Developing a community-based genetic nomenclature for anole lizards

Background Comparative studies of amniotes have been hindered by a dearth of reptilian molecular sequences. With the genomic assembly of the green anole, Anolis carolinensis available, non-avian reptilian genes can now be compared to mammalian, avian, and amphibian homologs. Furthermore, with more than 350 extant species in the genus Anolis, anoles are an unparalleled example of tetrapod genetic diversity and divergence. As an important ecological, genetic and now genomic reference, it is imperative to develop a standardized Anolis gene nomenclature alongside associated vocabularies and other useful metrics. Results Here we report the formation of the Anolis Gene Nomenclature Committee (AGNC) and propose a standardized evolutionary characterization code that will help researchers to define gene orthology and paralogy with tetrapod homologs, provide a system for naming novel genes in Anolis and other reptiles, furnish abbreviations to facilitate comparative studies among the Anolis species and related iguanid squamates, and classify the geographical origins of Anolis subpopulations. Conclusions This report has been generated in close consultation with members of the Anolis and genomic research communities, and using public database resources including NCBI and Ensembl. Updates will continue to be regularly posted to new research community websites such as lizardbase. We anticipate that this standardized gene nomenclature will facilitate the accessibility of reptilian sequences for comparative studies among tetrapods and will further serve as a template for other communities in their sequencing and annotation initiatives.


Background
As the rate of generating new sequence assemblies continues to accelerate, the final bottleneck that remains is annotation. While automated pipelines have been developed, it is still up to community initiatives to pool, evaluate, integrate, and disseminate the necessary resources required for functional and comparative annotations that support research needs. The presence of multiple tools and resources, and changing assemblies and annotations, presents "moving-target" challenges for those attempting to assign function, orthology, nomenclature and other common vocabulary to genetic loci. One challenge is that many assemblies are, or will be, periodically updated due to resequencing efforts that aim to fill in ever-present gaps, initiatives to provide a consensus reference sequence that takes into account the polymorphism present in a species, or a re-deployment of different assembly algorithms. The second challenge is that the generation of confidently assigned gene models on a fixed assembly generally correlates with the amount of effort that a community puts into annotating their genome of interest. A third challenge relates to the principle that orthologous (and by association, functional) assignments are interdependent on the quality and quantity of annotations from closely related genomes.
The recent publication of the genome sequence of the green anole, Anolis carolinensis, offers a rich trove of opportunities for biologists [1]. Comparing vertebrate genomes holds the promise to solve such questions as unmasking the genetic basis of human disease in addition to understanding common evolutionary processes. Whole genome sequencing efforts in vertebrates have been carried out for 39 species of mammals (10 primates, 8 rodents, 12 laurasiatherians, 3 afrotherians, 2 xenarthrans, 3 marsupials, 1 monotreme), 3 birds (avian reptiles), 1 amphibian, and 5 teleost species [2,3]. Non-avian reptiles are missing from this taxonomic survey of genomes, and the publication of a whole genome assembly for the green anole helps to fill this gap [1]. As a complement to this effort, a growing number of online resources are available for the Anolis community (Table 1).
Mammals, birds, and non-avian reptiles are grouped as amniotes, due to shared features including a characteristic egg adapted to terrestrial reproduction. Within the amniotes, mammals are estimated to have diverged over 300 million years ago (mya) from the reptiles [4]. Within the Reptilia are three major lineages: the Archosauria, which contains crocodilians, dinosaurs and birds and whose most recent common ancestor lived approximately 250 mya; the Lepidosauria, which contains the Squamata (lizards and snakes) and the tuatara (a lizardlike reptile found only in New Zealand); and the Anapsida or turtles. For comparative genomic analysis, this first non-avian reptile sequence will be invaluable as an outgroup for comparative analyses of an increasing number of amniote sequences.
For the past century, A. carolinensis, which is native to the southeastern US, has been a lizard of choice for comparative studies in ecology, evolutionary biology, behavior, physiology and neuroscience. With genomic and transcriptomic sequences available, A. carolinensis is also emerging as an important model organism for cellular, molecular, developmental and regenerative studies. Furthermore, A. carolinensis is only one of over 350 described species of Anolis, making it a member of one of the most species-rich clades of tetrapods [4].
Comparative genomic research at all taxonomic levels would be facilitated by a consistent system of gene nomenclature for A. carolinensis as the first sequenced non-avian reptile. Towards this goal, members of the Anolis research community have established the Anolis Gene Nomenclature Committee (AGNC) to generate and maintain standardized gene vocabularies. As a companion to the publication of the first non-avian reptile genome, we present this report as the first step in an evolving document.

