RFT1 encodes a polytopic endoplasmic reticulum membrane protein of the MOP (multidrug/oligosaccharidyl-lipid/polysaccharide) transporter superfamily that acts on Man(5)GlcNAc(2)-PP-dolichol (M5-DLO), the lipid-linked oligosaccharide intermediate of the N-glycosylation pathway. The protein spans the membrane roughly 12-14 times with both termini facing the cytosol, is not itself N-glycosylated, and is essential for viability in yeast and mammalian cells. Purified human RFT1 reconstituted into proteoliposomes translocates M5-DLO across the bilayer by an ATP-independent, bidirectional (scramblase-like) mechanism, with selectivity for the Man5 over shorter Man3 intermediates. Whether this translocase activity is the essential physiological role of the protein remains actively disputed: M5-DLO flipping persists in Rft1-depleted microsomes and in vesicles reconstituted with Rft1-free ER protein, Rft1-null Trypanosoma brucei is viable and glycosylates normally, and a yeast mutant engineered to block the lipid portal still supports robust growth, leading to an alternative model in which RFT1 acts as an M5-DLO chaperone that routes the intermediate for luminal completion. What is not in dispute is the pathway step at which the protein is required: loss of RFT1 function stalls dolichol-linked oligosaccharide assembly at M5-DLO and causes RFT1-CDG (CDG-In, OMIM 612015), a severe multisystem congenital disorder of glycosylation with neurodevelopmental abnormalities, hypotonia, epilepsy and sensorineural deafness.
Definition: Catalysis of the movement of a glycolipid from one leaflet of a membrane bilayer to the other, in either direction, by an ATP-independent mechanism.
Justification: GO currently offers no ATP-independent glycolipid translocase term. GO:0034202 (glycolipid floppase activity) and GO:0140327 (flippase activity) both require energy from ATP hydrolysis, and GO:0017128 (phospholipid scramblase activity) is restricted to phospholipid substrates. RFT1 and its archaeal counterpart Agl23 translocate Man(5)GlcNAc(2)-PP-dolichol in reconstituted liposomes with no ATP present, so no existing term states the demonstrated activity at its true specificity; curators are forced either to use a definitionally wrong ATP-dependent term or to retreat to the generic parent GO:0140303. The same gap affects other MOP-superfamily glycolipid scramblases.
Parent term: intramembrane lipid transporter activity
Supporting Evidence:
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005789 endoplasmic reticulum membrane | IBA GO_REF:0000033 | ACCEPT | Summary: Phylogenetic (PAINT) propagation of ER membrane localization across the Rft1 family (PTN000319164), with yeast RFT1 (SGD:S000000116) and human RFT1 itself among the donors. ER residence is the one part of the RFT1 picture that no party to the flippase controversy disputes, and it now has direct human evidence - Hirata et al. localized human Rft1 and mapped its topology, showing a polytopic ER protein with both termini cytosolic. Reason: ER membrane is the well-established site of action and is independently supported by yeast genetics, human topology mapping, and the reconstitution work. The appearance of Q96AA3 in its own WITH/FROM reflects that the human gene's own experimental annotations helped seat the ancestral node, which is the expected pattern, not circularity. Supporting Evidence: PMID:39025454 We show that it is a multispanning membrane protein located in the ER, with its N and C termini facing the cytoplasm. PMID:18313027 The definition of the RFT1 defect establishes the functional conservation of the DolPP-GlcNAc(2)Man(5) translocation process in eukaryotes. |
| GO:0034203 glycolipid translocation | IBA GO_REF:0000033 | ACCEPT | Summary: Phylogenetic propagation of the glycolipid translocation process to the Rft1 family. The node placement is defensible - the biochemical capacity is conserved from archaea (HhAgl23) through yeast to human, as Chen et al. showed by reconstituting purified ScRft1, HsRft1 and HhAgl23 in the same assay. Reason: The biological process term is directionally and energetically neutral, so it survives the controversy over whether RFT1's essential in-vivo role is transport or chaperoning - the family demonstrably translocates M5GN2-PP-Dol in vitro. Note this is a weaker claim than the MF rows make, and is the level at which the family assertion is safe. Supporting Evidence: PMID:38886340 Translocation was ATP-independent, occurred with similar kinetics for both dolichol or phytanol lipid carriers, and importantly, was dependent on addition of purified Rft1 or archaeal Agl23 |
