annotations. We searched for it manually with a variety of methods in species where it was not annotated (see Materials and Methods) and were able to find it in the mitochondrial genomes of all species except Chuniphyes multidentata, Diphyes dispar, Abylopsis tetragona, and C. appendiculata. Despite extensive variation at the sequence level, the 3D structure of a single alpha helix is quite conserved.

Most siphonophore mitochondrial genomes we assembled have two tRNAs, one each for methionine (Met) and tryptophan (Trp). This is consistent with the tRNA inventory of many other cnidarians (Kayal et al. 2015). The exceptions were D. dispar, A. tetragona, and C. appendiculata, for which we identified no tRNAs. These three species form a clade, consistent with a shared loss of tRNA in this group. In addition, tRNATrp was not found in Frillagalma vityazi.

Discussion

Our results confirm that siphonophores have the largest known nuclear genomes within Cnidaria and reveal striking variation in genome size within the group. We were able to confidently estimate genome sizes for six samples that had good model fit (Table 1 and Fig. 3). These specimens include some species with larger genomes that we sequenced more deeply and others that have relatively smaller genomes for siphonophores. All, though, have very large genomes relative to other cnidarians (Fig. 6). For the 25 specimens that had no peaks in the k-mer spectra at all, we estimated their minimum genome sizes to be between 1.4 and 5.6 Gb (Table 1). These minima were obtained by dividing the number of base pairs sequenced by 20, since our incremental k-mer counting analyses found a peak in all k-mer spectra of samples, which had good model fit by 15.65× coverage (supplementary table S4, Supplementary Material online). These genome size minima are based on approximate methods and are quite conservative (in part because we rounded up to 20), meaning the genomes could be considerably larger.

Fig. 6.

Fig. 6.

Open in a new tab

Cnidarian genome sizes and genome size estimates arranged by group. All Siphonophora represented here are newly sequenced for this project and are bolded.

The smallest observed genomes are scattered across the siphonophore phylogeny and nested within groups with much larger genomes. Thus, our findings suggest that the most recent common ancestor of siphonophores already had a very large genome relative to other cnidarians, and that there have been multiple independent reductions in genome size within isolated lineages. The most striking reduction of genome size we identified was within Nanomia. We included two specimens, one from California and one from North Carolina. The California specimen has a genome size of 1.5 Gb, within the range of many other siphonophores, while the North Carolina specimen has a genome of only 700 Mb, one of the two smallest siphonophore genomes we found (Fig. 3 and Table 1). Remarkably, both of these specimens come from populations that are commonly referred to as a single species, N. bijuga, though the branch length separating them is longer than for other conspecific specimens (Dunn et al. 2005). Haeckel (1888) described eight distinct species, which Bargmann and Totton (1965) later synonymized as all N. bijuga, indicating that there are morphological differences among Nanomia sp. populations. Taken together, this evidence strongly suggests these are different species, though a taxonomic revision is beyond the scope of this paper. In the meantime, we refer to the North Carolina specimen as N. bijuga and the California specimen as Nanomia California. These results show that genome skimming, even without genome assembly, can provide insight into questions about systematic biology. It is notable that the genomes differ in size by approximately 2-fold, and that N. bijuga North Carolina has a lower fraction of repeat sequences, 44.7%, than does Nanomia California, 63.4% (Table 1). This suggests that a possible reduction in the size of the N. bijuga North Carolina could be due to a reduction in repetitive regions. Full genome sequences for both of these taxa will be a fascinating opportunity to further understand radical changes in cnidarian genome size over relatively short evolutionary time scales.

Genome sizes were previously estimated with cell flow cytometry for two siphonophores collected in Japan, A. elegans at 3.482 Gb and Physalia at 3.247 Gb (Adachi et al. 2017). These sizes differ considerably from the estimates we obtained of 2.3 Gb for A. elegans (Pacific) and 1.7 Gb for Physalia Guam. The specimens were collected at different locations for the two studies and, similar tor Nanomia, there may be more diversity than is commonly appreciated within these species, including variation in genome sizes across populations, which may even reflect different species. Alternatively, there could be technical variations between cell flow cytometry and k-mer analyses. Additional samples will be required to resolve this.

Siphonophore mitochondrial genome assemblies revealed remarkable conservation of gene order in early siphonophore evolution, followed by bursts of changes in gene order within Eucladophora (Damian-Serrano et al. 2021). Previous studies have identified 4 main mitochondrial gene orders in Medusozoa (Kayal et al. 2015), but we found more than 10 gene orders in siphonophores. Mitochondrial gene Order 1 (Fig. 5) spans the most recent common ancestor of siphonophores and is also present in outgroup taxa, suggesting that this is the ancestral mitochondrial gene order of siphonophores. The mitochondrial gene order within medusozoans is well conserved, and the predicted Medusozoa ancestral mitochondrial gene order is shared by most of the Scyphozoa, Staurozoa, Cubozoa (ignoring genome fragmentation), and Trachylina (a subclade of Hydrozoa). However, the order underwent some changes in the ancestor of Hydroidolina (the clade that includes siphonophores), such as loss of two nonstandard protein-coding genes polB and ORF314 and inversion and translocation of cox1 (Kayal et al. 2015). The ancestral mitochondrial gene order of siphonophores (Order 1) is consistent with the ancestral mitochondrial gene order of Hydroidolina.