Report and Discussion
Establishing evolutionary metrics to help evaluate orthology between anoles and other vertebrates As an approach in the annotation process, finding orthologous relationships across species has become an important tool to evaluate gene identity [5]. However, determining gene orthology is not a trivial exercise. Vertebrate genomes have experienced a dynamic flux of activity from countless deletions and duplications, a constant stream of genomic rearrangements (including at least two whole genome duplications), and divergence in both gene expression and protein function. Fortunately, for many genes, orthologs can be reliably determined based on reciprocal protein similarity. For other genes, divergence in sequence requires data from synteny (gene order) conservation and functional analysis to also be considered. Below, we present the challenges involved in maintaining an evolving and communityaccepted record of gene ancestry, and briefly review the current state of assigning orthology using presently available resources and tools. Proposed criteria for evaluating gene orthology and paralogy are offered below with an aim to present a multi-metric summary for each Confidence in aligned assemblies from nearby taxa The paucity of amphibian and reptilian sequences compared with mammalian genomes presents a challenge for comparative analysis. When entire vertebrate clades depend on the annotations of a single genome, errors in comparative analysis are likely. As more annotated assemblies become available, we should be able to test and refine current assignments of orthologous and paralogous relationships. Yet, not all annotations are created equally, with model organisms such as chicken, mouse, rat and zebrafish having more comprehensive annotations due to greater allocated resources and larger active research communities. Therefore, the challenge is to develop an annotation approach that keeps pace with the rapidly expanding number of whole genome sequences being produced.
Currently available orthology pipelines Ancestral relationships between loci from selected species can be extracted via a variety of ready-built pipelines. The major databanks provide orthology/paralogy relationships for completed genomes through the implementation of well-established data workflows. Ensembl's orthology and paralogy relationships are based on a maximum likelihood tree-building algorithm, TreeBeST [7]. NCBI's Homologene uses a clustering approach based on an initial blastp search [8]. The UCSC Genome Browser also generates a comparative genomic table on selected sequenced species [9,10]. A number of other databases that specifically identify orthology/homology include the Orthologs Matrix Project (OMA) [11,12], InParanoid [13,14], TreeFam [15,16], Optic [17,18], and Evola [19,20]. Interestingly, HUGO (Human Genome Organization) has constructed a meta-comparison tool, HCOP (Human Gene Nomenclature Committee Comparison of Orthology Predictions), that records whether an orthology call has evidence in each of the beforementioned pipelines, hence, providing a valuable evaluative resource to assess overall confidence [21]. A major challenge for bioinformatics research is to keep up with an ever-changing landscape of software tools. Workflow evaluations must be performed on a regular basis by computer-savvy researchers but, most importantly, the results must be validated by knowledgeable biologists.

