To examine changes in opsin expression across Pteriomorphia, we identified species pairs with both publicly available annotated genomes and RNA-seq data collected at three developmental stages (i.e., trochophore, veliger, pediveliger) and from adult tissues. All RNA-seq data needed to be (1) based on Illumina paired-end sequencing with (2) relatively high and similar sequence depth across studies. Seven species from four families met our criteria: mussels Mytilus edulis and M. coruscus (Mytilidae); oysters Crassostrea gigas and Cr. angulata (Ostreidae); the pearl oyster Pinctada fucata (Margaritidae); and scallops Chlamys farreri and Pecten maximus (Pectinidae; Supplementary Table S1). For some species (e.g., Cr. gigas, Pi. fucata, and Pe. maximus), a single study did not include both larval and adult tissues, so a second study was obtained for the larval—adult comparison. Only data from control treatments were used for our analyses. Biological replicates were available for all tissue types across focal species with the exception of M. edulis (larval stages), M. coruscus (mantle), Pi. fucata (all tissue), and Ch. ferreri (larval stages; Supplementary Table S1). All transcriptomic annotated data was retrieved from the NCBI Sequence Read Archive (SRA; Supplementary Table S1), except the Pi. fucata data, which was downloaded from Takeuchi et al. (2012). We used the sratoolkit v3.0.0 (Heldenbrand et al., 2017) to download the RNA-seq datasets from the NCBI SRA database and fastp v0.23.2 (Chen et al., 2018) was used to ensure quality control by eliminating low-quality reads and adapters from the downloaded FASTQ files.

McElroy et al. (2023) demonstrated extensive lineage-specific opsin expansions in Mollusca, with bivalves having highly variable opsin content. To place opsins from our focal bivalve species into proper phylogenetic context, we collected opsin sequences from 23 high-quality pteriomorphian genome assemblies (Supplementary Table S2, species bolded used for expression analysis). Building on the results of McElroy et al. (2023), we used the gene-family assembly pipeline BITACORA v1.3 (Vizueta et al., 2020), incorporating Gene Model Mapper (GeMoMa; Keilwagen et al., 2016, 2018), to de novo predict genes based on alignments of the same high-quality molluscan opsin protein sets. We ran the predicted genes through the Phylogenetically Informed Annotation Pipeline (PIA; Speiser et al., 2014; modified version downloaded)¹ to identify opsins based on the Light Interacting Toolkit (LIT_1.1; r_opsin_20_rtrans.fas). We also aligned the high-quality curated reference opsin protein sequences (McElroy et al., 2023) from close relatives to additional genome assemblies (e.g., Crassostrea gigas for Cr. angulata) using miniprot v0.7 (Li, 2023) and then extracted transcripts and protein sequences for each gene using gffread v0.12.7 (Pertea and Pertea, 2020). We inspected alignments in MEGA X (Kumar et al., 2018) to combine results from these two approaches and aid in manually completing gene models (here, a complete GPCR Class A 7tm_1 domain) along with tblastn (NCBI BLAST+ v2.13.0; Camacho et al., 2009) hits in their respective genomes. All candidate opsins had a retinal-binding lysine residue homologous to K296 in bovine rhodopsin.

Recently, a closely related 7-transmembrane GPCR was identified in mollusks, annelids, and nemerteans as being more closely related to opsins than melatonin receptors and named “pseudopsins” (De Vivo et al., 2023). For outgroup sequences, we used these “pseudopsins,” along with melatonin receptors, and the opsin-like GPCRs from the placozoan Trichoplax adhaerens referred to as “placopsins” (XP_002113363.1, XP_002112437.1). To collect pseudospin and melatonin receptor sequences from additional species, we similarly mapped protein sequences from close relatives to the genome assemblies (e.g., Cr. gigas for Cr. angulata).

We then used mafft v7.481 (Kuraku et al., 2013) to align the opsin and outgroup amino acid sequences using the EINSI strategy (−maxiterate 1,000 –genafpair), then generated a phylogenetic tree using maximum likelihood analysis with IQ-TREE2 v2.1.3 (Minh et al., 2020) using the protein substitution model JTT + F + R9, and 1,000 ultrafast bootstrap for node support. For the purposes of visualization, we pruned the resulting tree using the R package ape v5.7.1 (Paradis et al., 2004) that only the opsin sequences from the seven species analyzed here for gene expression are present in the topology.

