{"id":"65fb6908-6108-49e1-9a82-1d482da3101d","arxiv_id":"1908.05919","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Non-trivial θ-curve motifs appear in 52 protein structures, falling into 7 (or 9 chiral) topologies, with some evidence linking them to enzymatic function.","lead":"This paper defines θ-curves, a new class of topological motifs in proteins where a loop and a connecting bridge form a theta shape, and finds 52 protein structures with 7 non-trivial types. It links these motifs to enzyme function and organism origin, and uses simulations to examine how the knotted loop affects folding and stability.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The count of seven θ-curve topologies includes two classes (θ54, θ8n) found only in 4dou, whose ion-mediated edges the authors themselves call dubious; the headline classification is thus not robust to those artifacts.","rationale":"The central claim has two layers: the existence of non-trivial θ-curves in proteins, and the specific enumeration of 7 topologies (9 with chirality). The first layer is supported by at least the covalent coagulogen example (1aoc), which gives θ31 and θ01#31 without ion mediation. The second layer, however, is directly weakened by the authors' own admission that the two 4dou-specific topologies θ54 and θ8n may stem from a dense, possibly automatically assigned ion network with a 'highly dubious' Val-Ca interaction. Since those two entries are the sole representatives of their topology classes in Table 1, the headline '7 different topologies' is not robust under the authors' own caveat. This is not a mere disagreement with consensus; it is an internal tension between the stated result and the stated uncertainty. The reader identified the same weakest assumption, so the verdict should remain CONDITIONAL: the paper should be accepted only if the ion-mediated classes are validated or explicitly removed from the headline count. A concrete computational check—deleting all ion-mediated edges and re-running the scan—would settle whether the count drops and which topologies survive. If the covalent-only classes remain, the paper's core concept still stands, but the claimed enumeration and the 'unknotting number 2' novelty would need revision.","tokens_in":12715,"tokens_out":7668,"duration_ms":81057,"concrete_test":"Recompute the identification pipeline on the same PDB snapshot with all ion-mediated edges removed, keeping only covalent disulfide bridges and chain-closure edges, and compare the surviving topology list against Table 1. If θ54, θ8n, θ41, and θ01#52 disappear, the 7-topology claim is an artifact of including unvalidated ion coordinates as edges.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the Results section 'Types of θ-curves in proteins', the authors report 7 topologically different motifs (9 with chirality), with θ54 and θ8n listed in Table 1 as unique to chain 4douA (human adiponectin). In the same paragraph they write: 'both cases seem to be rather the result of a dense net of interactions of four spatially close ions, possibly assigned automatically (the θ54, features the highly dubious interaction Val-Ca).' These two topologies are the only representatives of their classes, and they are also the basis for the statement that proteins can contain a constituent knot with unknotting number 2. The central count therefore includes classes that the authors themselves are not confident are real. The problem is not limited to 4dou: θ41 (3ulkA/4wkkA) uses Mg498 and chain closure, and θ01#52 (3ihrA) uses Na331 and chain closure, so several of the seven classes similarly depend on ion-mediated edges being accepted as stable graph edges. If the ion assignments are crystallization or automatic-assignment artifacts, the number of topologies drops and the 'first unknotting number 2' observation disappears. The paper gives no validation of ion stability (occupancy, B-factors, conservation, or MD) and no code to re-run the assignment.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a systematic search for non-trivial θ-curve motifs in protein structures, treating the protein as a spatial graph with backbone edges, covalent (disulfide) bridges, ion-mediated interactions, and chain-closure edges. Applying three topological invariants (Yamada polynomial, Kauffman boundary links, constituent knots) implemented in the Topoly package to the PDB, the authors identify 52 non-redundant one-chain proteins and 4 multichain cases, which they group into 7 topological classes (9 including chirality). They further analyze the relation of these motifs to protein function and organism of origin, perform coarse-grained folding simulations of the covalent θ-curve protein coagulogen, quantify bridge-induced unfolding barriers, and use a polymer model to estimate which θ-curve topologies could plausibly form in designed or mutated proteins.","tokens_in":12992,"tokens_out":3390,"duration_ms":37348,"significance":"If the classification is robust, this is the first large-scale identification of non-trivial θ-curves in proteins and would substantially extend the known topological repertoire beyond main-chain knots, slipknots, and deterministic links. The paper's methodological strengths are its use of three independent topological invariants, its full-PDB scan, and its explicit local caveat about the dubious 4dou ion-mediated cases. The main limitation is that several of the reported classes rest on ion-mediated edges whose biological validity is not independently