Towards community-driven evaluations of orthology
With an accelerated increase in genomic sequence data, even a well-organized mechanism to assign orthology can be overwhelmed. A community-driven effort to characterize a gene's evolutionary history as well as our confidence in summarizing it will be useful to the community and beyond. We propose that the Anolis research community work together to initiate and ultimately complement these efforts to build a pipeline that follows a common set of guidelines and relationships with the large genomic databanks. Towards this end, the AGNC has established working relationships with representatives from a network of relevant databases. Developing a common set of guidelines is the major focus of the AGNC in the upcoming year. Ultimately, we aim to generate a weighted point system, considering the different types of characteristics being compared. In situations where there is still substantial ambiguity, the AGNC plans to work with the researchers and database community for preliminary assignments. In the interim, we propose the following framework as a starting point: Species/taxa for comparative analysis Multiple alignment programs such as ClustalW [22], MUSCLE [23] and T-COFFEE [24] provide accessible tools to align multiple species. The presence or absence of reliable alignments can tell us which lineage this gene is limited to. All comparative analyses should include a common starting set of genomes to align to: • Mammals: 2 eutherians, preferably mouse and human, plus marsupial and monotreme genes if available.
• Birds (avian reptiles): zebra finch and chicken • Non-avian reptiles: Any additional gene sequences as available, particularly for non-squamate species (turtles or crocodilians) • Amphibians: Xenopus tropicalis and additional genomes as available • Teleosts: Zebrafish and Fugu rubripes or Tetraodon nigroviridis should be included. Additional teleosts (stickleback, medaka) can also be analyzed.
• Non-vertebrate chordates: Either Ciona intestinalis or savignyi can serve as a stem alternative to Drosophila melanogaster, if available.
Protein sequence analysis Sequence analysis programs such as MEGA [25] and PAML [26] provide accessible tools to analyze protein alignment across multiple species. Protein divergence will be estimated using dN (amino acid divergence) and dS (silent site divergence) using a codon-substitution matrix. There will be much variation in divergence estimates across proteins; however, confidence in alignment can be evaluated by comparing these estimates to other proteins. In particular, dS will serve as a neutral divergence marker among vertebrates while dN will provide a rough indicator of sequence alignment quality across larger phylogenetic distances. Orthology/Paralogy relationships Using the alignments, it will be informative to extract copy number information for each gene. A number of databases also provide this information (e.g., Ensembl) in their orthology pipelines. Relationships such as 1:1, 1:n, n:n (where n is an integer) are instructive to users interested in gene families and how they evolved between lizards and a reference genome such as chicken.
Predicted transcript sequence analysis Building on an approach used by FlyBase [6], each transcript receives a score based on a single-digit octal notation and the sum of the following categories (to an 8 point maximum): • 1 point if one or more aligned EST sequences aligns to the annotated transcript, • 2 points if an annotated exon intersects a region of aligned protein similarity (of course, similarity to self is excluded), • 4 points if there is any gene prediction that is fully consistent with the annotated transcript, and • 8 points if one or more aligned cDNAs are fully consistent with the annotated transcript.
Experimentally defined transcript sequence and alternative splicing EST or full-length cDNA transcript sequence is highly preferable to predicted annotations and should be used at every opportunity. Suggested parameters are currently as defined above. For alternative splicing, the identification of similar patterns of alternative splicing in the species being compared greatly increases confidence that there is an orthologous relationship. Synteny conservation Minimally, orthology could be recognized by the presence of at least 2 orthologous genes, from Gallus gallus, on either the 5' or 3' flanking sequences and in sequential order. Confidence increases with additional orthologous genes on one flank, or synteny conservation on both flanking regions. Gene expression Following gene duplication events, divergence of regulatory control regions can lead to differentiation in tissue specificity and timing of gene expression in paralogous genes. These regulatory regions are considered part of the gene being compared, but it is not straightforward to assign a score to this divergence. Genes that appear to be orthologous by the measures above can still display strikingly different gene expression, raising the question of whether the regulatory gene functionality has diverged in an opposing fashion to that of the protein coding sequence. This is one of the most difficult comparisons to evaluate, and as more comparative analyses are reported, the AGNC aims to develop proposals regarding how genes should be annotated when sequence and expression suggest contradictory findings about the descent of gene functionality.
Much of the above information can be collated into a single colon-separated string that provides the AGNC with a single metric to evaluate nomenclature, and the user with an instant confidence metric. Since this evolutionary character code (ECC) would change depending on the input data, the metric would simply be linked to the gene as a separate feature. As an example, a hypothetical "gene2" would be annotated with the gene description, gene2:chordates:80,55:1-1:5:3,4:TS, meaning that gene2 has orthology only within chordates with, respectively, 80% and 55% overall protein and nucleotide identity (alternatively, dN and dS can be used), it doesn't possess paralogs within and between species (chicken), it has both gene prediction and EST evidence (an octal score of 5), 3 genes upstream with synteny conservation with the reference species and 4 genes downstream, and tissue-specific expression in a cross-species comparison (e.g., with mouse).
With the adoption of a reliable set of orthologous relationships, downstream functional and comparative annotations and alignments that can be used by the entire community could quickly be generated. As an example, gene ontologies (GO) can be easily transferred after orthologies are assigned. Since the chicken genome is one of the twelve "reference" genomes that the Gene Ontology database is carefully annotating with controlled ontological vocabulary [27], the A. carolinensis genome is in an excellent position to be annotated reliably with associated GO terms.
These data must be quickly disseminated to the community via regularly updated databases. The Anolis community currently has a database that is preparing for the next generation of data sets. lizardbase [28] is the primary community website and anole resource that includes a mapping portal for both geographical and genome-based data. It is critical that such communityserving databases coordinate the effort to provide consensus datasets.