There are several striking features of the diversity of mitochondrial gene order within Siphonophora. We found that tRNAMet(CAU) and tRNATrp(UCA) are lost in three species, D. dispar, A. tetragona, and C. appendiculata, that form a clade in Calycophorae. In addition, tRNATrp(UCA) could not be found in F. vityazi. There are documented cases of tRNA loss in other nonbilaterian species (Lavrov and Pett 2016), including in ctenophores (Pett et al. 2011; Pett and Lavrov 2015), some sponge lineages (Wang and Lavrov 2008), and some cnidarians (Kayal et al. 2012). Octocorals lacking tRNATrp(UCA) rarely use TGA codons in their mitochondrial coding sequences (Pont-Kingdon et al. 1998; Brugler and France 2008; Pett and Lavrov 2015). However, this phenomenon was not found in tRNATrp(UCA)-deficient D. dispar, A. tetragona, C. appendiculata, and F. vityazi, which, like N. bijuga, C. multidentata, G. lapislazula, and R. ornicephala with tRNATrp(UCA), preferred to use TGA rather than TGG to encode Trp in mitochondrial coding sequences (supplementary table S5, Supplementary Material online). This indicates that there are may be atypical tRNATrp(UCA) in the mitochondrial genomes of D. dispar, A. tetragona, C. appendiculata, and F. vityazi that we have not been able to identify. Alternatively, there may be unknown mechanisms for decoding mitochondrial TGA codon, such as in Leishmania tarentolae, and the tRNATrp(CCA) in nuclear genome undergoes a specific C-to-U nucleotide modification in the first position of the anticodon to allow the decoding of the mitochondrial TGA codons as Trp (Alfonzo et al. 1999), which may also be present in ctenophore Mnemiopsis leidyi (Pett et al. 2011).

The remarkable conservatism of mitochondrial gene order across Hydrozoa and many siphonophores contrasts sharply with the radical rearrangements seen within Forskalia sp., C. ordinatum, and all members of Calycophorae (Fig. 5). They are so rearranged that it is difficult to find any pattern, including any correlation with nuclear genome size because of the few species with good model fits. With the exception of Praya and Craseoa lathetica, specimens with highly reorganized mitochondrial genomes also have exceptionally long branches in the mitochondrial protein sequence phylogeny (supplementary fig. S6, Supplementary Material online). Rearrangements of the mitochondrial gene order in siphonophores therefore tend to be associated with an elevated rate of evolution of the gene sequences themselves (Fig. 5). Previous studies have shown that many members of Hydroidolina have duplicated cox1 at each end of their mitochondrial chromosome(s) (Kayal et al. 2012, 2015). We found that the mitochondrial genomes of Cystonectae (including Physalia Guam, R. eysenhardtii, and Rhizophysa filiformis), B. elongata, S. amphytridis, Erenna sirena, and Stephalia dilata have a complete cox1 at one end (downstream of cob) and a very short duplicated 3′ end of cox1 (20 to 30 amino acids) at the other end (downstream of rrnL). However, similar gene duplication events were not detected in other siphonophores, suggesting that duplicated genes at the ends of mitochondrial chromosomes are not universal in Hydroidolina. Another notable feature is that Physonect sp. and the two R. ornicephala specimens share a unique gene order, but are not monophyletic, suggesting convergence.

The assembler (NOVOPlasty) indicated that our assembled mitochondrial genome of C. appendiculata was circular. This would be very surprising given the consistent linear mt genomes of Medusozoa (Bridge et al. 1992). Further inspection with read mapping indicated that this genome is linear (supplementary fig. S5, Supplementary Material online), and that the circular annotation was spurious. We note that some other medusozoan genomes deposited in GenBank are annotated as circular, such as Nemopilema nomurai (KY454767), Rhopilema esculentum (KY454768), Catostylus townsendi (OK299144), Chrysaora pacifica (MN448506), and Haliclystus antarcticus (KU947038), among others. We suspected that these were also spurious annotations. SRA data is available for N. nomurai, R. esculentum, and C. townsendi so we tested whether the reads for these species passed through the head–tail junction. They did not pass through the junctions (supplementary fig. S8, Supplementary Material online), corroborating our hypothesis that these mitochondrial genomes are actually linear. These results show that assembler software seems to be biased toward circular annotations, and these results should always be critically evaluated.

