The chordata olfactory receptor database

Graphical abstract

Introduction of database

Olfaction is one of the oldest chemosensory systems in chordates, playing crucial roles in their foraging, predator evasion, social communication, mating and parental care (Guo et al., 2023; Li and Liberles, 2015; Liberles, 2014). The initial step of olfaction is the binding and activation of olfactory receptors (ORs) by odorants in a combinatorial way (Malnic et al., 1999). In chordates, ORs are encoded by OR genes (olfrs). Generally, olfrs are expressed on the cilia of olfactory sensory neurons, more and more olfrs have been found to be ectopically expressed in various non-olfactory tissues and some of them are disease-related (Buck and Axel, 1991; Drew, 2022; Massberg and Hatt, 2018). For example, Orecchioni et al. had shown that the octanal receptors Olfr2 in mouse and OR6A2 in human vascular macrophages, upon interaction with toll-like receptor 4, initiate inflammasome activation in the presence of octanal, which in turn leads to the production and secretion of IL-1α and IL-1β proteins (Orecchioni et al., 2022). This response, along with the involvement of other inflammatory cytokines, is likely responsible for the pronounced effects of Olfr2 observed in mouse models of atherosclerosis (Orecchioni et al., 2022). Cheng et al. elucidated the binding mechanism of the insulin peptide insB:9-23 to the OR Olfr109 and demonstrated that a pepducin-based antagonist targeting Olfr109 ameliorated glucose homeostasis in mouse models of diabetes and obesity (Cheng et al., 2022). This suggests that ORs not only form the basis of olfactory function but also hold significant importance in understanding chordate biology and advancing disease treatments.

In chordates, the rapid evolution of olfrs has been suggested to be influenced by the unique living environments of specific species (Nei et al., 2008; Niimura, 2009). This diversity is evident in the varying number of olfrs among species: the African elephant genome encodes around 2,000 functional ORs, compared to about 1,200 in mouse, approximately 400 in humans, and just over 100 in zebrafish (Han et al., 2022; Niimura et al., 2014). Additionally, most species’ genomes include a high-proportion of OR pseudogenes. In humans, for instance, the genome contains over 400 OR pseudogenes, accounting for about 50% of the total olfrs (Han et al., 2022). These factors collectively increase the complexity of annotating olfrs based on sequence homology. Furthermore, the quality of genome sequencing significantly impacts the accuracy of olfr annotation. Poor sequencing and assembly quality, such as sequencing errors, incomplete sequences, or misjoins, can lead to erroneous annotations of olfrs. In fact, only a few species have had their olfrs thoroughly annotated, with data primarily recorded in six chordate OR-related databases: ORDB (Marenco et al., 2013), OlfactionDB (Modena et al., 2011), ODORactor (Liu et al., 2011), HORDE (Olender et al., 2013), OlfactionBase (Sharma et al., 2022), and M2OR (Lalis et al., 2024). However, with advancements in sequencing technology and reductions in sequencing costs, approximately 3,000 chordate species have now had their whole genomes sequenced, many of which are of high-quality (Satam et al., 2023; Sayers et al., 2023). However, the OR data for most species have not been effectively annotated or systematically integrated.

To address the lack of effective annotation and systematic integration of OR data across numerous species, we developed Genome2OR. This tool, which is based on a hidden Markov model (HMM) of DNA sequences, provides a rapid and sensitive solution for annotating olfrs from genomes (Han et al., 2022). Recently, we updated the Genome2OR to simplify the annotation process and employed it to perform exhaustive annotations of thousands of chordate genomes, yielding over one million olfrs. Leveraging these annotations, along with a multitude of derived datasets and additional external data we curated, we created the chordata olfactory receptor database (CORD, website of CORD). CORD is an online resource that offers a rich and comprehensive collection of OR data. With its user-friendly interface and an intelligent, efficient underlying database system, users can easily access and retrieve the data they need. We believe that CORD will be a valuable resource for the research community in the OR field, and it will also engage the broader public interested in the science of olfaction.

Overview of database

CORD offers a comprehensive and high-quality resource for chordate ORs, covering a wide range of data from species diversity to the molecular level. It features a user-friendly interface and a robust underlying database system, ensuring that users can easily access, analyze, and retrieve the information they need.