The opsin literature has a long list of synonymies for opsin types. Here, we use common names and the short-hand synonyms that often indicate that opsin’s G-protein signaling pathway: r-opsin = G_q-opsin, which includes the arthropod and cephalopod visual opsins and the vertebrate melanopsin; xenopsin = G_x-opsin, an opsin type found in lophototrochozoans; G_o-opsin; neuropsin = Opn5; retinochrome = RTC, and peropsin.

To make orthologous gene comparisons among species and to distinguish genes resulting from paralogous duplication, we developed a nomenclature based on the phylogenetic topology of pteriomorphian opsins. Our nomenclature only applies to this study, as adding additional opsin sequences to a phylogenetic analysis could alter the placement of gene duplications that we identified. However, we think that future attempts at a comprehensive opsin nomenclature should be grounded in phylogenetics. Briefly, the first three letters of a gene name are determined by the first capital letter of the genus and the first two letters in lowercase of the species name (e.g., “Med” for Mytilus edulis). A period separates the abbreviated Latin binomial from the alphanumeric code identifying the opsin type (homolog), such as “xenopsin” (e.g., “opnGx”). The next part of the name is a single letter capitalized indicating the opsin clade membership if the opsin type is divided into multiple clades, for example, clades A versus B in xenopsin (e.g., “opnGx.B”). If there is a paralogous gene duplication, it is shown as an Arabic numeral with the clade letter (e.g., “opnGx.B1”). A period separates the clade membership with estimated time of when the paralogous duplication occurred. “MY” specifies a duplication along the “Mytilidae” lineage (e.g., “opnGx.B1.MY”).

Typically, bivalve larvae are sampled by hundreds or thousands of individuals per time point. Many of the studies used here had multiple pooled samples at the same time point or had two collection times within a single developmental stage, for example, 17 and 21 days post-fertilization (dpf) across the pediveliger stage. In these situations, we did a single mapping process with multiple samples and then averaged these data to get a single transcripts-per-million (TPM) value representing that developmental stage (“pediveliger”; Supplementary Table S1; e.g., Crassostrea angulata). We applied the same approach when there were multiple RNAseq data for adult tissues (Supplementary Table S1; e.g., Mytilus edulis). Another caveat with the data is that the length developmental stages can vary among species (hours to days) or within a species when influenced by environmental inputs like temperature (reviewed in Cragg, 2016). Thus, there may be changes in gene expression during a prolonged stage that were not captured when examining a single collection time point.

We combined nucleotide sequences of curated opsins for each species with their publicly available genome annotations, removing any redundancies created by the opsin sequence addition. We then used Salmon v1.9.0 (Patro et al., 2017) for pseudo-alignment-based quantification of each SRA dataset (Supplementary Table S1) and collected the transcripts-per-million (TPM) values for downstream comparisons. To account for possible noise, we then categorized opsins as expressed (present) in each sample if the TPM value was above the 10th percentile of values from each study (Supplementary Table S3).

Finally, we selected four well-established housekeeping genes to compare with opsin expression: actin (ACTB), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), succinate dehydrogenase (SDHA), and polyubiquitin-C (UBC; Silver et al., 2008; Huan et al., 2016). All the housekeeping genes were extracted from the seven focal species and tblastn (NCBI BLAST+ v2.13.0; Camacho et al., 2009) was used to find the hits of these protein sequences in their respective genomes. All four housekeeping genes were recovered except UBC in Crassostrea angulate, GAPDH from Mytilus coruscus, and SDHA and UBC in Pinctada fucata. Finally, the transcripts-per-million (TPM) values for these housekeeping genes were extracted from the same SRA datasets from which we determined opsin expression (Salmon v1.9.0; Patro et al., 2017).

A ML tree was generated to place opsins from target pteriomorphian species into the opsin types identified in McElroy et al. (2023). Broadly, we recapitulated previous relationships of molluscan opsin groups (Figure 1A; Supplementary Data 1) and evidenced lineage-level duplications of many opsins in this group of bivalves recently demonstrated in McElroy et al. (2023). This phylogeny provided a framework to identify putative paralogs and to estimate in which taxonomic lineages gene duplications or losses may have occurred. Ultimately, this phylogenetic framework allowed for more accurate comparations of gene expression among species.