In addition to the biological insight gained with these genome skimming analyses, they have technical value for prioritizing future siphonophore genome projects. Though largely relative to other cnidarian genomes, siphonophore genomes for which we obtained confident genome size estimates can feasibly be sequenced and assembled to chromosome scale using standard modern approaches. Given their tractable genome sizes, phylogenetic distribution, and interesting biology, all six species with good model fit (Fig. 3) are promising priorities for future de novo siphonophore genome assembly projects.

Though the roughly 190 described siphonophores make up a tiny fraction of the more than 10,000 cnidarian species, this deep dive within this single clade greatly enriches our understanding of nuclear and mitochondrial genome diversity in Cnidaria. Other poorly known clades may have similar surprises to offer and may benefit from genome skimming to better understand their genome diversity in the context of evolution.

Materials and Methods

Summary of Genomes across Cnidaria

Cnidarian genome sizes were retrieved in November 2023 from https://www.ncbi.nlm.nih.gov/data-hub/genome/.

Sampling

Out of the 35 specimens examined in this study, 23 were collected in a previous study of siphonophore phylogenetics (Dunn et al. 2005). The remaining 12 specimens are new. Data for each specimen can be found in Supplementary table S1. Most were collected by deep-sea submersible off the coast of California by Monterey Bay Aquarium Research Institute from the research vessel R/V Western Flyer. We also collected a Physalia in Guam that was preserved in ethanol. Specimens collected by the R/V Western Flyer were frozen with liquid nitrogen, homogenized by mortar and pestle, and gDNA extracted with E.Z.N.A. Mollusc DNA Kit (Cat. #D3373-01). The Physalia specimen was ground up via mortar and pestle and then gDNA extracted using the QIAGEN DNeasy Blood & Tissue Kit (Cat. #69504). Samples were quantified via NanoDrop and Qubit.

Sequencing

Whole-genome shotgun (WGS) library preparation and sequencing was completed by the Yale Center for Genome Analysis (YCGA). The WGS libraries were prepared with the IDT xGen DNA Library Prep EZ Kit (Cat. #10009821). They were then sequenced on an Illumina NovaSeq6000 instrument using a S4 flow cell in a 2 × 150 bp paired-end run. The numbers of reads and bases sequenced for each specimen are indicated in Table 1.

Genome k-mer Analysis

We developed a Snakemake pipeline to complete the following analysis (Köster and Rahmann 2012). Low-quality regions and adapters were trimmed with Trimmomatic v0.39 (Bolger et al. 2014) prior to all downstream k-mer analyses. Only reads that remained paired after trimming were retained. Potential contamination was evaluated with Kraken v2.1.3 (Wood and Salzberg 2014) and its standard database of 100 GB, which includes bacteria, archaea, virus, and human genomes along with a collection of known vectors. This partitioned the full set of reads into two read sets, classified reads with similarity to organisms in the Kraken database and therefore potential contamination and unclassified reads for which there is no identified potential contamination. Unless otherwise noted, analyses were conducted on the full set of reads. Given the high false-positive rate of Kraken2 for human sequences (Garrido-Sanz et al. 2022), we also mapped 1 million reads from each sample to the human reference genome GRCh38 with bwa v0.7.17 (Li and Durbin 2009) to more rigorously evaluate the possibility of human contamination.

We used Jellyfish v2.3.0 (Marçais and Kingsford 2011) to count k-mers (k = 21) and GenomeScope 2.0 (Vurture et al. 2017; Ranallo-Benavidez et al. 2020) to analyze k-mer spectra. We conducted Jellyfish + GenomeScope analyses on the full read set and on the unclassified read set (i.e. reads that were not classified as contaminants by Kraken). In preliminary analyses, we also tested k = 17 on the full read set, but found little difference from k = 21 analyses and did not pursue these analyses further.

Manual Genome Size Estimation

In addition to the parametric models of GenomeScope2, we also applied a standard manual method for genome size estimation from k-mer spectra with one or more peaks. This generally follows the approach outlined at https://bioinformatics.uconn.edu/genome-size-estimation-tutorial/. The histogram produced by Jellyfish has two columns, which we refer to as X and Y. X is the count of k-mer occurrences in the input sequences, and Y is the number of k-mers with that count. We identified two X values by manual inspection of each histogram. Xmin is the X position of the first minimum. All counts to the left of Xmin are considered to be sequencing errors and are excluded from further consideration. Xpeak is the X position of the first peak. We then estimated the haploid genome length L as follows:

L=∑n=Xmin10,000XnYn2Xpeak.

The 2 is in the denominator since the first peak is the heterozygous peak. The sum is through 10,000 since this is the maximum included in the Jellyfish histogram output.

Incremental k-mer Counting and Minimum Genome Sizes

To identify the read coverage (i.e. the total length of sequencing reads divided by the haploid genome size) required to observe a peak in the k-mer histogram and to assess the stability of the genome size estimates relative to depth of sequencing, we performed incremental k-mer analyses that assessed progressively larger subsets of the data for the eight samples that had poor fit or good fit. These data subsets ranged in size from 1% to 100% of the data in 1% increments.