CORD integrates data from 2,776 species across seven major evolutionary clades: lancelets, jawless fish, jawed fish, amphibians, reptiles, birds, and mammals, and includes 1,176,818 olfrs (Table 1). These genes comprise 663,380 functional olfrs and 513,438 OR pseudogenes, showcasing the depth and breadth of the data (Table 1). Furthermore, CORD features data on odorants, OR–odorants pairs and high-precision protein models (Ahdritz et al., 2024), snake diagrams (Isberg et al., 2016), BLAST (Camacho et al., 2009), profiles, WebLogos (Crooks et al., 2004), and sequence similarity networks, providing robust support for structural and functional studies of ORs.

The user interface of CORD is designed to be intuitive, offering a user-friendly querying experience. Interactive data visualizations and advanced search options enhance the efficiency of data retrieval and analysis. Notably, CORD emphasizes data localization and cross-referencing, allowing users to download the entire dataset while offering extensive links to both internal and external databases. This feature facilitates in-depth exploration and cross-database applications, providing users with comprehensive research capabilities.

In summary, CORD significantly enhances the accuracy and convenience of data retrieval and analysis by expanding the coverage of OR data in chordates and optimizing data visualization and user interface design. This provides robust data support for research in olfactory science and related biomedical fields.

CORD interface and navigation

CORD is an open-access web server offering free data browsing and download services. The platform features nine functional menus: Home, BLAST, WebLogo, Profile, Network, Analysis, Tool, Statistics, and Help (Fig. 1). Additionally, it includes two key information display pages: the search results page and the detail page. With its intuitive interface and robust search functionality, users can easily access, analyze, and retrieve the OR data they need (Fig. 1).

Home menu: intuitive navigation and advanced search features

The Home menu provides intuitive access to information through the search box and navigation cards, enabling users to quickly retrieve data (Fig. 2A). The search box features a built-in fuzzy matching algorithm that enables rapid searches across 18 fields. For users requiring more precise queries, the “Advanced” button next to the search box offers access to advanced search options, allowing for exact or fuzzy searches based on logical combinations in specific fields, thus meeting detailed query needs. The navigation cards serve as shortcuts to popular sections of the database, simplifying user navigation. The design of the search box and navigation cards fully embodies CORD’s user-centric design philosophy, ensuring that both new users and experienced researchers can efficiently browse the database.

Search result page: personalized search result customization

In CORD, user-initiated searches are directed to a well-designed search results page. This page intuitively displays search results in a table format, featuring up to 35 customizable attribute fields such as “Organism,” “Gene Name,” and “UniProt ID,” providing a multidimensional data view. Users can sort, filter, and customize the display fields and their order according to their personal needs, achieving a personalized data interface. The filtering panel on the page supports various filtering criteria, enhancing search accuracy. Additionally, users can easily download search results in different file formats, facilitating data export and analysis.

Detail page: comprehensive OR information display

The detail page is the core component of CORD for presenting OR information, with each OR having its dedicated page. The page comprises eight modules that comprehensively introduce receptor information. The “basic information” module provides an overview of essential OR data; the “Sequences” module displays protein and DNA sequences in a color-coded format and includes links to the BLAST menu; the “snake diagram” module highlights the main structural features of the OR, using color coding for enhanced recognition (Pandy-Szekeres et al., 2023); the “structure” module offers structural prediction results and model quality assessments, including predicted local distance difference test, predicted aligned error, and multiple sequence alignment (Varadi et al., 2024); the “taxonomy” module presents basic species information and OR distribution; the “sequence similarity network” module displays the sequence similarity network among ORs within a species; the “similar proteins” module lists similar proteins based on sequence identity; and for ORs with OR–odorant pairs data, the detail page also features a “ligands” module, which illustrates the relationships between ORs and ligands. The detail page not only displays comprehensive data but also enhances interactivity and data export functions, enabling researchers to easily access, analyze, and share information.

BLAST menu: homology sequence search

The BLAST menu integrates homology sequence search functionality based on the NCBI blast + toolkit (v2.15.0) (Camacho et al., 2009) (Fig. 2B). Users can submit query sequences to search for similar protein or DNA sequences within the CORD, utilizing the BLAST menu’s powerful search capabilities. This page provides flexibility, enabling users to adjust key parameters of the BLAST algorithm, such as E-value, word size, and scoring matrix, to optimize search results according to their research needs (Camacho et al., 2009). Additionally, the download and export functions of the BLAST page greatly streamline data retrieval and subsequent analysis.