| GO:0005789 endoplasmic reticulum membrane | IEA GO_REF:0000044 | ACCEPT | Summary: Electronic annotation from the UniProtKB/Swiss-Prot subcellular location vocabulary (SL-0097). UniProt's own CC line carries this by similarity to yeast Rft1 (ECO:0000250|UniProtKB:P38206), but direct human evidence now exists. Reason: Correct localization, consistent with every other evidence line and with direct human localization/topology data. Supporting Evidence: PMID:39025454 We show that it is a multispanning membrane protein located in the ER, with its N and C termini facing the cytoplasm. |
| GO:0006488 dolichol-linked oligosaccharide biosynthetic process | IEA GO_REF:0000002 | ACCEPT | Summary: InterPro2GO mapping from IPR007594 (Rft1 family). This is the safest process term for RFT1 - it states involvement in LLO assembly without committing to the mechanism by which RFT1 is required. Reason: Loss of RFT1 stalls dolichol-linked oligosaccharide assembly at the M5GN2-PP-Dol intermediate in both yeast and human cells, so pathway involvement is beyond dispute even where the mechanism is not. The domain-to-term mapping is sound. Supporting Evidence: PMID:18313027 RFT1 deficiency in both yeast and human cells leads to the accumulation of incomplete DolPP-GlcNAc(2)Man(5) and to a profound glycosylation disorder in humans. PMID:19701946 The pathogenic character of the novel mutations was illustrated by the accumulation of Man(5)GlcNAc(2)-PP-dolichol and by reduced recombinant DNase 1 secretion. |
| GO:0016020 membrane | IEA GO_REF:0000002 | KEEP AS NON CORE | Summary: InterPro2GO mapping to the root membrane term. Correct but uninformative relative to the GO:0005789 annotations. Reason: Technically true - RFT1 is a polytopic membrane protein - but subsumed by the more specific ER membrane annotations. Harmless to retain; it should not be treated as a distinct localization claim. Supporting Evidence: PMID:39025454 We show that it is a multispanning membrane protein located in the ER, with its N and C termini facing the cytoplasm. file:human/RFT1/RFT1-deep-research-falcon.md Human RFT1 localizes throughout the ER. DeepTMHMM and experimental tagging/topology mapping indicate ~14 transmembrane spans |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | REMOVE | Summary: Nine IPI rows derived from the HuRI all-by-all binary yeast two-hybrid interactome map, with partners AQP6, BEST2, CNR2, CREB3L1, ERGIC3, MUC1, RNF144A, TMX2 and TSPAN12. All are membrane proteins; none has an established role in dolichol-linked oligosaccharide assembly or in N-glycosylation, and no follow-up validation of functional relevance to RFT1 has been reported. Reason: Bare "protein binding" conveys nothing about molecular function and is explicitly discouraged by project curation guidelines. These are unvalidated high-throughput Y2H hits between membrane proteins, a class prone to artefact, and none of them illuminates RFT1's activity or its contested essential role. If any of these interactions is later shown to be functional, the appropriate annotation would be a specific term, not GO:0005515. Supporting Evidence: PMID:32296183 Here we present a human 'all-by-all' reference interactome map of human binary protein interactions, or 'HuRI'. |
| GO:0034202 glycolipid floppase activity | IDA PMID:38886340 Rft1 catalyzes lipid-linked oligosaccharide translocation ac... | MODIFY | Summary: IDA from Chen et al. 2024, who reconstituted purified recombinant human RFT1 (HsRft1-FLAG) into proteoliposomes and measured M5GN2-PP-Dol translocation by coupled alpha1-2 mannosidase digestion, obtaining ~90% conversion in 2 h with selectivity for M5 over M3 substrate. The underlying activity is directly demonstrated on the human protein, so the evidence is sound. The term is not: GO:0034202 is defined as "Catalysis of the movement of a glycolipid from the cytosolic to the exoplasmic leaflet of a membrane, using energy from the hydrolysis of ATP", whereas the assayed activity is explicitly ATP-independent and was run in reactions containing no ATP. Reason: The energetics in the GO:0034202 definition contradict the experiment that grounds the annotation. RFT1 is an ATP-independent, bidirectional scramblase-type carrier, not an ATP-driven floppase; the Menon-lab literature consistently calls it a scramblase. GO:0017128 has the right energetics but the wrong substrate class (phospholipid), and GO:0140327 flippase activity is both ATP-dependent and the wrong direction. GO:0140303 intramembrane lipid transporter activity is the correct ATP-agnostic parent, and matches the wording UniProt already uses for Q96AA3 ("Intramembrane glycolipid transporter"). A new ATP-independent glycolipid scramblase child term is proposed under proposed_new_terms so the activity can eventually be stated at full specificity. Note this MODIFY is about term energetics and is independent of the separate, still-open question of whether this activity is RFT1's essential in-vivo role. Proposed replacements: intramembrane lipid transporter activity Supporting Evidence: PMID:38886340 As with ScRft1, proteoliposomes reconstituted with HsRft1 showed almost 90% conversion of M5GN2 to M3GN2 after 2 h of incubation. PMID:38886340 we sought to reconstitute liposomes with purified ScRft1 or HhAgl23 in reactions that contained Ξ±1-2 mannosidase, without ATP PMID:38886340 this transbilayer translocation is rapid, ATP-independent, protease-sensitive, and occurs with exquisite substrate specificity |