We initially attempted to construct these incremental subsamples by subsetting reads and rerunning Jellyfish on each subset, but that was not tractable given the large computational and disk space needs. We therefore wrote sharkmer (https://github.com/caseywdunn/sharkmer), a new incremental k-mer counter implemented in rust. Commit aed4b426 was used for the analyses presented here. It counts the k-mers in n hashes and then combines these chunks to provide incremental k-mer counts. We set n = 100 so that we could obtain k-mer counts at 1% intervals. This strategy is faster than creating incrementally larger data sets and analyzing each with Jellyfish, but requires a large amount of RAM. We therefore capped the number of reads considered at 1 billion. This truncated the very large data sets of A. elegans Pacific, Physalia Guam, and R. ornicephala T898-D2 (Supplementary table S3). We compare the results of sharkmer to those of KMC and Jellyfish, using the truncated data sets where the read number exceeded 1 billion, to validate this new tool.

The output was used to generate videos that show how the k-mer spectra change as data are added, as well as summary statistics that reveal how many reads are needed to estimate genome size and how stable those estimates are.

Mitochondrial Genome Assembly and Annotation

Most mitochondrial genomes were de novo assembled with NOVOPlasty v4.3.1 (Dierckxsens et al. 2017). The mitogenomes of Bargmannia lata and F. vityazi were assembled by GetOrganelle v1.7.6 (Jin et al. 2020). Partial mitogenomes of H. hippopus and L. fluoracantha were assembled by SPAdes v3.13.0 (Prjibelski et al. 2020). To determine whether the siphonophore mitochondrial genome is linear or circular, we split the genome sequences at nad5 (nad5 was chosen because it is the longest gene in the mitochondrial genome) and manually reconnected the head and tail of the original sequence, with the head of the new sequence being the second half of nad5 and the tail being the front half of nad5. Sequencing reads were then mapped onto these spliced sequences. If reads spanned the head–tail junction, it indicated that the sequence is circular; otherwise, it is linear (supplementary fig. S5, Supplementary Material online). The software used for read mapping were Bowtie 2 v2.4.1 (parameters and other information are in supplementary tables S6 to S8, Supplementary Material online; Langmead and Salzberg 2012) and SAMtools v1.17 (Danecek et al. 2021) with Tablet v1.21.02.08 (Milne et al. 2013) visualization.

Annotation of the mitochondrial genomes was conducted with MITOS2 web server (Donath et al. 2019) with the Mold, Protozoan, and Coelenterate Mitochondrial Code and the Mycoplasma/Spiroplasma Code (NCBI translation table 4), using default settings and canceling the circular setting. Then, genome annotation of start and stop positions of each gene was manually adjusted using Artemis (Rutherford et al. 2000). We verified the presence and absence of tRNA in siphonophores using the tRNA annotation tools MITOS2 and tRNAscan-SE.

We also used read mapping to verify the assembly of mitochondrial genomes that showed evidence of rearrangement; 10M to 50M reads (supplementary table S7, Supplementary Material online) were mapped to each of these genomes using Bowtie 2 v2.4.1 in end-to-end mode. These were assessed in two ways. First, by visual inspection (supplementary fig. S7, Supplementary Material online). Second, we wrote a python script (analyses_mitochondria/count_spanning_reads.py in the git repository) that parses cigar strings from each read mapping and counts the number of mappings that span from one location to the adjacent location to its right. A value of zero indicates that there are no mapped reads that connect a position in the assembly to the position to its immediate right.

ATP8 had low sequence similarity to known sequences in some species, which required a manual search process. This involved using a combination of methods to find atp8, including BLAST version 2.2.26 and BLAST+ 2.13.0 (Altschul et al. 1990, 1997; Camacho et al. 2009) and a deep learning–based homolog detection tool, PROST (Rives et al. 2021; Kilinc et al. 2023), and looking for atp8 start and stop sites within open reading frames. Each potential atp8 sequence was folded using AlphaFold2 and ColabFold v1.5.3 (Jumper et al. 2021; Mirdita et al. 2022; Varadi et al. 2022) to see if they matched the single alpha helix shape of other annotated atp8 genes of siphonophores. Additionally, we noted a pattern of atp8 genes being neighbors with atp6, cox3, and nad2, so we looked for open reading frames in the gaps between these genes. Mitochondrial genomes were updated to reflect newly found atp8 genes as seen in Fig. 5.