WebLogo menu: sequence conservation analysis

The WebLogo menu leverages the WebLogo (v3.7.8) tool to provide users with an intuitive interface for visualizing sequence conservation patterns across different OR families or species evolutionary lineages (Crooks et al., 2004). The page displays sequence conservation through stacked letters, with letter sizes representing residue frequency, where larger letters indicate higher conservation (Fig. 2C). Users can customize the view of conservation patterns based on specific structural regions, OR families and species evolutionary lineages. The page also supports the insertion of multiple WebLogos, enabling visual comparisons of conservation patterns across different OR families or species evolutionary lineages.

Profile menu: residue distribution revealing

The Profile menu and WebLogo menu collectively undertake the task of analyzing conservation patterns within OR families (Fig. 2D). Although these two functional modules differ in their modes of representation, they both aim to uncover common features and patterns within OR families. The profile menu provides specific numerical values for residue distribution at each position, offering more detailed information, albeit in a less intuitive format compared to WebLogo (Fig. 2C).

Network menu: OR network explorer

The Network menu offers an interactive network visualization tool for exploring the complex relationships between ORs (Fig. 2E). Users can select specific OR families or species using the control panel and adjust the style and color of data points. This tool simplifies the browsing and analysis of OR networks, thus aiding users in gaining a deeper understanding.

Analysis menu: mapping the distribution of ORs

The tools in the Analysis menu assist researchers in exploring the distribution of ORs (Fig. 2F). By selecting species evolutionary lineages using the toggle panel, researchers can use histograms and heatmaps to visualize OR distribution. Histograms provide detailed distributions across different species or receptor families, while heatmaps offer a quick overview of overall distribution patterns.

Tool menu: accessing and utilizing the annotation tool

The tool menu aids users in comprehensively understanding and utilizing Genome2OR, an OR genome annotation tool we developed (Han et al., 2022). This menu provides workflow diagrams, user guides, algorithm processes, and dependency lists, ensuring that users can easily download, install, and efficiently use the tool. These resources help users grasp the tool’s algorithmic details and effectively apply it in their genomic annotation and analysis research.

Statistics menu: database data overview

The Statistics menu provides users with comprehensive statistics and visual representations of the database data (Fig. 2G). Through pie charts, histograms, and tables, users can quickly grasp the basic data overview of CORD. Additionally, this page showcases the data growth trajectory from previous versions to the current version, offering a clear comparison of data volume and highlighting the continuous improvement in the content and data quality of CORD.

Help menu: comprehensive support hub

The Help menu is a comprehensive support center designed to help users thoroughly comprehend and efficiently use the database. It offers sections such as guides, data introduction, FAQs, contact us, cite us, acknowledgments, and revision history. These resources enable users to fully leverage the functionalities and resources of CORD.

Comparison with previous work

CORD significantly advances the field of OR databases, improving data richness, species coverage, and technical architecture. Compared to our previous work, this update includes over 60% more data. Additionally, CORD introduces new data on odorants, OR–odorant pairs, high-precision protein models, and conceptual translation sequences of OR pseudogenes.

In a comparative analysis with other existing chordate OR databases, CORD demonstrates significant advantages across key metrics (Table 1). Compared to ORDB, the number of functional ORs in CORD has increased more than 35-fold, and species coverage has expanded from 70 to 2,776 species, representing nearly a 40-fold increase (Table 1). Regarding pseudogenes, although ODORactor and HORDE databases provide related data, CORD exhibits a clear advantage in the inclusion of OR pseudogenes. GPCRdb is undeniably a leading resource in the G protein-coupled receptor field, and we are enthusiastic users of this database. The comprehensive content and intuitive user interface of GPCRdb have been a significant source of inspiration for us, driving the development of CORD. While GPCRdb provides detailed and professional data on most human G protein-coupled receptors, it largely overlooks ORs, offering only limited information on a few dozen human ORs (Table 1). Additionally, CORD records 3,118 OR–odorant pairs and 23,690 odorants, the highest among all databases, providing a rich chemical and molecular foundation for the functional study of ORs.

Database construction

Update on genomic data acquisition for CORD

CORD continues the automated genomic data download process that was initiated in our previous work, ensuring a seamless continuation of data acquisition and integration (Han et al., 2022). This update focuses on the collection of incremental data, specifically all chordate genomic data published through the NCBI Assembly database from January 2021 to April 2023 (Taxon ID: 7711). Utilizing the NCBI Datasets tool (v15.10.0), we have newly downloaded 1,048 chordate genomic datasets, bringing the total number of species in CORD to 2,776 (Sayers et al., 2023).