Across Mollusca, genomes contain opsins from as many as seven distinct clades, but lack c-opsins and cnidopsins (McElroy et al., 2023). We phylogenetically classified 447 opsins sequences mined from 23 pteriomorphian genomes, including 119 from our seven focal species into the seven types of opsins: canonical or noncanonical G_q-opsins (= r-opsins), neuropsin, G_o-opsin, xenopsin (= G_x-opsin), peropsin, and retinochrome (Figure 1A). All identified opsin sequences possessed a retinal-binding lysine residue homologous to K296 in the bovine rhodopsin positional naming system indicating the capacity to form a photopigment. Gene duplications were observed in xenopsin (Figure 1B), both canonical and non-canonical G_q-opsins (Figure 1C), G_o-opsin and neuropsin (Figure 1D). Some of these duplication events appear to be deep within the Pteriomorphia before the split of the four families examined (e.g., xenopsin clade B), while others are at the family-level, such as r-opsin paralogs in Pectinidae and Mytilidae (Figure 1C). Multiple rounds of gene duplication were estimated to occur in the xenopsin clade B for Mytilus and pectinid species, while Crassostrea has a single duplication event and Pi. fucata has only one gene from that xenopsin clade (Figure 1B). The xenopsin clade A appears to be less expansion-prone than xenopsin clade B, but gene duplication is evident in Pi. fucata, Crassostrea, and Mytilus (Figure 1B). Duplications of neuropsin were only observed in Mytilus, which has a lineage-level expansion resulting in four opsins vs. one in the other species examined here. As in McElroy et al. (2023), a single retinochrome was found in these pteriomorphian genomes (Figure 1D), and only Crassostrea and Pi. fucata had a copy of peropsin (Figure 1D).

Some general patterns emerged from the opsin expression data (Figure 2; Supplementary Data 2). First, all opsin types are expressed across the three larval stages, trochophore, veliger, and pediveliger, in most species. Second, few opsin types are expressed in the adult mantle tissue. This was observed across all seven species. At one extreme, neuropsin (opn5) is below the threshold of expression in the adult mantle for all focal taxa and treated as “off” (Figure 2). Of the three species that have a peropsin gene (Cr. angulata, Cr. gigas, and Pi. fucata), expression occurs during the trochophore stage in Cr. angulata and Pi. fucata, veliger and pediveliger in all three species, but only the mantle tissue of Cr. gigas (summarized in Figure 2, see also Figure 3; Supplementary Figure S2). This type of pattern is also prevalent with xenopsins. For example, of the 13 xenopsin in the M. edulis genome which are commonly expressed during larval stages, only two copies are present in the adult mantle (summarized in Figure 2, see also Supplementary Figure S1). Third, the number of opsin genes expressed for a given type is higher in the veliger and pediveliger stages than in the trochophore for most species. To summarize, 76 opsins were expressed in the veliger stage across all focal species versus 57 and 76 genes in the trochophore and pediveliger stages, respectively (Figure 2). While not a strong trend, this pattern is notable for the xenopsin (opnG_x) in mytilid species M. coruscus and M. edulis (Figure 2) with nine and six xenopsins being expressed during the veliger stage, respectively, versus six and five xenopsins in the pediveliger stage.

When looking at changes in relative expression level of specific opsins rather than presence/absence of expression, no clear patterns emerge, except that retinochrome (RTC) was the most highly expressed gene across the focal species (Figures 3, 4; Supplementary Figures S1, S2) when considering non-eye tissue samples. Instead, changes during larval development are largely lineage-specific among our seven focal species. For example, when comparing opsin expression between the two oyster species, Cr. angulata and Cr. gigas, only two of the 12 genes, a G_o-opsin, opnGo.A.OS, and a xenopsin, opnGx.B1.OS, have similar changes (Figure 3). Seven of the opsin genes have opposing expression profiles (e.g., opn5.OS, opnGq-nc.B.OS, and opnGq.B.OS; Figure 3). Comparing the oysters to their most closely related family, Margaritidae (Pi. fucata), gene expression is dissimilar for the orthologous neuropsin (opn5), which is not expressed in any of the Pi. fucata samples (Supplementary Figure S2), and the orthologous xenopsin (opnGx.B1.MA) is below the expression threshold for the Pi. fucata (Supplementary Figure S2). The three remaining orthrologs, opnGq.B, peropsin, and RTC, have grossly similar expression patterns in the larvae with highest levels of expression in pediveliger (opnGq.B and peropsin) or trochophore (RTC; Supplementary Figure S2; Figure 3). The remaining Pi. fucata opsins cannot be directly compared to oyster opsins due to lineage-specific duplications in xenopsins, G_o-opsins, and noncanonical r-opsins for each family (Figures 1B–D).