Phylogenetic Analysis

The amino acid sequences of 13 mitochondrial protein encoding genes for siphonophores (33 specimens of the complete mitochondrial genome were assembled), 22 other medusozoans, and 23 anthozoans were used for phylogenetic analysis (see supplementary table S5, Supplementary Material online, for accession numbers of previously published genomes included in these analyses). Each gene was individually aligned in MAFFT v7.310 (Katoh and Standley 2013) and trimmed using Trimal v1.4.1 (Capella-Gutiérrez et al. 2009) with default parameters to discard ambiguously aligned sites. Gene alignments were then concatenated into a final supermatrix. Unconstrained maximum likelihood (ML) phylogenetic analysis was then conducted using IQ-Tree v2.0.3 (Minh et al. 2020) with the option -m MFP and evaluated with 1,000 fast-bootstrapping replicates (supplementary fig. S6, Supplementary Material online). Given the broader sequence sampling of the transcriptome phylogeny, we ran constrained inferences after clamping the five nodes (squares in Fig. 5) to be consistent with the topology of the tree in Munro et al. (2018). Given these constraints, the present paper should not be considered an independent analysis of phylogenetic relationships within Siphonophora.

Supplementary Material

evae048_Supplementary_Data
evae048_supplementary_data.pdf (6.6MB, pdf)

Acknowledgments

Thanks to Lourdes Rojas and Eric Lazo-Wasem at the Yale Peabody Museum of Natural History. Thank you to the NIH T32 Training Grant in Genetics at Yale. All library preparation and sequencing were conducted at the Yale Center for Genomic Analysis. All analyses were performed on computer clusters operated by the Yale Center for Research Computing.

Contributor Information

Namrata Ahuja, Department of Ecology and Evolutionary Biology, Yale University, New Haven, CT, USA.

Xuwen Cao, Institute of Marine Science and Technology, Shandong University, Qingdao 266237, China.

Darrin T Schultz, Department of Neuroscience and Developmental Biology, University of Vienna, Vienna 1010, Austria.

Natasha Picciani, Department of Ecology and Evolutionary Biology, Yale University, New Haven, CT, USA.

Arianna Lord, Museum of Comparative Zoology, Department of Organismic and Evolutionary Biology, Harvard University, Cambridge, MA, USA.

Shengyuan Shao, Institute of Marine Science and Technology, Shandong University, Qingdao 266237, China.

Kejue Jia, Department of Molecular, Cellular and Developmental Biology, Yale University, New Haven, CT, USA.

David R Burdick, University of Guam Marine Laboratory, Mangilao, GU.

Steven H D Haddock, Monterey Bay Aquarium Research Institute, Moss Landing, CA, USA.

Yuanning Li, Institute of Marine Science and Technology, Shandong University, Qingdao 266237, China.

Casey W Dunn, Department of Ecology and Evolutionary Biology, Yale University, New Haven, CT, USA.

Supplementary Material

Supplementary material is available at Genome Biology and Evolution online.

Author Contributions

N.A.: new sample collection, new DNA extractions, coordination with sequencing center, implemented k-mer analysis workflow, and manuscript writing. D.T.S.: project design, developed k-mer analysis workflow, and interpretation of k-mer spectra. C.W.D.: project design, data analysis, implementation of sharkmer, data submission, and manuscript writing. N.P.: implemented k-mer analysis workflow and manuscript edits. A.L.: summary of existing cnidarian genomes. X.C.: assembly, annotation, and analysis of mitochondrial genomes and manuscript writing. S.H.D.H.: new sample collection and project design. Y.L.: assembly, annotation, and analysis of mitochondrial genomes and manuscript writing. K.J.: atp8 annotation in siphonophore mitochondrial genomes. D.R.B.: collected Physalia specimen. SS: assistance with mitochondrial genome assembly.

Funding

Support for this project came from multiple sources including the National Science Foundation grants NSF-OCE to C.W.D. and NSF-OCE 1829805 to S.H.D.H, the Tal Waterman Fund at Yale, the Shandong Provincial Natural Science Foundation (grant number: ZR2023QC268), and the Qingdao Postdoctoral Applied Research Project (grant number: QDBSH20220202163). D.T.S. was supported by European Research Council (ERC) under the European Union’s Horizon 2020 Programme (grant number 945026). S.H.D.H. was supported by the David and Lucile Packard Foundation. D.R.B was supported by NSF Established Program to Stimulate Competitive Research (EPSCoR) grant (grant number OIA-1946352).

Data Availability

Code and other analysis components are available at https://github.com/dunnlab/siph_skimming. Sequence reads have been deposited in NCBI SRA as BioProject PRJNA925656. Annotated mitochondrial genomes have been deposited at NCBI with accession numbers OQ957189 to OQ957220.