Updates and optimizations of Genome2OR

We have implemented significant updates and enhancements to the Genome2OR annotation tool, aimed at elevating the efficiency and automation of genome annotation. The new version of Genome2OR simplifies the annotation process by rewriting sections of the code, merging tasks that previously required three separate steps into a single step (Han et al., 2022). Furthermore, the new version incorporates the HMM Profile into the code, allowing users to swiftly invoke it by specifying parameters, thereby streamlining the annotation process. In addition, to accommodate users’ needs for customized HMM Profiles, we have retained an input interface that enables users to upload their customized HMM Profile files for personalized annotation. This improvement not only boosts the usability of Genome2OR but also provides users with greater flexibility.

Conceptual translation of pseudogenes using FASTA suite

We employed the fasty tool from the FASTA suite to perform the conceptual translation of pseudogenes (Pearson, 2016). This process relies on a carefully curated and deduplicated target database that integrates OR protein sequences from nearly 3,000 species, ensuring its uniqueness and comprehensiveness. By matching the query DNA sequences against the protein database, we were able to identify the closest reference sequences, significantly enhancing the accuracy of the conceptual translation. Thanks to the richness and diversity of our database, we have successfully achieved conceptual translation for 99% of the pseudogenes in CORD.

Implementation of protein sequence clustering for CORD

We implemented protein sequence clustering for CORD, using the same method as the Uniclust database, which is widely recognized in the industry for its efficiency and sensitivity (Jumper et al., 2021; Mirdita et al., 2017). We used the uniclust-pipeline (github.com/soedinglab/uniclust-pipeline) for protein sequence clustering, with core commands from the MMseqs2 (Mirdita et al., 2021) and HHsuite3 (Steinegger et al., 2019) suites. To accommodate the specific needs of CORD, we made necessary modifications to the original process code, especially in terms of clustering sensitivity and sequence identity threshold parameters. We chose more sensitive clustering parameters, for example, adjusting the ‘-s’ parameter in the mmseqs prefilter command from 1 to 6 to enhance the sensitivity of prefiltering, although this requires more computational resources. In addition, based on the classification knowledge of the OR family, we provided customized settings for the sequence consistency threshold parameters (Olender et al., 2020). By applying sequence consistency thresholds of 30%, 40%, 50%, 60%, 80%, and 90%, we successfully generated a series of clustering datasets from CORDclust30 to CORDclust90. It is worth noting that our clustering analysis not only covers functional OR protein sequences but also includes sequences obtained from the conceptual translation of OR pseudogenes.

High-precision protein structure prediction of ORs with OpenFold

We employed the OpenFold tool to perform high-precision protein structure prediction on the 663,380 functional ORs present in our database (Ahdritz et al., 2024). Our workflow adhered to a stringent set of bioinformatics standards, ensuring the accuracy and reliability of the prediction process. Initially, we identified candidate sequences with sequence homology to the target ORs by conducting a sequence search on the CORDclust50 database using the HMMsearch algorithm (Eddy, 2011). Subsequently, these candidate sequences were realigned in a MSA the MAFFT software to ensure the high-accuracy of the alignment results (Katoh and Standley, 2013; Nakamura et al., 2018). Following this, we used the HHalign tool in the HHsuite3 toolkit to convert the alignment results from FASTA format to A3M format, meeting the input requirements of OpenFold (Steinegger et al., 2019). Ultimately, these standardized MSAs were input into OpenFold, where our advanced deep learning model facilitated protein structure prediction.

OR community detection and network embedding in CORD

Initially, we performed MSAs of 663,380 functional OR protein sequences from the CORD and calculated their similarities, thereby constructing a global OR sequence similarity network (Han et al., 2022). Subsequently, we employed the parallel Louvain method algorithm from the NetworKit toolkit for community detection within this network (Angriman et al., 2022; Christian and Henning, 2016). Additionally, we utilized the LargeVis algorithm to project the network, reducing its 663,380 dimensions to 3 dimensions to enable visualization and analysis (Tang et al., 2016). Ultimately, we partitioned the functional ORs in CORD into 20 distinct communities.