| GO:0034203 glycolipid translocation | IDA PMID:38886340 Rft1 catalyzes lipid-linked oligosaccharide translocation ac... | ACCEPT | Summary: IDA for the biological process from the same fully reconstituted assay. Purified human RFT1 was necessary and sufficient for M5GN2-PP-Dol movement across the bilayer in protein-free liposomes. Reason: This is the strongest and least contested annotation on the gene. Unlike the molecular function row, the process term makes no claim about energy coupling or direction, so it is fully satisfied by the reconstitution data. It also remains true under the alternative model - even authors who argue that scrambling is a moonlighting activity accept that the activity has been demonstrated with purified protein. Supporting Evidence: PMID:38886340 Translocation was ATP-independent, occurred with similar kinetics for both dolichol or phytanol lipid carriers, and importantly, was dependent on addition of purified Rft1 or archaeal Agl23 PMID:42417535 While this activity has been demonstrated in liposomes reconstituted with purified Rft1, biochemical evidence of additional M5-DLO scramblases and the viability of Rft1-null Trypanosoma brucei suggest that scrambling may be a moonlighting function of Rft1 rather than its essential cellular role. |
| GO:0034202 glycolipid floppase activity | TAS Reactome:R-HSA-4570573 | MARK AS OVER ANNOTATED | Summary: TAS from the Reactome disease reaction "Defective RFT1 does not flip the N-glycan precursor". Reactome asserts the canonical flippase model, which is a reasonable reading of the genetics but inherits the same ATP-dependence mismatch as the IDA row, and additionally states as settled a mechanism that the primary literature still treats as open. Reason: This row has the same definitional problem as the IDA row - GO:0034202 requires ATP hydrolysis and RFT1's translocase activity is ATP-independent - but re-terming it is not sufficient, because the row's problem is not only the term. A TAS to a pathway database is weak evidence for a mechanism that is actively contested; Rft1-depleted sealed microsomes retain full M5-DLO flippase activity, and Rft1-null trypanosomes glycosylate normally, so "defective RFT1 does not flip the N-glycan precursor" is a model rather than an established fact. The IDA row already carries the real evidence for this activity and has been re-termed to GO:0140303 intramembrane lipid transporter activity; re-terming this row as well would preserve it as a second, independent-looking assertion of the same contested mechanism, inflating the apparent support. It is therefore marked as an over-annotation rather than modified. Supporting Evidence: PMID:19494107 no difference was seen in the level of M5-DLO flippase activity in sealed wild type and Rft1-depleted microsomal vesicles PMID:23720757 They have normal steady-state levels of mDLO and significant N-glycosylation, indicating robust M5-DLO flippase activity. |
| GO:0005789 endoplasmic reticulum membrane | IGI PMID:18313027 Human RFT1 deficiency leads to a disorder of N-linked glycos... | ACCEPT | Summary: IGI with yeast RFT1 (UniProtKB:P38206) from Haeuptle et al., who showed that human RFT1 cDNA complements the Deltarft1 yeast growth and glycosylation defect, implying that the human protein reaches and functions in the same compartment. Reason: Cross-species complementation is legitimate genetic support for shared site of action, and the conclusion is corroborated by direct human localization and topology data. Nothing in the flippase controversy bears on where the protein sits. Supporting Evidence: PMID:18313027 Despite the low sequence similarity between the yeast and the human RFT1 proteins, we demonstrated both their functional orthology and the pathologic effect of the human p.R67C mutation by complementation assay in Deltarft1 yeast cells. PMID:39025454 We show that it is a multispanning membrane protein located in the ER, with its N and C termini facing the cytoplasm. |