Opsin expression between the pair of scallop species, Ch. farreri and Pe. maximus, appears to be more conserved than in Ostreidae. When genes were above the expression threshold, expression patterns were more similar among larval stages and when those stages were compared to the adult mantle tissue (e.g., opnGo.B.PE, opnGq.B.PE.2; Figure 4). However, expression levels of many scallop opsins were low, and often only one of the species pair had expression above its species-specific threshold. For example, low expression of xenopsin (= opnG_x) was observed for both species in genes opnGx.A.PE.2, opnGx.B1.PE.2, but only Pe. maximus has expression above the threshold for opnGx.B1.PE.3 and opnGx.B2.PE.1 (Figure 4).

Opsin expression between the mytilid species, M. coruscus and M. edulis, was the most conserved. Both relative level and expression pattern across all opsin types were mirrored between the species (Supplementary Figure S1). However, genomic content varied, most notably for the xenopsin type (Figure 2). Mytilus edulis had additional copies of xenopsin that ranged from one new copy in the A clade to five more copies in B2 clade (e.g., opnGx.A.MY.2b, opnGx.B2.MY.3e, opnGx.B2.MY.3f; Supplementary Figure S1). These copies are the result from a series of paralogous duplication events within the Mytilidae (Figure 1B).

Opsins were expressed at relatively lower levels in adult mantle tissue than in larvae (Figures 3, 4; Supplementary Figures S1, S2). This was observed across all opsin types in all focal species with the exceptions of two non-canonical r-opsins and one xenopsin (opnGx.A.OS.1) in Cr. angulata (Figure 3) and one of the pectinid-specific r-opsin paralogs, opnGq.B.PE.2, in the two scallop species (Figure 4). In contrast, when eyes were present, opsin expression was higher in eye tissue than in mantle or any larval stage. The relative expression of 12 of the 18 scallop opsins were higher in the adult eye samples of Ch. farreri (Figure 5). These 12 opsins represent the six opsin types that scallops possess (pectinids do not have a peropsin; Figure 2), and 10 of these opsin genes are pectinid-specific paralogs from the xenopsin A and B1 (Figure 1B) clades, non-canonical and canonical r-opsin clades (Figure 1C), and G_o-opsin clade (Figure 1D). One copy of the paralog pairs of xenopsin (opnGx.A.PE.2), G_o-opsin (opnGo.B.PE) and non-canonical r-opsin (opnGq-nc.B.PE.1) have relatively higher expression in eye tissue than in the larvae, while expression all four canonical r-opsin paralogs increased between 5.6 to 18.5 K fold in the eye (Figure 5). To assess whether the variation in opsin expression observed between larval stages and eyes extended throughout the system, we examined the expression levels of four housing keeping genes from the same samples. Our findings revealed consistently similar expression patterns of housekeeping genes between eyes and each larval stage, suggesting that the difference in opsin expression between larval stages and eyes are biologically meaningful and not the result of RNAseq data artifacts (Figure 5; Supplementary Data 2).

We first identified and phylogenetically placed opsin genes from the genomes of seven focal species (Figure 1). Of the 119 opsin genes from our focal species, all but nine were expressed in at least one larval stage indicating that opsins were important to general larval function. When an opsin copy was not expressed in the larvae, these genes were almost always paralogous duplicates for that taxon or family (except for neuropsin in Pi. fucata; Supplementary Figure S2), suggesting that paralogs have diverged in function after duplication. Presence of opsin expression varied across developmental stages and species, but the greatest number of opsins was expressed in the two later larval stages veliger and pediveliger (76 and 76 out of 119 genes, respectively) versus 57 opsin genes in the trochophore stage (Figure 2). The only other bivalve study to examine opsin expression in larvae is from a non-pteriomorphian and eyeless species, the razor clam Sinonovacula constricta (Infraclass: Heteroconchia; Kong et al., 2023). Like our results, the majority of S. constricta opsins (17 out of 23 genes) was expressed in the larvae. Both number of opsins expressed, and relative expression levels increased from the trochophore to pediveliger stage (Kong et al., 2023). Also, like our findings (except in scallop eyes), opsins were generally lowly expressed in the adult tissues. These results provide an independent data point of opsin expression coinciding with the timing of metamorphic competence and support our hypothesis that opsins play a role in identifying the cues involved in settlement.