Data collection in CORD

To build the CORD database, we collected various types of data from multiple reliable sources, including odorants and their properties, OR–odorant pairs data, species lineage data, and olfr names. Notably, we obtained 76 attribute data of 23,690 odorants from The Good Scents Company (TGSC) and the PubChem database (Creek; Kim et al., 2023). Simultaneously, by integrating and analyzing seven databases—ORDB (Marenco et al., 2013), OlfactionBase (Sharma et al., 2022), M2OR (Lalis et al., 2024), OlfactionDB (Modena et al., 2011), ODORector (Liu et al., 2011), InAct (Del et al., 2022), and Glass (Chan et al., 2015), we obtained 3,118 OR–odorant pairs data entries after redundancy removal. To ensure the accuracy of species lineage data, we employed the TaxonKit tool for matching and downloading (Shen and Ren, 2021). Additionally, we obtained data such as olfr names from UniProt, adding important information dimensions to the CORD (UniProt Consortium, 2023).

Web implementation of CORD

The web implementation of CORD adopts a front-end and back-end separation development concept, which allows the front-end user interface and back-end server logic to be developed and deployed independently. The user interface design of CORD is inspired by the UniProt database, aiming to provide a clear, intuitive, and professional navigation layout to ensure that users can efficiently access and analyze data (UniProt Consortium, 2023). The front-end uses the Vue.js (v3.3) framework and Element Plus (v2.3) component library to implement a responsive and componentized user interface (Kyriakidis et al., 2016). The back-end chooses Nginx (v1.21) as the web server and reverse proxy, combined with the module packaging optimization of Webpack (v5.88), ensuring the stability, security, and fast page loading time of the application (Reese, 2008). At the database level, CORD uses the NebulaGraph (v3.6) graph database, optimizing graph data storage and query performance, especially suitable for handling complex bioinformatics networks. The integration of Elasticsearch (v8.13) provides powerful full-text search capabilities, while MySQL (v8.0) ensures the consistency and integrity of transactional data (Gormley and Tong, 2015; Workbench, 2019). In terms of data visualization, by integrating AntV G2 (v4.2) and ECharts (v5.4) graphics libraries, CORD provides a wealth of data visualization options, not only enhancing the readability of data but also supporting researchers in deeply understanding complex bioinformatics data (Li et al., 2018).

Future development

The CORD is dedicated to continuously integrating various data related to chordate ORs, aiming to serve scientists and the public interested in olfactory research. With the release of the first experimentally determined chordate OR structure (Billesbolle et al., 2023) and several consensus chordate OR structures (Choi et al., 2023), our understanding of OR structures has entered a new stage. Clearly, more and more OR structures are on their way. In the future, we plan to integrate experimental structural information of ORs into the CORD. Meanwhile, we plan to use new molecular simulation technologies such as AlphaFold3 to build odor–OR complex structures and put them into our database (Abramson et al., 2024). Additionally, we will continue updating new ORs from newly sequenced chordate genomes, introduce a more systematic classification system for ORs based on DNA phylogeny, and retain multiple previous classification systems to enable scientists to delve into the evolutionary relationships of these two major classes of ORs (Niimura et al., 2014; Olender et al., 2020). Functionally, we look forward to adding a genome browser feature in future versions of CORD, enabling users to analyze olfrs and their neighboring genes on the genome.

Conclusion

CORD is a comprehensive and highly integrated OR database. This update has achieved significant growth in data volume and has made important advances in data types, user interface experience, and backend architecture performance. Through the carefully designed web interface and advanced search algorithms, CORD greatly enhances the efficiency of users in retrieving and analyzing data, providing a powerful tool for OR research.

Our database has achieved a 64% increase in the number of functional ORs, with the total number of functional genes reaching 663,380. We have also expanded species coverage by 64%, bringing the total number of species to 2,776, and added key data types such as odorants, OR–odorants and protein models. These rich data resources, combined with our continuous optimization of the Genome2OR tool, provide a solid foundation for cross-species comparative research. Furthermore, the backend architecture of CORD has been thoroughly restructured, adopting the NebulaGraph database and Elasticsearch full-text search engine, significantly improving search efficiency and intelligence.

We believe that the launch of CORD will benefit researchers and the public in the field of ORs. At the same time, we look forward to the addition of a genome browser feature in future versions, which will further enhance the database’s analytical capabilities and provide researchers with deeper insights.

References

[1] Abramson J, Adler J, Dunger J et al. Accurate structure prediction of biomolecular interactions with AlphaFold 3. Nature 2024;630:493-500.