established; because the authors themselves discount the two classes unique to 4dou, the headline count of seven topologies is not yet on solid ground.","major_comments":[{"comment":"The reported count of seven topologies includes θ54 and θ8n, each of which is supported solely by chain 4douA, in a structure whose ion-mediated interactions the authors themselves describe as 'highly dubious' and 'possibly assigned automatically'. Because these two classes are also the basis for the claim that a protein θ-curve can contain a constituent knot with unknotting number 2, the central classification is not robust to ion-assignment artifacts. The manuscript should either validate the 4dou ion coordination sites (e.g., via occupancy, B-factors, coordination geometry, or a stability calculation) or exclude these two classes and revise the counts and the unknotting-number claim accordingly.","section":"Results, 'Types of θ-curves in proteins'; Table 1"},{"comment":"The classes θ41 and θ01#52, which are not restricted to 4dou, nevertheless depend on a single ion-mediated edge (Mg498 in 3ulkA/4wkkA and Na331 in 3ihrA) in addition to chain closure. The paper provides no evidence that these ion-mediated edges are stable, biologically meaningful interactions rather than crystallization or automated-assignment artifacts. Since several of the seven classes share this dependency, the paper should report how the classification changes under stricter ion-bond criteria and, if possible, provide occupancy/B-factor or simulation-based support for the retained ion-mediated edges.","section":"Results, 'Types of θ-curves in proteins'; Table 2"}],"minor_comments":[{"comment":"There are multiple typos and grammatical slips, including 'travesting' for 'traversing', 'detrministic' for 'deterministic', 'priciple' for 'principle', 'bacterie' for 'bacteria', and 'reacher' for 'richer'; these should be corrected in a revision.","section":"Throughout"},{"comment":"The phrase 'cite-sulkowska2012conservation,dabrowski2017topological' is an unresolved LaTeX citation; the intended references should be inserted.","section":"Results, 'Function, origin and conservation of the θ-curve motif'"},{"comment":"The entries for the Topoly package and the knotoid classification [30] and [36] contain unresolved '??' placeholders for journal and page information.","section":"References"},{"comment":"The caption does not define the symbols '...' and '↔', nor the meaning of 'Cls'; these are explained only in the main text and should be restated in the caption for readability.","section":"Table 2 caption"},{"comment":"The sentence 'we removed all structures with artificially long bonds or improbable gap filling' would benefit from concrete thresholds or a reference to the SI, since this filtering step affects the final counts.","section":"Results, 'The algorithm'"},{"comment":"The statement that 'no new deterministic θ-curves may be found in proteins by introducing a bridge' is presented as a general result, but it depends on the contact-cutoff and representative-selection procedure described only in the SI; a brief restatement of the cutoff and a caveat about its dependence would help the reader assess that negative result.","section":"Discussion"}],"recommendation":"major_revision","confidential_remarks":"The paper's dependence on self-cited tools (Topoly, Knoto-ID, previous knotoid work) is legitimate and does not affect my assessment. The main issue is that the authors' own admission about 4dou undercuts the headline count; this is fixable within the scope of the manuscript, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe paper's real contribution is the first systematic scan of θ-curves in proteins. It reports 52 non-redundant chains with non-trivial topologies, classifies them with three independent invariants (Yamada, Kauffman boundary link, constituent knots), and adds a knotoid-based descriptor for local geometry. That's a genuine extension of the protein-knot/lasso/link program, and the authors are careful to separate what is deterministic (covalent bridges) from what is probabilistic (chain closure). The polymer-model catalogue of attainable motifs is a nice bonus.\n\nThe biggest soft spot is the headline count. Two of the seven topologies, θ54 and θ8n, appear only in 4dou and rest on ion-mediated edges the authors themselves describe as dubious ('highly dubious interaction Val-Ca'). Those are also the basis for the 'unknotting number 2' observation. The paper keeps them in the main count anyway. Several other classes (θ41, θ01#52) also depend on chain closure or ion-mediated edges, so their robustness is weaker than the covalent θ31 in coagulogen, which is solid. The paper gives no validation of ion binding (occupancy, B-factors, conservation) and no code to rerun the assignment, which makes it hard to check the borderline cases.\n\nThe function-origin correlations are the thinnest part: '70% enzymes' compared to a 30% background with no statistical test. The chirality-origin correlation is suggestive but not quantified. These are minor relative to the classification, but they should be tightened.\n\nThe central claim—that non-trivial θ-curves exist in proteins and can be systematically classified—holds up. The issue is the exact count and the biological claims built on the borderline cases. The paper is for people working in protein topology and structural bioinformatics. It deserves a serious referee, but the revision should validate or exclude the ion-dependent topologies and release the detection pipeline.