The Organelle DNA Lineages (ODL) software was written R, version 4.0.3. It is available at https://github.com/csmatyi/odl together with all supplementary files and figures. The script is called odl5.R. As input, it takes a list with two columns: the first column contains the Latin name of the species in the study. The second column is the NCBI accession number of the oDNA sequence. Species and their corresponding accession numbers can be downloaded at the Organelle Genome Browser at https://www.ncbi.nlm.nih.gov/genome/browse#!/organelles/.

The software removes duplicate species entries. Using the “getAnnotations-GenBank” function from the “ape” package, it downloads and stores the annotation for each accession. Using the “read.GenBank” function, it retrieves the genome sequence for each accession. The software adds all of the organelle genome sequences into a DNAStringSet, and then calls the “msa” function to create a multiple sequence alignment. Then, in an all-versus-all pairwise fashion, the script calculates the sequence similarity between all possible pairs of genome sequences. The sequence similarity values are then stored in a symmetric square sequence similarity matrix.

The gene, tRNA and rRNA elements are also written to an output file, which is compatible with the input format of the CREx software [19]. The output can be directly uploaded to the CREx website at http://pacosy.informatik.uni-leipzig.de/crex. The GenBank records for all sequences are also written to a subdirectory as part of the output. These can be used as input for the GenomeVx visualization software [20].

A gene order distance matrix is calculated for each species pair by comparing the indexes of each gene/tRNA/rRNA between the two species using the following equation:

d a , b = ∑ i = 1 k | mean ( index a ( i ) ) − mean ( index b ( i ) ) | + unique ( a ) + unique ( b ) (1)

In Equation (1), the difference between the mean index value of element i (meaning genes, tRNAs, and rRNAs) between species a and b is summed up for all common elements between the two species. The sum also includes the unique number of elements pertaining to species a and b, respectively.

Figure 2 depicts a simple example of calculating the gene order distance between three hypothetical species. For example, the distance between species 1 and species 3 is 3 + 1 + 2.5 + 3 = 9.5. The distance in the index value for gene A is 4 – 1 = 3. For gene B it is 3 – 2 = 1. For gene C it is (4 + 3)/2 – 1 = 2.5. For gene D it is 5 – 2 = 3.

It is possible that a certain gene/rRNA/tRNA is present in multiple copies. In this instance, the average index value is taken for that particular gene/rRNA/tRNA. The distance value between two species may be greater than 1, since we are

looking at indexes. The gene order distance matrix is then normalized by dividing each element in the distance matrix by the maximum value of the matrix, and then subtracted from 1 to derive a gene order similarity matrix.

The genome sequence similarity matrix and the gene order similarity matrix are then visualized on a heatmap. Furthermore, these two matrixes are also combined into a “combined” matrix, giving equal weight to both the genome sequence similarity matrix and the gene order similarity matrix. The “heatmap” function is called, using the “ward.D2” method to perform species clustering. A Silhouette plot is also created by the software, which depict the average silhouette width for each possible number of clusters, showing the optimal number of clusters with a dashed line, where the average silhouette width is highest. The mean silhouette width shows how close the points in one cluster are to points in another cluster.

Clustering is performed using the “ward.D2” method on the sequence similarity matrix. Several statistics measures are written to an output file, such as the Hopkins clustering statistic, which denotes how well the matrix forms clusters. Other parameters are also recorded for each of the clusters (which correspond to monophyletic groups), including the number of species in the cluster, the average oDNA length ± one standard deviation, the minimum, mean, and maximum similarity value, as well as the standard error of the mean (SEM), as well as the p-value, which calculates statistical significance between similarity values between species within the cluster and between species within the cluster and outside the cluster. This p-value is a good measure of discontinuity between monophyletic groups.

The main output of the software is a gene order map, showing the order of genes/rRNAs/tRNAs along the organelle genome per species per cluster. A color-coded legend helps the user identify a given gene/rRNA/tRNA along the organelle genome of a given species.

GenomeVx was used to create the image of the human reference mitogenome in Figure 1.

The mitogenomes for 51 cephalopod species were downloaded from the Organelle Genome Browser, specified in the previous subsection. If a given species occurred more than once, only one of its accession number was randomly selected.

There are a lot of different databases and programs which are tailored for storing and visualizing genome and synteny. Some of these programs include software which are amenable for the visualization of organelle genomes and gene order. In the following, some of these software and databases will be reviewed. A list of the most commonly used mitogenome and plastid genome annotation tools are listed in Table 4.

GenBank and RefSeq are the standard databases used for storing sequence data, including organelle genomes. Refseq offers curated and non-redundant data for users. Several databases exist, which are built upon GenBank and RefSeq, which store data derived from these databases and which have improved annotation and data quality. Such databases include OGRe [21], MetAMIGA [22], MitoZOA [23], Mamit-tRNA [24], MamMiBase [25], and tRNAdb [26].

CREx is a software which calculates the hypothetical most parsimonious gene order rearrangement between two organelle genomes, and then visualizes the output. Possible rearrangements include transpositions, reverse transpositions, reversals and tandem duplication random losses (TDRLs) [19].

Similar to CREx, EqualTDRL is a software which compares two input genomes (i.e. two mt genomes), calculates different series of tandem duplication random losses (TDRL), which hypothetically transforms the gene order of one genome into the other. It creates a diagonally symmetric map showing all the possible series of TDRLs between the two genomes [27].

GenomeVx is a web-based software which allows the user to create editable, colorful, publication-ready images of circular genomes, such as that of mitochondrial and plastid genomes as well as large plasmids. As input, it takes raw feature positions or GenBank files, which the user can upload to the GenomeVx website [20].