[2] Ahdritz G, Bouatta N, Floristean C et al. OpenFold: retraining AlphaFold2 yields new insights into its learning mechanisms and capacity for generalization. Nat Methods 2024;21:1514-1524.

[3] Angriman E, van der Grinten A, Hamann M et al. Algorithms for large-scale network analysis and the NetworKit toolkit. In: Bast H, Korzen C, Meyer U, Penschuck M (eds.), Algorithms for Big Data: DFG Priority Program 1736. Cham: Springer Nature Switzerland, 2022, 3-20.

[4] Billesbolle CB, de March CA, van der Velden W et al. Structural basis of odorant recognition by a human odorant receptor. Nature 2023;615:742-749.

[5] Buck L, Axel R. A novel multigene family may encode odorant receptors: a molecular basis for odor recognition. Cell 1991;65:175-187.

[6] Camacho C, Coulouris G, Avagyan V et al. BLAST+: architecture and applications. BMC Bioinformatics 2009;10:421.

[7] Chan WK, Zhang H, Yang J et al. GLASS: a comprehensive database for experimentally validated GPCR-ligand associations. Bioinformatics 2015;31:3035-3042.

[8] Cheng J, Yang Z, Ge XY et al. Autonomous sensing of the insulin peptide by an olfactory G protein-coupled receptor modulates glucose metabolism. Cell Metab 2022;34:240-255.e10.

[9] Choi C, Bae J, Kim S et al. Understanding the molecular mechanisms of odorant binding and activation of the human OR52 family. Nat Commun 2023;14:8105.

[10] Christian LS, Henning M. Engineering parallel algorithms for community detection in massive networks. IEEE Trans Parallel Distrib Syst 2016;27:171-184.

[11] Creek O. The Good Scents Company Information System. 2021.

[12] Crooks GE, Hon G, Chandonia JM et al. WebLogo: a sequence logo generator. Genome Res 2004;14:1188-1190.

[13] Del TN, Shrivastava A, Ragueneau E et al. The IntAct database: efficient access to fine-grained molecular interaction data. Nucleic Acids Res 2022;50:D648-D653.

[14] Drew L. Olfactory receptors are not unique to the nose. Nature 2022;606:S14-S17.

[15] Eddy SR. Accelerated profile HMM searches. PLoS Comput Biol 2011;7:e1002195.

[16] Gormley C, Tong Z. Elasticsearch. The Definitive Guide: A Distributed Real-time Search and Analytics Engine. Sebastopol, CA: O’Reilly Media, Inc, 2015.

[17] Guo L, Cheng J, Lian S et al. Structural basis of amine odorant perception by a mammal olfactory receptor. Nature 2023;618:193-200.

[18] Han W, Wu Y, Zeng L et al. Building the Chordata Olfactory Receptor Database using more than 400,000 receptors annotated by Genome2OR. Sci China Life Sci 2022;65:2539-2551.

[19] Isberg V, Mordalski S, Munk C et al. GPCRdb: an information system for G protein-coupled receptors. Nucleic Acids Res 2016;44:D356-D364.

[20] Jumper J, Evans R, Pritzel A et al. Highly accurate protein structure prediction with AlphaFold. Nature 2021;596:583-589.

[21] Katoh K, Standley DM. MAFFT multiple sequence alignment software version 7: improvements in performance and usability. Mol Biol Evol 2013;30:772-780.

[22] Kim S, Chen J, Cheng T et al. PubChem 2023 update. Nucleic Acids Res 2023;51:D1373-D1380.

[23] Kyriakidis A, Maniatis K, You E. The Majesty of Vue.js. Birmingham, UK: Packt Publishing Birmingham, 2016.

[24] Lalis M, Hladis M, Khalil SA et al. M2OR: a database of olfactory receptor-odorant pairs for understanding the molecular mechanisms of olfaction. Nucleic Acids Res 2024;52:D1370-D1379.

[25] Li Q, Liberles SD. Aversion and attraction through olfaction. Curr Biol 2015;25:R120-R129.

[26] Li D, Mei H, Shen Y et al. ECharts: a declarative framework for rapid construction of web-based visualization. Visual Informatics 2018;2:136-146.

[27] Liberles SD. Mammalian pheromones. Annu Rev Physiol 2014;76:151-175.

[28] Liu X, Su X, Wang F et al. ODORactor: a web server for deciphering olfactory coding. Bioinformatics 2011;27:2302-2303.

[29] Malnic B, Hirono J, Sato T et al. Combinatorial receptor codes for odors. Cell 1999;96:713-723.