| GO:0006487 protein N-linked glycosylation | IMP PMID:18313027 Human RFT1 deficiency leads to a disorder of N-linked glycos... | ACCEPT | Summary: IMP from RFT1-CDG patient fibroblasts homozygous for p.R67C, which show protein hypoglycosylation that is corrected by lentiviral expression of wild-type RFT1 cDNA. Three further patients with independent missense alleles (p.R67C, p.K152E, p.E298K) reproduce the phenotype and the rescue. Reason: A core, uncontested biological process annotation with mutant-phenotype plus rescue evidence in human cells. This term states that RFT1 is required for N-glycosylation without asserting the mechanism by which it is required, which is exactly the right level of commitment given the open mechanistic question. Supporting Evidence: PMID:18313027 The causality of the RFT1 p.R67C mutation was further established by restoration of normal glycosylation profiles in patient-derived fibroblasts after lentiviral expression of a normal RFT1 cDNA. PMID:19701946 The pathogenic character of the novel mutations was illustrated by the accumulation of Man(5)GlcNAc(2)-PP-dolichol and by reduced recombinant DNase 1 secretion. |
| GO:0006488 dolichol-linked oligosaccharide biosynthetic process | IMP PMID:18313027 Human RFT1 deficiency leads to a disorder of N-linked glycos... | ACCEPT | Summary: IMP for LLO assembly - RFT1-deficient patient cells accumulate DolPP-GlcNAc2Man5 and fail to build the mature Glc3Man9GlcNAc2-PP-dolichol precursor, pinpointing the step at which the pathway stalls. Reason: This is the most defensible core process annotation for RFT1 and the one that best represents current knowledge. It captures the established requirement at the M5-DLO step while remaining agnostic between the transporter and chaperone models, both of which predict exactly this accumulation phenotype. Supporting Evidence: PMID:18313027 RFT1 deficiency in both yeast and human cells leads to the accumulation of incomplete DolPP-GlcNAc(2)Man(5) and to a profound glycosylation disorder in humans. PMID:23720757 rather than facilitating M5-DLO flipping, Rft1 facilitates conversion of M5-DLO to mDLO by another mechanism, possibly by acting as an M5-DLO chaperone. |
| GO:0034203 glycolipid translocation | IGI PMID:18313027 Human RFT1 deficiency leads to a disorder of N-linked glycos... | MODIFY | Summary: IGI with yeast RFT1 (UniProtKB:P38206) inferring the translocation process from complementation of Deltarft1 by human RFT1 cDNA. The experiment shows that the human protein restores pathway flux in cells lacking yeast Rft1; it does not distinguish transport from any other essential contribution to M5-DLO handling. Reason: This is precisely the inferential step the field disputes. Genetic loss of Rft1 causes M5-DLO accumulation, but the same genetic background retains full flippase activity in sealed microsomes (Rush 2009), Rft1-null Trypanosoma brucei flips M5-DLO and glycosylates normally (Jelk 2013), and a yeast Rft1 mutant engineered to block the lipid portal - and so predicted to lack scramblase activity - still supports robust growth (Chiduza 2026). Complementation therefore supports involvement in dolichol-linked oligosaccharide biosynthesis, the term the same paper already grounds by IMP, rather than the specific transport step. The transport claim is not lost by this change - it is carried, with far better evidence, by the Chen 2024 IDA rows. Proposed replacements: dolichol-linked oligosaccharide biosynthetic process Supporting Evidence: PMID:19494107 although Rft1 may play a critical role in vivo, depletion of this protein does not impair the transbilayer movement of M5-DLO in sealed microsomal fractions prepared from disrupted cells. PMID:23720757 We report that TbRft1-null procyclic trypanosomes grow nearly normally. PMID:42417535 Strikingly, the portal-blocking mutant which is predicted to lack scramblase activity supported robust growth. |
| GO:0005789 endoplasmic reticulum membrane | TAS Reactome:R-HSA-4570573 | ACCEPT | Summary: TAS localization from the same Reactome disease reaction. Unlike the molecular function assertion in that reaction, the localization claim is uncontroversial. Reason: ER membrane residence is agreed by all parties and directly demonstrated for the human protein, so the traceable author statement is accurate even though the mechanistic framing of the parent Reactome reaction is not. Supporting Evidence: PMID:39025454 We show that it is a multispanning membrane protein located in the ER, with its N and C termini facing the cytoplasm. |
Loading supporting contentβ¦
Download this section (compressed HTML)Q: Should GO annotate the reconstituted, ATP-independent M5-DLO translocase activity of RFT1 as its molecular function while the in-vivo essentiality of that activity is contested, or should the molecular function be held at a binding/carrier term until a separation-of-function allele is available?