One of the few opsins with consistent expression patterns across different species and developmental stages was retinochrome (Figure 2), which was often the most highly expressed opsin in these datasets (Figures 3–5; Supplementary Figures S1, S2). This opsin, first discovered in cephalopods (Hara and Hara, 1965; Hara et al., 1967), acts primarily as a photoisomerase for converting all-trans to 11-cis retinal (reviewed in Terakita and Nagata, 2014; Vöcking et al., 2022). That is, it likely does not drive phototransduction and instead acts to resupply 11-cis retinal for rhodopsin (Vöcking et al., 2021; Kong et al., 2023). Retinochrome is found across all mollusk clades (e.g., Ramirez et al., 2016; McElroy et al., 2023) and in other lophotrochozoans, though its function is only known from mollusks (Vöcking et al., 2021). Unlike other groups of opsins, retinochrome does not regularly duplicate and diversify; it is almost typically represented by a single gene in mollusks, indicating that it is likely functionally restricted (though see examples in Kong et al., 2023; McElroy et al., 2023). As in McElroy et al. (2023), no duplications of retinochrome were seen in pteriomorphian bivalves in this study. In addition to resupplying 11-cis retinal, retinochrome has been hypothesized to act as a storage protein for retinal (Ozaki et al., 1983). These critical functions may drive demand for retinochrome presence in all light-responsive cells, but currently little is known about opsin expression across development and tissue types in a broad range of mollusks.

The other opsin type in mollusks expected to act as an isomerase is the molluscan peropsin (Ramirez et al., 2016; Vöcking et al., 2021). Like retinochrome, this opsin is largely resistant to duplications, but has been lost numerous times (McElroy et al., 2023). Here, the two Crassostrea species and Pinctada fucata are the only taxa whose genomes encode peropsin. In both groups of species, we found peropsin expressed across all larval stages with apparently increasing expression levels from trochophore through pediveliger (Figure 3; Supplementary Figure S2). Determining if peropsin functions similarly to retinochrome in the classic molluscan visual cycle (Terakita et al., 1989) and whether it can drive phototransduction are important first steps in defining the role for this opsin. Furthermore, in species with both retinochrome and peropsins, visual (e.g., immunohistochemistry) or transcriptomic (e.g., single-cell RNA-seq) analysis should be conducted to determine if photoreceptors and other cell types express both opsins. Together, these investigations should help shed light on why some lineages maintain these putative photoisomerases, while other species lose it.

Opsin may have a role in larval exploration of suitable settlement sites. We found relatively higher levels of opsin expression in the veliger and pediveliger larval stages for peropsin and some of the paralogs of G_o-opsin, canonical and noncanonical r-opsins (opnG_q), and xenopsin (opnG_x). Increased number and expression levels of opsins in these later stages may be related to increasing sensory needs as the larva approaches metamorphic competency. It has been demonstrated that larvae alter their response to light at different developmental stages, going from positive phototaxis in veligers to negative phototaxis in pediveligers (e.g., Mytilus edulis in Bayne, 1964). This likely is opsin-based, as opsin has been shown to be expressed in the larval eyespots of other marine invertebrates [Polyplacophora (Vöcking et al., 2015); Platyneresis dumerilli (Randel et al., 2013); the flatworm Maritigrella crozieri (Rawlinson et al., 2019)]. While the specific location of where each opsin expressed in pteriomorphian larvae is still unknown, the photosensitive eyespots are ubiquitous among molluscan larvae, forming in the late veliger or early pediveliger stages of bivalves (reviewed in Cragg, 2016). These simple organs located in the anterior aspect of each gill bar consist of two cells, a photoreceptor cell and a pigment cell, and can sense direction and intensity of light, but lack spatial vision (Hodgson and Burke, 1988). Both “visual” opsins, those expressed in adult image-forming eyes (e.g., G_q-opsins in Randel et al., 2013; Vöcking et al., 2015), as well as opsins that have not been demonstrated to have a role in vision (e.g., xenopsins in Rawlinson et al., 2019), have been shown to be expressed in larval eyespots.