[30] Marenco LN, Bahl G, Hyland L et al. Databases in SenseLab for the genomics, proteomics, and function of olfactory receptors. Methods Mol Biol 2013;1003:3-22.

[31] Massberg D, Hatt H. Human olfactory receptors: novel cellular functions outside of the nose. Physiol Rev 2018;98:1739-1763.

[32] Mirdita M, von den Driesch L, Galiez C et al. Uniclust databases of clustered and deeply annotated protein sequences and alignments. Nucleic Acids Res 2017;45:D170-D176.

[33] Mirdita M, Steinegger M, Breitwieser F et al. Fast and sensitive taxonomic assignment to metagenomic contigs. Bioinformatics 2021;37:3029-3031.

[34] Modena D, Trentini M, Corsini M et al. OlfactionDB: a database of olfactory receptors and their ligands. Adv Life Sci 2011;1:1-5.

[35] Nakamura T, Yamada KD, Tomii K et al. Parallelization of MAFFT for large-scale multiple sequence alignments. Bioinformatics 2018;34:2490-2492.

[36] Nei M, Niimura Y, Nozawa M. The evolution of animal chemosensory receptor gene repertoires: roles of chance and necessity. Nat Rev Genet 2008;9:951-963.

[37] Niimura Y. Evolutionary dynamics of olfactory receptor genes in chordates: interaction between environments and genomic contents. Hum Genomics 2009;4:107-118.

[38] Niimura Y, Matsui A, Touhara K. Extreme expansion of the olfactory receptor gene repertoire in African elephants and evolutionary dynamics of orthologous gene groups in 13 placental mammals. Genome Res 2014;24:1485-1496.

[39] Olender T, Nativ N, Lancet D. HORDE: comprehensive resource for olfactory receptor genomics. Methods Mol Biol 2013;1003:23-38.

[40] Olender T, Jones T, Bruford E et al. A unified nomenclature for vertebrate olfactory receptors. BMC Evol Biol 2020;20:42.

[41] Orecchioni M, Kobiyama K, Winkels H et al. Olfactory receptor 2 in vascular macrophages drives atherosclerosis by NLRP3-dependent IL-1 production. Science 2022;375:214-221.

[42] Pandy-Szekeres G, Caroli J, Mamyrbekov A et al. GPCRdb in 2023: state-specific structure models using AlphaFold2 and new ligand resources. Nucleic Acids Res 2023;51:D395-D402.

[43] Pearson WR. Finding protein and nucleotide similarities with FASTA. Curr Protoc Bioinform 2016;4:3-9.

[44] Reese W. Nginx: the high-performance web server and reverse proxy. Linux J 2008;2008:2.

[45] Satam H, Joshi K, Mangrolia U et al. Next-generation sequencing technology: current trends and advancements. Biology-Basel 2023;12:997.

[46] Sayers EW, Bolton EE, Brister JR et al. Database resources of the national center for biotechnology information in 2023. Nucleic Acids Res 2023;51:D29-D38.

[47] Sharma A, Saha BK, Kumar R et al. OlfactionBase: a repository to explore odors, odorants, olfactory receptors and odorant-receptor interactions. Nucleic Acids Res 2022;50:D678-D686.

[48] Shen W, Ren H. TaxonKit: a practical and efficient NCBI taxonomy toolkit. J Genet Genomics 2021;48:844-850.

[49] Steinegger M, Meier M, Mirdita M et al. HH-suite3 for fast remote homology detection and deep protein annotation. BMC Bioinformatics 2019;20:473.

[50] Tang J, Liu J, Zhang M et al. Visualizing large-scale and high-dimensional data. In: Bourdeau J (ed.), Proceedings of the 25th International Conference Companion on World Wide Web. Republic and Canton of Geneva: CHE, 2016, 287-297.

[51] UniProt Consortium. UniProt: the universal protein knowledgebase in 2023. Nucleic Acids Res 2023;51:D523-D531.

[52] Varadi M, Bertoni D, Magana P et al. AlphaFold protein structure database in 2024: providing structure coverage for over 214 million protein sequences. Nucleic Acids Res 2024;52:D368-D375.

[53] Workbench M. MySQL Workbench 8.0. USA: Oracle Corporation, 2019.

RIGHTS & PERMISSIONS

The Author(s) 2024. Published by Oxford University Press on behalf of Higher Education Press.