Literature Cited

  1. Adachi K, Miyake H, Kuramochi T, Mizusawa K, Okumura S-I. Genome size distribution in phylum Cnidaria. Fish Sci. 2017:83(1):107–112. 10.1007/s12562-016-1050-4. [DOI] [Google Scholar]
  2. Alfonzo JD, Blanc V, Estevez AM, Rubio MA, Simpson L. C to U editing of the anticodon of imported mitochondrial tRNATrp allows decoding of the UGA stop codon in Leishmania tarentolae[J]. EMBO J. 1999:18(24):7056–7062. 10.1093/emboj/18.24.7056. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. J Mol Biol. 1990:215(3):403–410. 10.1016/S0022-2836(05)80360-2. [DOI] [PubMed] [Google Scholar]
  4. Altschul SF, Madden TL, Schäffer AA, Zhang J, Zhang Z, Miller W, Lipman DJ. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 1997:25(17):3389–3402. 10.1093/nar/25.17.3389. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bargmann HE, Totton AK. A synopsis of the Siphonophora. By A.K. Totton assisted by H.E. Bargmann, etc. London: Trustees of the British Museum (Natural History); 1965. [Google Scholar]
  6. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 2014:30(15):2114–2120. 10.1093/bioinformatics/btu170. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bridge D, Cunningham CW, Schierwater B, DeSalle R, Buss LW. Class-level relationships in the phylum Cnidaria: evidence from mitochondrial genome structure. Proc Natl Acad Sci U S A. 1992:89(18):8750–8753. 10.1073/pnas.89.18.8750. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Brugler MR, France SC. The mitochondrial genome of a deep-sea bamboo coral (Cnidaria, Anthozoa, Octocorallia, Isididae): genome structure and putative origins of replication are not conserved among octocorals. J Mol Evol. 2008:67(2):125–136. 10.1007/s00239-008-9116-2. [DOI] [PubMed] [Google Scholar]
  9. Camacho C, Coulouris G, Avagyan V, Ma N, Papadopoulos J, Bealer K, Madden TL. BLAST+: architecture and applications. BMC Bioinformatics 2009:10(1):421. 10.1186/1471-2105-10-421. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Capella-Gutiérrez S, Silla-Martínez JM, Gabaldón T. Trimal: a tool for automated alignment trimming in large-scale phylogenetic analyses. Bioinformatics 2009:25(15):1972–1973. 10.1093/bioinformatics/btp348. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Damian-Serrano A, Haddock SHD, Dunn CW. The evolution of siphonophore tentilla for specialized prey capture in the open ocean. Proc Natl Acad Sci U S A. 2021:118(8):e2005063118. 10.1073/pnas.2005063118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Danecek P, Bonfield JK, Liddle J, Marshall J, Ohan V, Pollard MO, Whitwham A, Keane T, McCarthy SA, Davies RM, et al. Twelve years of SAMtools and BCFtools. GigaScience. 2021:10(2). 10.1093/gigascience/giab008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Deorowicz S, Debudaj-Grabysz A, Grabowski S. Disk-based k-mer counting on a PC. BMC Bioinformatics 2013:14(1):160. 10.1186/1471-2105-14-160. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Deorowicz S, Kokot M, Grabowski S, Debudaj-Grabysz A. KMC 2: fast and resource-frugal k-mer counting. Bioinformatics 2015:31(10):1569–1576. 10.1093/bioinformatics/btv022. [DOI] [PubMed] [Google Scholar]
  15. Dierckxsens N, Mardulyn P, Smits G. NOVOPlasty: de novo assembly of organelle genomes from whole genome data. Nucleic Acids Res. 2017:45(4):e18. 10.1093/nar/gkw955. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Donath A, Jühling F, Al-Arab M, Bernhart SH, Reinhardt F, Stadler PF, Middendorf M, Bernt M. Improved annotation of protein-coding genes boundaries in metazoan mitochondrial genomes. Nucleic Acids Res. 2019:47(20):10543–10552. 10.1093/nar/gkz833. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Dunn CW, Pugh PR, Haddock SHD. Molecular phylogenetics of the siphonophora (Cnidaria), with implications for the evolution of functional specialization. Syst Biol. 2005:54(6):916–935. 10.1080/10635150500354837. [DOI] [PubMed] [Google Scholar]
  18. Garrido-Sanz L, Àngel Senar M, Piñol J. Drastic reduction of false positive species in samples of insects by intersecting the default output of two popular metagenomic classifiers. PLoS One 2022:17(10):e0275790. 10.1371/journal.pone.0275790. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Haeckel E. Report on the Siphonophorae collected by H.M.S. Challenger during the years 1873-1876. UK: H.M. Stationery Office; 1888. [Google Scholar]
  20. Heath-Heckman E, Nishiguchi MK. Leveraging short-read sequencing to explore the genomics of sepiolid squid. Integr Comp Biol. 2021:61(5):1753–1761. 10.1093/icb/icab152. [DOI] [PubMed] [Google Scholar]