\n\nRecommendation: send to peer review; ask for robustness checks on the ion-mediated edges.","headline":"First systematic survey of θ-curves in proteins, with a solid classification pipeline but a headline count that leans on dubious ion-mediated edges the authors themselves flag.","tokens_in":13532,"tokens_out":2730,"would_cite":true,"duration_ms":23561,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Non-trivial θ-curves occur naturally in 52 protein chains, across seven topologies.","keywords":["θ-curves","protein topology","disulfide bridges","ion-mediated interactions","knotoids","Yamada polynomial","spatial graphs","knotted proteins"],"falsifier":"Re-examine the human adiponectin structure (code 4dou) at higher resolution or in solution: if the Val-Ca contact and the cluster of four nearby ions prove to be crystallization or automated-assignment artifacts, then $\\theta_{54}$ and $\\theta_{8n}$ leave the census, and the number of observed θ-curve topologies falls from seven to five. The complementary check is to mutate the residues that coordinate the ions and observe whether the θ-curves disappear in a functional assay.","tokens_in":1897,"feed_emoji":"🪢","tokens_out":2296,"duration_ms":94266,"temperature":0.7,"pith_summary":"Proteins are usually discussed as strings that may tie themselves into knots, but their covalent disulfide bridges and ion-mediated contacts add extra strands, so the whole assembly is a spatial graph. This paper asks whether that graph can contain a non-trivial θ-curve: a Y-shaped structure made of three arcs joining two junction points, which cannot be deformed to the planar letter θ without the arcs crossing. By scanning all protein structures deposited up to March 2018, the authors identify 52 non-redundant single-chain proteins and 4 multi-chain complexes with non-trivial θ-curves, representing 7 distinct topologies (9 if handedness is counted). The survey matters because it moves protein topology beyond knots and lassos to multi-vertex motifs, and it provides a first look at whether such shapes are functional or accidental.","feed_headline":"52 protein structures carry non-trivial θ-curves","feed_subtitle":"A scan of every known protein fold finds seven θ-curve topologies made from backbone, disulfide, and ion bridges.","key_machinery":"A θ-curve is an embedding of the letter θ in three dimensions: two trivalent vertices joined by three arcs. The load-bearing machinery is a pipeline that converts a protein into a spatial graph on Cα atoms (with edges for the backbone, disulfide bridges, ion-mediated contacts, and a chain-closure point at infinity), enumerates all cycles, and searches for an external arc connecting two residues of a cycle, which turns that cycle into a θ-curve. Each candidate is relaxed, simplified, and classified by three independent invariants: the Yamada polynomial of the spatial graph, the boundary-link invariant of θ-curves, and the constituent knots of the three arc pairs; the paper then assigns each motif a triple of knotoids, called its knotoid content, to distinguish different spatial realizations of the same topology.","core_discovery":"On the paper's own terms, the central discovery is that the spatial graph formed by a protein main chain plus its disulfide bridges and ion-mediated contacts does contain non-trivial θ-curves, and that all known examples fall into seven topological classes: the prime $\\theta_{31}$ and $\\theta_{41}$, the composite $\\theta_{01}\\#3_1$, $\\theta_{01}\\#4_1$, $\\theta_{01}\\#5_2$, and two ion-mediated curves in human adiponectin, $\\theta_{54}$ and an unnamed 8-crossing curve called $\\theta_{8n}$. Most of these motifs sit in proteins whose backbone is unknotted, and several of them also appear as main-chain knots in other proteins. The authors classify each curve's spatial realization by its constituent knotoids, and they show that the function and organism of origin correlate with the motif type: roughly 70% of the host proteins are enzymes, and animal proteins show a preference for one handedness of the trefoil-based curves. They also report that formation of the knotted loop in the purely covalent example (horseshoe-crab coagulogen) is not the rate-limiting step of folding, unlike main-chain knots.","pith_inferences":["A direct extension of the census would be to redo the scan as more structures are deposited; if the ion-mediated θ-curves in 4dou disappear under better ion assignment, the topology count drops from seven to five, while a growing list would show the motifs are more common than this first pass suggests.","The statistical model implies that engineered proteins with many clustered disulfide bonds (more than about ten) should be able to realize deterministic θ-curves that have never been observed; this is a concrete design target for de novo protein synthesis.","Because the same θ-curve topology can carry different knotoid triples, knotoid content could be used as a finer evolutionary fingerprint, letting researchers ask whether functional convergence acts on the spatial realization rather than on the bare topology.","The authors' closing suggestion to classify the full protein graph (all bridges at once) would generalize this work from θ-curves to arbitrary spatial graph motifs, and the Yamada polynomial used here is already the natural invariant for such a project."],"forward_implications":["The census gives the first concrete answer to whether non-trivial θ-curves are possible in proteins: they are rare but