Mitofish is a database of annotations and re-annotations for mitochondrial genome for a large number of fish species. Its companion program, Mitoannotator is an extremely rapid, high quality annotation pipeline for fish mitogenomes [28].

The MITOchondrial genome annotation Server (MITOS) is a well-known, high-quality mitochondrial genome annotation pipeline. MITOS uses standardized gene names and gene boundary designations. It uses BLAST to compare existing mitochondrial proteins to the raw genome sequence to annotate the location of these genes. It also has a web server where users can upload their raw genome sequences. After selecting a translation code, the server will calculate and present results for the user. These results include tabular summary of annotated genetic elements and visualization of genes within the uploaded genome sequence. The results can then be downloaded by the user [29] [30]. DOGMA is another web service, which uses BLAST against internal databases to detect protein-coding genes and also rRNAs, and use tRNAscan-SE to discover tRNAs, however, according to Guyeux et al. [31] it is outdated. DOGMA and MITOS both use metazoan databases data [29] [32].

For plastid genomes, the Chloroplast Genome Database (CpGDB) stores plastid genome and individual genes sequences, and also annotation records for 3823 species [33]. Plastid genome annotation tools include CpGAVAS (Chloroplast Genome Annotation, Visualization, Analysis, and GenBank Submission) and CpGAVAS2 [34] [35], CGAP (Chloroplast Genome Annotation Platform) [36], and GeSeq [37].

The ChloroMitoSSRDB database is an open-source repository of perfect and imperfect microsatellites, two to six nucleotides long. Information include the position of repeats, size, motif and length polymorphisms. The repeat sequences are hyperlinked to annotated gene regions at NCBI [38].

The current software differs from all the previous methods in that it clusters the organelle genomes based on both gene order and sequence similarity. It then depicts a linear organelle genome map for each cluster and species in the study showing the position of each gene. The software also produces accompanying statistics files and also heatmaps showing species relationships based on gene order and sequence similarity. Other software calculates genome distance by the number of rearrangements needed to transform the organelle genome of one species into another [39]. The present software does this by calculating the difference between the index of the order of each gene.

Since the main goal of this paper is to present a new bioinformatics tool, usage of the ODL software will be showcased here in the mitochondrial mapping of cephalopods. Cephalopods are a class of species in the phylum Mollusca (mollusks). They have two subclasses, Nautiloidea (nautiluses) and Coleoidea, which is made up of two superorders, Decapodiformes (squids and cuttlefish), and Octopodiformes (octopuses and argonauts). The mitogenomes of 51 species as well as the outlier Danio rerio were analyzed to discover putative monophyletic groups.

After running the software, ten putative monophyletic groups were discovered, besides Daniorerio, the outlier species. A list of species and accession numbers can be found in Supplementary File 1. The statistics for each cluster can be seen in Table 5. The matrixes, clusters, and statistics for the mitochondrial gene order, sequence similarity, and combined analyses can be found in Supplementary Files 2 - 4. Here results for only the combined matrix are reported. The mitogenome map can be seen in Figure 3, and the corresponding combined heatmap can be seen in Figure 4. The Hopkins clustering statistic for the combined matrix is 0.81, which denotes good clustering. Supplementary Figures 1-3 show the Silhouette plot for clustering based on gene order similarity, mitogenome sequence similarity, and combining both methods.

Argonauta and Nautilus form two separate clusters. Nautilus also has a significantly large non-coding region between the tRNAs for glutamine (Q) and threonine (T) [40]. Nautilus also has a mean GC% of 40% ± 0.26%, whereas for Argonauta this value is 22.9% ± 0.05%.

Octopodidae with 17 species is the largest and most significant monophyletic group, with a unique morphology of eight legs. These species come from the genera Amphioctopus, Callistoctopus, Cistopus, Hapalochlaena, and Octopus. The 17 octopus species in this study have a very conserved gene order and genome length between 15,479 bp and 16,084 bp. The octopus genomes have a mean GC% of 24.5% ± 0.88%. The enigmatic Vampyroteuthis infernalis [41] is classified as a member of this group.

Ten species of cuttlefish from the genera Sepia and Sepiella also form a monophyletic group. Their gene order is very conserved, but very different from all other cephalopod groups. The genome length is also very conserved, between 16,163 to 16,244 bp. The GC% is 24.6% ± 1.47%. Takumiya et al. [42] also found significant differences between coleoid cephalopods based on mtDNA, separating cuttlefish from all other groups.

Following are several squid groups, including four monophyletic groups and two species which don’t belong anywhere else. The first of the four groups is Loliginidae (pencil squids), with species from the genera Doryteuthis, Heterololigo, Loliolus, Sepioteuthis, and Uroteuthis. The species Sepioteuthis lessoniana is an outlier due to the translocation of three tRNA sequences (for isoleucine, valine and tryptophan) and l-rrna dn s-rrna in its mt genome. Yokobori et al. [43] also found that this squid group is monophyletic.

Four species, Architeuthis dux, Dosidicus gigas, Stenoteithis oualaniensis, and Todarodes pacificus form a separate group. These four species have a mean GC% of 29.8% ± 1.4% and a genome length from 20,254 to 20,331 bp. Xu, et al. [44] also separate them from all other squids.

Bathyteuthis abyssicola (the deepsea squid) belongs to its own family, Bathyteuthidae, in the suborder Oegopsina. It possibly even belongs to its own order, according to Uribe and Zardoya [45]. Kawashima et al. [46] found that the mitogenome structure of Semmirossia patagonica (Patagonian bobtail squid), from the family Sepiolidae is unique among decapods. For example, like Nautilus species, the ATP6 and ATP8 genes are not adjacent to one another. Ommastrephes bartrami (the neon flying squid) and Watasenia scintillans (the firefly squid) and Chiroteuthis picteti and Ilex argentinus cluster together into the same putative groups.