  21. Hetherington ED, Damian-Serrano A, Haddock SHD, Dunn CW, Choy CA. Integrating siphonophores into marine food-web ecology. Limnol Oceanogr Lett. 2022:7(2):81–95. 10.1002/lol2.10235. [DOI] [Google Scholar]
  22. Hogan RI, Hopkins K, Wheeler AJ, Yesson C, Allcock AL. Evolution of mitochondrial and nuclear genomes in Pennatulacea. Mol Phylogenet Evol. 2022:178:107630. 10.1016/j.ympev.2022.107630. [DOI] [PubMed] [Google Scholar]
  23. Jin J-J, Yu W-B, Yang J-B, Song Y, dePamphilis CW, Yi T-S, Li D-Z. GetOrganelle: a fast and versatile toolkit for accurate de novo assembly of organelle genomes. Genome Biol. 2020:21(1):241. 10.1186/s13059-020-02154-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Jumper J, Evans R, Pritzel A, Green T, Figurnov M, Ronneberger O, Tunyasuvunakool K, Bates R, Žídek A, Potapenko A, et al. Highly accurate protein structure prediction with AlphaFold. Nature 2021:596(7873):583–589. 10.1038/s41586-021-03819-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Katoh K, Standley DM. MAFFT multiple sequence alignment software version 7: improvements in performance and usability. Mol Biol Evol. 2013:30(4):772–780. 10.1093/molbev/mst010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Kayal E, Bentlage B, Cartwright P, Yanagihara AA, Lindsay DJ, Hopcroft RR, Collins AG. Phylogenetic analysis of higher-level relationships within Hydroidolina (Cnidaria: Hydrozoa) using mitochondrial genome data and insight into their mitochondrial transcription. PeerJ 2015:3:e1403. 10.7717/peerj.1403. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Kayal E, Bentlage B, Collins AG, Kayal M, Pirro S, Lavrov DV. Evolution of linear mitochondrial genomes in medusozoan cnidarians. Genome Biology and Evolution. 2012:4(1):1–12. 10.1093/gbe/evr123. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Kilinc M, Jia K, Jernigan RL. Improved global protein homolog detection with major gains in function identification. Proc Natl Acad Sci U S A. 2023:120(9):e2211823120. 10.1073/pnas.2211823120. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Kokot M, Długosz M, Deorowicz S. KMC 3: counting and manipulating k-mer statistics. Bioinformatics 2017:33(17):2759–2761. 10.1093/bioinformatics/btx304. [DOI] [PubMed] [Google Scholar]
  30. Köster J, Rahmann S. Snakemake—a scalable bioinformatics workflow engine. Bioinformatics 2012:28(19):2520–2522. 10.1093/bioinformatics/bts480. [DOI] [PubMed] [Google Scholar]
  31. Langmead B, Salzberg SL. Fast gapped-read alignment with Bowtie 2. Nature Methods. 2012:9(4):357–359. 10.1038/nmeth.1923. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Lavrov DV, Pett W. Animal mitochondrial DNA as we do not know it: mt-genome organization and evolution in nonbilaterian lineages. Genome Biol Evol. 2016:8(9):2896–2913. 10.1093/gbe/evw195. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Li H, Durbin R. Fast and accurate short read alignment with Burrows–Wheeler transform. Bioinformatics. 2009:25(14):1754–1760. 10.1093/bioinformatics/btp324. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Ling MK, Yap NWL, Iesa IB, Yip ZT, Huang D, Quek ZBR. Revisiting mitogenome evolution in Medusozoa with eight new mitochondrial genomes. iScience 2023:26(11):108252. 10.1016/j.isci.2023.108252. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Mapstone GM. Global diversity and review of Siphonophorae (Cnidaria: Hydrozoa). PLoS One 2014:9(2):e87737. 10.1371/journal.pone.0087737. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Marçais G, Kingsford C. A fast, lock-free approach for efficient parallel counting of occurrences of k-mers. Bioinformatics 2011:27(6):764–770. 10.1093/bioinformatics/btr011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Milne I, Stephen G, Bayer M, Cock PJA, Pritchard L, Cardle L, Shaw PD, Marshall D. Using Tablet for visual exploration of second-generation sequencing data. Briefings in Bioinformatics. 2013:14(2):193–202. 10.1093/bib/bbs012. [DOI] [PubMed] [Google Scholar]