real, appearing in 52 of more than 128,000 known chains.","If the enzyme enrichment is not a detection artifact, θ-curves may rigidify part of the chain and help build active sites, a role already proposed for main-chain knots.","For the covalent θ-curve in horseshoe-crab coagulogen, folding is not slowed by the knotted loop: the loop forms late (at native-contact fraction around 0.8) and is not tied to the main free-energy barrier.","Bridge-removal simulations show the disulfide bridges that stabilize the protein are those that hold chain segments together; the knotted loop itself is not the stabilizing feature.","Adding plausible new disulfide bridges to all known structures produces no new deterministic θ-curve topologies, but would create new probabilistic ones in some knotted proteins, such as $\\theta_{66}$ or $\\theta_{65}$ from the $6_1$-knotted hydrolase 3bjx."],"supporting_citations":[{"why":"The reference enumerating θ-curves with up to seven crossings supplies the classification scheme that names the observed motifs.","marker":"[29]"},{"why":"Yamada's invariant of spatial graphs is one of the three invariants used to determine θ-curve topology.","marker":"[31]"},{"why":"The boundary-link invariant of θ-curves provides a second, independent classification method used to validate the results.","marker":"[32]"},{"why":"Supplies the topology calculation and simplification routines used throughout the search.","marker":"[30]"},{"why":"Introduces the knotoid formalism used to assign each θ-curve its constituent knotoid triple.","marker":"[33]"},{"why":"Assigns the knotoid types used to distinguish spatial realizations of each θ-curve.","marker":"[59]"},{"why":"Provides the crystal structure of horseshoe-crab coagulogen, the purely covalent θ-curve example used for folding and stability simulations.","marker":"[43]"},{"why":"Supplies the earlier observation that knots in proteins may aid enzyme function, used here as the comparison for the enzyme-enrichment analysis.","marker":"[22]"}],"fun_headline_variants":["θ-curves in proteins: seven topologies, one scan","Seven θ-curve topologies emerge from protein backbone and bridges","Seven θ-curve topologies found across known protein structures","Disulfide and ion bridges make seven θ-curve protein topologies","Seven non-trivial θ-curve topologies in known protein structures"],"cache_read_input_tokens":15616,"weakest_assumption_plain":"The count of seven θ-curve topologies assumes that the metal ions used as junction points in some structures (calcium, magnesium, sodium) really are held there by the protein and are not crystallization or software artifacts; the paper itself flags one such contact, Val-Ca in 4dou, as highly dubious.","fun_headline_variants_meta":{"raw":{"variants":["θ-curves in proteins: seven topologies, one scan","Seven θ-curve topologies emerge from protein backbone and bridges","Seven θ-curve topologies found across known protein structures","Disulfide and ion bridges make seven θ-curve protein topologies","Seven non-trivial θ-curve topologies in known protein structures"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001555,"raw_usage":{"total_tokens":6185,"prompt_tokens":883,"completion_tokens":5302,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":499,"completion_tokens_details":{"reasoning_tokens":5215}},"tokens_in":499,"tokens_out":5302,"duration_ms":34326,"temperature":1.0,"reasoning_tokens":5215,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:00:43.237603+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-examine the human adiponectin structure (code 4dou) at higher resolution or in solution: if the Val-Ca contact and the cluster of four nearby ions prove to be crystallization or automated-assignment artifacts, then $\\theta_{54}$ and $\\theta_{8n}$ leave the census, and the number of observed θ-curve topologies falls from seven to five. The complementary check is to mutate the residues that coordinate the ions and observe whether the θ-curves disappear in a functional assay.","supporting_citations":[{"cited_title":"An enumeration of theta-curves with up to seven crossings","cited_arxiv_id":null,"evidence_quote":"The reference enumerating θ-curves with up to seven crossings supplies the classification scheme that names the observed motifs."},{"cited_title":"An invariant of spatial graphs","cited_arxiv_id":null,"evidence_quote":"Yamada's invariant of spatial graphs is one of the three invariants used to determine θ-curve topology."},{"cited_title":"& Zhao, P","cited_arxiv_id":null,"evidence_quote":"The boundary-link invariant of θ-curves provides a second, independent classification method used to validate the results."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the topology calculation and simplification routines used throughout the search."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the knotoid formalism used to assign each θ-curve its constituent knotoid triple."},{"cited_title":"& Stasiak, A","cited_arxiv_id":null,"evidence_quote":"Assigns the knotoid types used to distinguish spatial realizations of each θ-curve."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the crystal structure of horseshoe-crab coagulogen, the purely covalent θ-curve example used for folding and stability simulations."},{"cited_title":"& Sulkowska, J","cited_arxiv_id":null,"evidence_quote":"Supplies the earlier observation that knots in proteins may aid enzyme function, used here as the comparison for the enzyme-enrichment analysis."}],"review_version":1}