Suggested experts: GO molecular function ontology editors, glycobiology curators (SGD, UniProt)
Q: Is there appetite for an ATP-independent glycolipid scramblase branch in GO, parallel to GO:0017128 for phospholipids, to cover RFT1, archaeal Agl23 and related MOP-superfamily glycolipid scramblases?
Suggested experts: GO ontology editors
Q: Should Reactome R-HSA-4570573 ("Defective RFT1 does not flip the N-glycan precursor") be reworded to describe the evidenced failure point - stalling of LLO assembly at M5-DLO - rather than asserting a contested catalytic mechanism, as the corresponding disease-mechanism curation has done?
Suggested experts: Reactome curators
Experiment: Take the portal-blocking yeast Rft1 mutant that supports robust growth and assay it directly in the Chen et al. reconstituted proteoliposome system. If purified mutant protein is scramblase-dead in vitro yet complements in vivo, the transport activity is formally dispensable and the essential role lies elsewhere. Run the reciprocal test on the two central-cavity mutants that grow poorly.
Hypothesis: RFT1's essential in-vivo function is M5-DLO binding/routing rather than transbilayer transport.
Type: reconstitution biochemistry with separation-of-function alleles
Experiment: Apply activity-correlation profiling to detergent-solubilized human ER membranes depleted of RFT1 - fractionate, reconstitute each fraction into liposomes, assay M5-DLO scrambling, and correlate activity with quantitative proteomics across fractions to nominate candidate genes, then test the top candidates by knockout plus reconstitution.
Hypothesis: A protein other than RFT1 carries the physiological M5-DLO flippase activity in human ER.
Type: activity correlation profiling mass spectrometry
Experiment: Mutate the invariant near-axis arginine and the conserved basic vestibule position, then measure M5-DLO binding affinity and scramblase activity in parallel on purified protein in proteoliposomes. Conservation from human to fungi to Dictyostelium marks these positions as selected, so what they are selected for is the question that matters; the characterised RFT1-CDG substitutions at the same positions are convenient ready-made reagents for the binding arm. Residue position alone cannot answer it - substrate binding is required under both the transporter and the chaperone model, so only a measured dissociation between the two readouts is informative.
Hypothesis: The deeply conserved basic positions on the RFT1 pore axis serve substrate capture, and substrate capture is separable from transbilayer movement.
Type: structure-function analysis of conserved substrate-site residues
Experiment: Build the human equivalent of the portal-blocking salt-bridge mutant that supports robust growth in yeast, then run both halves of the test that has so far only been done in halves - purify the mutant and show it is scramblase-dead in the reconstituted proteoliposome assay, and separately show it rescues RFT1-null human cells. This is the sharpest available discriminator - a scramblase-dead mutant that still rescues would establish that transport is not the essential function in human cells, whereas loss of rescue would restore the canonical model.
Hypothesis: Blocking the lateral lipid portal of human RFT1 abolishes transport in vitro but leaves the essential cellular function intact.