Nomenclature for Anolis gene names and symbols
Analysis of the chicken and zebra finch genomes has demonstrated that while a majority of genes can be assigned clear orthologs, functional genes unique to the avian lineage require additional analysis [29]. With the A. carolinensis genome, the challenge is for gene nomenclature to both clearly point out orthology with other vertebrates and allow for identification of nonavian, reptile-specific genes. The AGNC has reviewed guidelines issued by gene nomenclature organizations from mammalian (Human Gene Nomenclature Committee, HUGO; International Committee on Standardized Gene Nomenclature for Mice), avian reptile (Chicken Gene Nomenclature Committee) [30], amphibian (Xenbase) [31,32], and teleost (ZFIN, Zebrafish Information Network) [33,34]

communities.
A major consideration for gene nomenclature in A. carolinensis is flexibility for comparisons with other amniote genomes. Given that the most frequent comparisons of Anolis genes would likely be with human, mouse, or chicken orthologs, the AGNC proposes using a gene symbol style that would allow the reader to infer the species based on the symbol alone. For a hypothetical gene named "gene2", likely species for cross-comparison are: GENE2, human (Homo sapiens): all capitals, italicized Gene2, mouse (Mus musculus): first letter capitalized, italicized GENE2, chicken (Gallus gallus): all capitals, italicized gene2, Xenopus tropicalis: all lower case, italicized gene2, zebrafish (Danio rerio): all lower case, italicized To make it easier to distinguish a reference to an Anolis gene in comparisons with human, mouse, and avian orthologs, the AGNC proposes a gene symbol style similar to Xenopus tropicalis and zebrafish, i.e., gene2, Anolis carolinensis: all lower case, italicized Further details of these guidelines are presented below.

Gene symbols
• Gene symbols for all Anolis species should be written in lower case only and in italics, e.g., gene2.
• Whenever criteria for orthology have been met (previous Section), the Anolis gene symbol should be comparable to the human gene symbol, e.g., if the human gene symbol is GENE2, then the Anolis gene symbol would be gene2. In situations where the human and mouse symbols differ, the AGNC requests that the investigator contact the AGNC through lizardbase to determine a suitable gene symbol for Anolis.
• Orthologous genes in other Anolis species should have the same gene symbol and name as A. carolinensis. A proposed abbreviation code system for comparisons within the genus covering Anolis species is presented below (see section below; Table 1).
• Gene symbols should only contain ASCII characters (Latin alphabet, Arabic numerals) • Punctuation (dashes, periods, slashes) should not be used unless they are part of a human or mouse gene symbol, e.g., if the human gene symbol is NKX3-1, then the Anolis gene symbol should be nkx3-1.
• Gene names: In other model systems, a unique database of gene symbols is typically maintained by a gene nomenclature committee, but there is more variability for the full gene name. Whenever possible, the human or mouse gene name should be used, but omitting references to homology or disease descriptions, e.g., "delta-like 1", not "delta-like 1 (Drosophila)". Provisional human or mouse gene names, e. g., KIAA# or C#orf, should not be used as the basis for a gene name in Anolis species.
• Novel gene names and symbols: If an orthologous gene cannot be identified in any currently sequenced genome, a novel name may be selected by the investigators. The name should ideally be brief and convey information about the gene expression or function but not include proper or commercial names, e.g., yep1, yolk expressed protein 1. References to molecular weight should be avoided, i.e., do not use p35, 35 kDal protein.
• Gene symbols should not start with an "A" or "Ac" as an abbreviation for Anolis carolinensis, i.e., not acgene2. Gene symbols may start with "a" or "ac" if the human or mouse ortholog starts with these letters, e.g., actb for beta-actin. • Using criteria for orthology described in the previous objective, duplication of the Anolis ortholog of a mammalian gene will be indicated by an "a" or "b" suffix, e.g., gene2a and gene2b. If the mammalian gene symbol already contains a suffix letter, then there would be a second letter added, e.g., gene4aa and gene4ab.