  38. Minh BQ, Schmidt HA, Chernomor O, Schrempf D, Woodhams MD, von Haeseler A, Lanfear R. IQ-TREE 2: new models and efficient methods for phylogenetic inference in the genomic era. Mol Biol Evol. 2020:37(5):1530–1534. 10.1093/molbev/msaa015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Mirdita M, Schütze K, Moriwaki Y, Heo L, Ovchinnikov S, Steinegger M. ColabFold: making protein folding accessible to all. Nat Methods. 2022:19(6):679–682. 10.1038/s41592-022-01488-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Munro C, Siebert S, Zapata F, Howison M, Damian-Serrano A, Church SH, Goetz FE, Pugh PR, Haddock SHD, Dunn CW. Improved phylogenetic resolution within Siphonophora (Cnidaria) with implications for trait evolution. Mol Phylogenet Evol. 2018:127:823–833. 10.1016/j.ympev.2018.06.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Munro C, Zapata F, Howison M, Siebert S, Dunn CW. Evolution of gene expression across species and specialized zooids in Siphonophora. Mol Biol Evol. 2022:39(2):msac027. 10.1093/molbev/msac027. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Pett W, Lavrov DV. Cytonuclear interactions in the evolution of animal mitochondrial tRNA metabolism. Genome Biol Evol. 2015:7(8):2089–2101. 10.1093/gbe/evv124. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Pett W, Ryan JF, Pang K, Mullikin JC, Martindale MQ, Baxevanis AD, Lavrov DV. Extreme mitochondrial evolution in the ctenophore Mnemiopsis leidyi: insight from mtDNA and the nuclear genome. Mitochondrial DNA 2011:22(4):130–142. 10.3109/19401736.2011.624611. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Pont-Kingdon G, Okada NA, Macfarlane JL, Beagley CT, Watkins-Sims CD, Cavalier-Smith T, Clark-Walker GD, Wolstenholme DR. Mitochondrial DNA of the coral Sarcophyton glaucum contains a gene for a homologue of bacterial MutS: a possible case of gene transfer from the nucleus to the mitochondrion. J Mol Evol. 1998:46(4):419–431. 10.1007/PL00006321. [DOI] [PubMed] [Google Scholar]
  45. Prjibelski A, Antipov D, Meleshko D, Lapidus A, Korobeynikov A. Using SPAdes de novo assembler. Current Protocols in Bioinformatics. 2020:70(1). 10.1002/cpbi.102. [DOI] [PubMed] [Google Scholar]
  46. Ranallo-Benavidez TR, Jaron KS, Schatz MC. GenomeScope 2.0 and Smudgeplot for reference-free profiling of polyploid genomes. Nat Commun. 2020:11(1):1432. 10.1038/s41467-020-14998-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Rives A, Meier J, Sercu T, Goyal S, Lin Z, Liu J, Guo D, Ott M, Zitnick CL, Ma J, et al. Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences. Proc Natl Acad Sci U S A. 2021:118(15):e2016239118. 10.1073/pnas.2016239118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Robison BH. Light in the ocean’s midwaters. Scientific American. 1995:273:60–64. http://www.jstor.org/stable/24981452. [Google Scholar]
  49. Rutherford K, Parkhill J, Crook J, Horsnell T, Rice P, Rajandream M-A, Barrell B. Artemis: sequence visualization and annotation. Bioinformatics 2000:16(10):944–945. 10.1093/bioinformatics/16.10.944. [DOI] [PubMed] [Google Scholar]
  50. Siebert S, Robinson MD, Tintori SC, Goetz F, Helm RR, Smith SA, Shaner N, Haddock SHD, Dunn CW. Differential gene expression in the siphonophore Nanomia bijuga (Cnidaria) assessed with multiple next-generation sequencing workflows. PLoS One 2011:6(7):e22953. 10.1371/journal.pone.0022953. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Varadi M, Anyango S, Deshpande M, Nair S, Natassia C, Yordanova G, Yuan D, Stroe O, Wood G, Laydon A, et al. AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models. Nucleic Acids Res. 2022:50(D1):D439–D444. 10.1093/nar/gkab1061. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Vurture GW, Sedlazeck FJ, Nattestad M, Underwood CJ, Fang H, Gurtowski J, Schatz MC. GenomeScope: fast reference-free genome profiling from short reads. Bioinformatics 2017:33(14):2202–2204. 10.1093/bioinformatics/btx153. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Wang X, Lavrov DV. Seventeen new complete mtDNA sequences reveal extensive mitochondrial genome evolution within the Demospongiae. PLoS One 2008:3(7):e2723. 10.1371/journal.pone.0002723. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Wood DE, Salzberg SL. Kraken: ultrafast metagenomic sequence classification using exact alignments. Genome Biol. 2014:15(3):R46. 10.1186/gb-2014-15-3-r46. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Zapata F, Goetz FE, Smith SA, Howison M, Siebert S, Church SH, Sanders SM, Ames CL, McFadden CS, France SC, et al. Phylogenomic analyses support traditional relationships within Cnidaria. PLoS One 2015:10(10):e0139068. 10.1371/journal.pone.0139068. [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

evae048_Supplementary_Data
evae048_supplementary_data.pdf (6.6MB, pdf)

Data Availability Statement

Code and other analysis components are available at https://github.com/dunnlab/siph_skimming. Sequence reads have been deposited in NCBI SRA as BioProject PRJNA925656. Annotated mitochondrial genomes have been deposited at NCBI with accession numbers OQ957189 to OQ957220.

Articles from Genome Biology and Evolution are provided here courtesy of Oxford University Press

RESOURCES