Type: separation-of-function mutagenesis with paired in-vitro and in-vivo readouts
Experiment: Mutate the most family-constrained cavity positions identified by the constraint analysis - R290, N435, Y378, E64, E298, Q186, N283, E156 and the S/T cluster at 257, 286 and 412 - alongside a size-matched and hydrophobicity-matched set of lipid-facing positions as controls, and score every variant in the reconstituted flipping assay. The prediction is a large asymmetry, cavity mutants impairing transport while matched surface mutants do not. The matched control set is the part that matters - without it the experiment cannot distinguish a functional cavity from the generic tolerance of lipid-facing positions to substitution.
Hypothesis: The conserved central cavity, not the lipid-facing surface, carries the residues required for M5-DLO handling.
Type: constraint-guided mutagenesis with matched surface controls
Experiment: Determine an experimental structure of human or yeast RFT1 with and without bound M5-DLO by cryo-EM. The cationic central cavity that underpins the current mechanistic model is a feature of substrate-docked predicted models; in the apo human AlphaFold model the innermost axis lining is net-acidic, so the electrostatics of the resting state and whether R67 and K152 actually contact the pyrophosphate are both unverified.
Hypothesis: The substrate-binding cavity of RFT1 becomes cationic only on substrate engagement.
Type: cryo-EM structure determination
What is not known β curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: It is not established which molecular activity of RFT1 is the one the protein evolved to perform and that is required for cell viability. The reconstituted M5-DLO scramblase activity may be a moonlighting capacity of a substrate-binding fold rather than its selected function, and no alternative activity (proposed candidates include chaperoning or routing M5-DLO on the cytosolic face) has been demonstrated.
NARROWING BIOLOGY MF_DARK
What is known: Firmly established - RFT1 is an essential polytopic ER membrane protein with cytosolic termini and a MOP/MATE-type two-lobed alternating-access architecture; it is required for dolichol-linked oligosaccharide assembly to proceed past M5-DLO in yeast and in mammalian cells; and purified human RFT1 translocates M5GN2-PP-Dol across a bilayer in vitro without ATP. Not established - that the in-vitro translocase activity is the activity required in vivo. The family-wide constraint result (selection concentrated on the polar central cavity rather than the lipid-facing surface) is discussed under resolution rather than here, because it comes from a single computational run that emitted no data artifacts and so cannot be recomputed from committed files.
Significance: Every molecular function annotation on this gene rests on the flippase attribution, as does the Reactome reaction that two of the GOA rows derive from. If the selected function is substrate capture and routing rather than transport, then a whole eukaryotic protein family is annotated with the wrong molecular activity, and the transport step of the N-glycosylation pathway is attributed to a protein that does not carry it.
What would resolve it: A separation-of-function allele would settle it - an RFT1 variant that retains scramblase activity in reconstituted liposomes but fails to support viability, or the converse. The portal-blocking mutant of Chiduza et al. is the first attempt at the second half of this and needs direct biochemical confirmation that it is indeed scramblase-dead. Identifying the additional M5-DLO scramblase activity present in Rft1-free membranes would also resolve which protein carries the transport step in vivo. Structure alone will not settle it - substrate binding is required under both models, so no amount of mapping residues onto the binding site can separate them; what is needed is a variant that separates binding from transport. An independent structural analysis makes the underlying point in the positive direction - the fold is a genuine MOP/MATE alternating-access architecture, and a family-wide constraint analysis shows selection has preserved the polar substrate cavity far more strongly than the lipid-facing surface, with the most constrained positions forming one compact pocket. Substrate capture is therefore an evolved feature of this protein whichever activity it ultimately serves. That result does not by itself settle the question either, for a reason the analysis does not draw - lipid-facing residues are the fastest-evolving class in essentially any membrane protein, so weak constraint along a lipid transit path is the null expectation and cannot show the path is unused.
Provenance (the field's own admissions):
Gap: The identity of the ER protein(s) responsible for the Rft1-independent M5-DLO flippase activity detected in microsomes and in reconstituted vesicles is unknown.
OPEN BIOLOGYCURATION MF_DARK
What is known: A specific, saturable, ATP-independent M5-DLO flippase activity is demonstrably present in ER membranes depleted of Rft1 and in vesicles reconstituted from Rft1-free ER protein, and Rft1-null trypanosomes retain it. No gene has been assigned to it.
Significance: Until this activity is assigned to a gene, the GO representation of the M5-DLO translocation step is incomplete, and RFT1 will keep absorbing annotations that may belong to another protein.
Provenance (the field's own admissions):
Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)