Protein symbols
• Protein symbols should be the same as the gene symbol except written in all upper case without italics, e.g., GENE2.

Nomenclature for Anolis non-coding sequences, including transposons and repetitive elements
The classification and nomenclature of transposable elements presents a particular challenge because of the large diversity of transposons in eukaryotic genomes. Several classification and naming schemes have been proposed but there is currently no consensus on how transposons should be annotated [35,36]. An ideal classification system of transposable elements should reflect the evolutionary relationships among elements [37]. However, as eukaryotic genomes are annotated independently from each other there has been a tendency to name transposon families by numbering them in the order they are discovered, without much consideration of their evolutionary affinities across genomes [38]. Although scientists agree on the major categories of transposable elements (DNA transposons, non-LTR retrotransposons and LTR retrotransposons), there is no consensus on their classification at lower levels (families and subfamilies) and on how to name newly discovered transposons. Thus, the nomenclature of transposons can be considered a work in progress. An International Committee on the Classification of Transposable Elements has been created and is aiming to build a classification that will reflect the structural and evolutionary affinities among elements, yet that will also be relatively easy to use. Until a consensus is reached within the transposable element community, we propose some simple guidelines for the nomenclature of transposable elements in A. carolinensis.
The general principles of the nomenclature follow the recommendations of Kapitonov and Jurka [37], with some minor modifications. Kapitonov and Jurka proposed to name elements by the super-family in which they belong, followed by a unique identifier (generally a number), a structural identifier if necessary, and end with a species identifier. For example, Helitron-1_Acar would be the name of family 1 of autonomous Helitron in A. carolinensis. If a non-autonomous family of helitron has been amplified by Helitron-1_Acar, its name will be Helitron-1N1_Acar, the N indicating its nonautonomous nature. However, the diversity within some super-families is relatively well known, at least in vertebrates, and we propose that the name of elements should reflect their evolutionary affinities below the super-family level. For instance, the hAT super-family contains several well-defined monophyletic lineages (e.g., hobo, Charlie, restless). In those cases where the diversity of the super-family is well characterized, we propose to name elements using the name of the clades. For instance, we propose to use the name hobo-1_Acar instead of hAT-1_Acar for a family that is unambiguously related to other hobo elements.
An additional difficulty in naming transposable elements results from the common occurrence of horizontal transfer. A consequence of horizontal transfer is that identical or very similar elements might be found in distantly related organisms [39][40][41][42]. Novick et al. [41] proposed to use the letter HT to indicate the fact that an element has been horizontally transferred from another species, e.g. hAT-HT1_Acar. However, this solution is not satisfactory as the same elements might carry different names in different organisms because genomes are annotated independently. For instance, the anole hAT-HT2_Acar is different from the hAT2_ML of bats but is identical to the hAT4 in Xenopus tropicalis. In those cases, we believe it is better to not use a numbering scheme but instead to choose a different name for those families that are found in distantly related taxa. A name that reflects at least partially the evolutionary affinities of the elements is preferable. The solution adopted in Thomas et al. [42] to name horizontally transferred helitrons seems satisfactory, e.g., Heligloria.
As mentioned earlier, the classification and nomenclature of transposons is a work in progress that will require a better knowledge of transposable element evolution below the super-family level and across genomes. It is the goal of the committee to regularly improve and update the classification of A. carolinensis elements.

Abbreviations for Anolis species and population groups
Comparative and functional genomics is rapidly progressing from broad-scale comparisons among model systems to fine-scale analyses among populations and closely related species [43][44][45]. Anolis is an ecologically, physiologically, and morphologically diverse genus of over 350 species that has a rich history of comparative studies [4]. While the nomenclature described above establishes guidelines for the model system, A. carolinensis, it is critical that the research community arrive at a common vocabulary to reference data from other Anolis species and among populations. The AGNC proposes the following guidelines with this aim: • All genus and species abbreviations for anoles will begin with the capital letter, 'A', followed by three lowercase italicized letters based approximately on the first letters of the species name, e.g., Anolis sagrei = Asag.
• In comparative analyses abbreviations will be added as a suffix to the proper gene names, e.g., gene2-Asag.
• The three-letter species abbreviation suffix (in lowercase) is generated by the first two letters of the species name and an identifying third letter unique to each species. In cases of redundancy in all of the first three letters of species names, precedence is given to the date of first publication. For the remaining species, the third letter will be replaced with the subsequent letter of the species name that generates a unique code. Examples: A. grahami = Agra since this species was first reported in 1845 [46]; A. gracilipes = Agrc; A. granuliceps = Agrn. A full listing of 378 abbreviations based on our current view of the species content of Anolis is found in Table 2 and posted to various anole community sites listed at the end of this report.
• Once established, modifications to the four letter abbreviations are strongly discouraged in order to maintain clarity, even in cases of renaming or reclassification.
• This system of nomenclature does not address subspecies designations or geographic 'races.' The AGNC is currently accepting community proposals for these designations.