In pteriomorphians, opsin may play an important role in coordinating with a yet-to-be-determined chemosensory system to orchestrate larval settlement, perhaps analogous to the cryptochrome-based photosensing system in the sponge, Amphimedon queenslandica (Say and Degnan, 2020). In the sponge, detecting the cessation of light is required for the larvae to respond to a highly inductive biochemical cue, otherwise, larvae are unable to settle if maintained in constant light. Light was shown to influence expression of nearly 180 genes critical for settlement (Say and Degnan, 2020). Many of these genes possessed known G-protein regulatory motifs that repress the GPCR signaling of chemotransduction in A. queenslandica and likely maintain larvae in a state that is unable to respond to biochemical cues until larvae transition in to the dark (Say and Degnan, 2020). Future work in Pteriomorphia should examine these light-mediated changes to gene expression profiles during settlement and metamorphosis.

Another critical sensory modality in metamorphic competency is chemoreception. For many diverse marine invertebrates, GPCRs, the same superfamily as opsin, are the chemoreceptors that regulate settlement. This has been demonstrated across diverse metazoans such as the gastropod Haliotis rufescens (Trapido-Rosenthal and Morse, 1986), the echinoderm Stronglylocentrotus purpuratus (Amador-Cano et al., 2006), the sponge Amphimedon queenslandica (Say and Degnan, 2020), and cnidarians Hydractinia echinata (Schneider and Leitz, 1994) and Acropora millepora (Strader et al., 2018), but see (Holm et al., 1998; Tran and Hadfield, 2012). Intriguingly, Baxter and Morse (1992) proposed that the chemosensor that induces settlement and metamorphosis in the gastropod Haliotis is not only a GPCRs, but likely is a member of the rhodopsin-like class of GPCRs, as is opsin, which comprises subfamily A16. Perhaps some portion of the large and diverse opsin repertoire in pteriomorphian larvae function as chemoreceptors?

There is a growing body of evidence that opsins have multimodal functionality (Feuda et al., 2022). Opsin has been shown to have light-independent sensory modalities including chemosensory (Leung et al., 2020), auditory (Senthilan et al., 2012), mechanoreception (Katana et al., 2019), and temperature reception (Shen et al., 2011; reviewed in Leung and Montell, 2017). A promising candidate is xenopsin. A recently described opsin type (Ramirez et al., 2016), xenopsin is an under-characterized opsin restricted to Lophotrochozoa (Ramirez et al., 2016; Vöcking et al., 2017). It is associated with ciliary photoreceptors and may be co-expressed with G_q-opsins (Vöcking et al., 2017; Döring et al., 2020). Xenopsin is particularly prone to large gene family expansions in both pteriomorphian and non-pteriomorphian bivalves (Figure 1; McElroy et al., 2023). Furthermore, these gene copies are most commonly expressed in the later developmental stages of pteriomorphian (summarized in Figure 2) and heteroconchian S. constricta larvae, with few expressed in adult tissue (Figure 3; Kong et al., 2023). For these reasons, we think xenopsins may be important for species-specific cues in development. Future work should target specific spatiotemporal expression patterns for xenopsins in bivalves across life stages.

Opsins are worthwhile proteins to explore in the context of life-stage triggers and decisions of settling in mollusks, which require multisensory inputs. Future work should be to test functions. A first step is to determine whether candidate opsins form photosensitive pigments when provided an appropriate chromophore. Assays to test if an opsin can form a functioning photopigment can be conducted in heterologous expression systems, where opsin is expressed outside of the animal and then and then absorbance spectra can be quantified (Faggionato and Serb, 2017; Smedley et al., 2022). Second, we can test whether the candidate opsin can perform as a chemoreceptor. Because GPCRs are one of the most common pharmaceutical targets (Sriram and Insel, 2018), there are high-resolution GPCR structures in dedicated repositories such as GPCRdb (Pándy-Szekeres et al., 2018) and GPCR-EXP (Chan and Zhang, 2020) available to investigate the molecular basis of GPCR structure–function relationship and characteristic features of ligand binding (reviewed in Venkatakrishnan et al., 2013). Furthermore, there are a wealth of protein ligand interaction databases that consists of a list of active site residues of a protein and the physio-chemical properties of ligands. Ligand compatibility can be examined with computational approaches allow modeling of ligand docking (e.g., GPCR-ModSim Esguerra et al., 2016) and ligand predictions based on protein models [pdCSM-GPCR (Velloso et al., 2021); others listed in Allen and Roth, 2011], such as the AlphaFill algorithm applied to Alphafold models (Hekkelman et al., 2023). These in silco studies could be followed up with in vitro testing of ligand binding to test for light-independent functions in an opsin (reviewed in Allen and Roth, 2011).