Abbreviations for conserved sequences
A subclass of sequences can be defined by their high degree of conservation across taxonomic levels [47,48].   Nomenclature for these conserved sequences (CSs) poses unique challenges because they lack defining content, such as that comprising transposons and repetitive elements. Additionally, CSs are not always completely conserved and occasional duplicate CSs are scattered throughout the genome. We propose to describe CSs in the Anolis genome using a combination of species code, unique identification number, length, percent conservation with other species, and characterization of species with which they are shared [49]. We recommend that: • CS names begin with the species code, Acar, to identify Anolis carolinensis as the species within which these sequences are described.
• A unique, 1-indexed, arbitrarily assigned number follow the species name.
• Abbreviated length class designations follow the CS number. We define the length classes as follows: (s) short ≤ 99 bp; (m) medium 100-499 bp; or (l) long ≥500 bp). Using this nomenclature, the 1,000th CS identified in the A. carolinensis genome that is 600 bp long having 100% conservation between A. carolinensis and chicken genomes would be named Acar1000l1SMB.

Abbreviations for Anolis genetic markers including microsatellite assays
The A. carolinensis genome contains many types of repetitive elements including mononucleotide tracts, microsatellites, minisatellites, and satellites. Many researchers focus on simple tandem repeats (STRs, also known as short tandem repeats, microsatellites or simple sequence repeats, SSRs). Some STRs have variable numbers of repeats (i.e., variable number tandem repeats, VNTRs). However, variation is often not reported with the genomic sequence and may be inconsistent among populations and species, and knowledge of variation can change through time as more individuals are sampled. Rather than subdividing and explicitly defining the different repeat types or using VNTR status, we provide a simple, unique nomenclature that can be applied to all STRs in any species of Anolis. This nomenclature is linked to a more descriptive, locus-specific annotation available from lizardbase. Additional detail regarding the challenges of explicitly defining various classes of STRs has been described [50].
We propose that Anolis STRs be assigned a name consisting of three fields separated by underscores: 1) the species code described in Part 4 above derived from the organism of origin, 2) the letters 'str' for simple tandem repeat, and 3) a unique, 1-indexed, identification number Using this nomenclature, the 8 th STR identified in the A. carolinensis genome would be coded as Acar_str_8. We will store additional, locus-specific information such as repeat unit, genomic location, and number of repeats in a separate database, linked to each STR using these unique names. The submission of STR markers and assignment of unique identification numbers will be handled through lizardbase by the AGNC or designated member.

Future objectives of the Anolis Gene Nomenclature Committee
The recently published green anole (A. carolinensis) genome [1] provides an example of how a community of researchers with both common and distinct interests can work together to build an enduring resource. This genomics resource now provides an opportunity for the community to advance a greater knowledge of gene function and orthology. As work progresses on Anolis species genomes, new and unforeseen nomenclature issues will certainly arise. The goal of the AGNC is to foster community-based discussion where these problems can be resolved. We have presented guidelines for three immediate objectives for the AGNC but we foresee the need to rapidly address the following objectives: • Nomenclature for populations and treatment of geographic variation • Creating a common nomenclature for genetic markers such as microsatellites and SNPs • Creating a common nomenclature for transposable elements The AGNC welcomes feedback from the community to raise overlooked issues and unforeseen conflicts. The AGNC views these recommendations as an evolving document, and current, archival, and proposed revisions will be posted to the anole community web sites: Anolisgenome [51] Anolis Newsletter [52] Anole Annals Blog [53] Correspondence to any member of the committee is welcomed. We also would like to elicit comments and suggestions from other research communities with unannotated genomes. It would be helpful to be able to develop and share such important resources and experiences together.