{"id":"c51be88d-3d52-47a8-ac18-fc8629e28582","arxiv_id":"1908.05126","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":3,"one_line_summary":"A screen of 44,860 pentapeptides by molecular dynamics identifies 1,225 conformationally stable sequences that cluster into 54 backbone topologies proposed as a minimal block set for protein design.","lead":"Using molecular dynamics, the authors screened 44,860 five-amino-acid peptides built from an alanine backbone with three positions mutated, and found 1,225 that kept a single shape for over 80% of the simulation time. These stable peptides clustered into 54 backbone topologies, which the authors propose as a minimal set of building blocks for designing protein structures.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Completeness claim rests on unproven three-substitution sampling assumption; the paper never delivers the promised proof.","rationale":"The reader's weakest-assumption analysis identifies exactly the point on which the paper's headline claim depends: the sufficiency of sampling only three non-alanine positions. My reading of the full text confirms that this is asserted (Section 2.1), promised to be proved (Section 3), and then never proved; the Discussion replaces proof with a plausibility statement. No other issue is more load-bearing. The threshold choice for stability (80% of trajectory) and the clustering threshold (0.15) are also post hoc, but even if those thresholds were accepted, the incomplete sampling would still undermine completeness. Conversely, if the three-substitution assumption were validated—e.g., by showing four-substituted pentapeptides add no new topologies—the 54-type catalog would become a meaningful minimal set, though the stability thresholds would still need sensitivity analysis. The reader's conditional verdict is therefore the appropriate one: the work is reproducible and the catalog useful, but the completeness claim is not established. My concern does not move the verdict because the conditional acceptance already reflects this gap; it strengthens the specific justification for the condition.","tokens_in":23681,"tokens_out":2404,"duration_ms":28021,"concrete_test":"Run the same molecular dynamics protocol (AMBER/OPLS, 300 K, 10 ns, extended start, clustering at squared-Euclidean threshold 0.15 in phi/psi space) on a random sample of 10,000 pentapeptides in which four of the five positions are substituted, i.e., only one alanine position remains fixed. Cluster the stable representatives and compare the resulting topological clusters with the 54 listed in Supplementary Table 1. If any new cluster appears that is not within the clustering threshold of an existing representative, the three-substitution assumption is falsified and the completeness claim must be withdrawn or substantially weakened.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that 54 topological types of stable pentapeptides form a 'complete set' needed for pre-folded protein conformations. This completeness depends on a sequence-space restriction introduced in Section 2.1: only the central position and two of the four flanking positions are substituted in an alanine pentapeptide, giving 44,860 sequences from the 3,200,000 possible pentapeptides. Section 3 says 'Later we shall prove the validity of this assumption,' but no proof appears anywhere in the manuscript. The Discussion only asserts that replacing frequent alanines with other functional groups would 'substantially replenish a set of structurally stable elements, but it is unlikely to change the set of topological types found.' That is an informal expectation, not a demonstration. The 1,225 stable peptides and their 54 topologies are drawn entirely from a subspace in which at most three positions are ever non-alanine. Any stable backbone topology that requires interactions among four or five side chains, or a specific combination of four functional groups, is excluded by construction. The paper provides no saturation argument—for example, no test showing that a fourth substitution leaves the cluster set unchanged. The single-protein application to Na,K-ATPase is generated by the same restricted protocol and therefore cannot validate completeness. Consequently, the 'complete set' claim collapses if the three-substitution assumption fails, and that assumption is currently unsupported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript describes a computational search for pentapeptide sequences that adopt a single dominant backbone conformation in molecular dynamics simulations. Starting from the alanine pentapeptide AAAAA, the authors generated 44,860 sequences in which at most three positions are mutated, simulated each for 10 ns in vacuum with AMBER/OPLS, and classified a peptide as structurally stable if more than 80% of the sampled conformations fall into a single cluster (1,225 peptides). Clustering the representative conformations of these stable peptides produced 54 clusters, which are interpreted as topological types that can initiate, maintain, or terminate secondary-structure elements. The paper claims that these 54 types form a complete set needed for the pre-folded state of protein chains and illustrates the protocol on human Na,K-ATPase.","tokens_in":23948,"tokens_out":5982,"duration_ms":58355,"significance":"If the completeness claim were established, the 54-type catalog would provide a compact, rationally designable library of local backbone motifs, with obvious value for protein design and local-structure prediction. The paper's strengths are its large simulation set (44,860 independent 10-ns trajectories), the detailed supplementary tables of stable peptides, cluster representatives, dihedral angles, and RMSDs, and the reproducible description of the MD protocol in the Supplement. The significance is currently limited because the completeness claim rests on an unproven sequence-space restriction and on thresholds that are calibrated from the same data used to produce the catalog.","major_comments":[{"comment":"Section 2.1 introduces the restriction to at most three substituted positions, and Section 3 states 'Later we shall prove the validity of this assumption,' but no proof is provided anywhere in the manuscript; the Discussion only asserts that additional substitutions would 'unlikely' change the set of topological types. Because the abstract's 'complete set' claim is defined relative to this restricted subspace, the paper needs either a discharge of the promised proof, a saturation experiment (e.g., a sample of four- and five-substituted pentapeptides showing no new topologies), or a qualified claim restricted to the three-substitution subspace.","section":"Section 2.1 and Section 3"},{"comment":"The two parameters that define the catalog are chosen from the observed distribution: the 80% stability threshold is selected because of a local maximum in the largest-cluster size distribution, and the 0.15 clustering threshold is stated without any derivation. No sensitivity analysis is reported for either parameter. Since 1,225 stable peptides and the resulting 54 clusters are determined by these choices, the completeness and minimality of the 54-type set are not established.","section":"Section 2.3, Figure 1"},{"comment":"The human Na,K-ATPase analysis is presented as a demonstration, but it cannot validate the catalog: stability is assigned with the same vacuum-MD protocol, the same 80% threshold, and the same clustering distance that generated the 54 types. The reported 79 stable peptides and 32.4% coverage are not mapped onto the 54 topologies or compared with any independent prediction, so this result is descriptive rather than confirmatory.","section":"Section 3, Na,K-ATPase paragraph"},{"comment":"Twenty-six of the 54 clusters contain exactly one pentapeptide (clusters 29-54). With 1,225 stable peptides and no saturation analysis, the 54-type catalog could change substantially if the sampled sequence set were enlarged or if the clustering threshold were varied; the 'complete set' claim therefore lacks empirical support even within the three-substitution subspace.","section":"Supplementary Table 2 and Section 2.4"},{"comment":"The transferability of vacuum-MD stability to protein contexts is asserted but not demonstrated: all trajectories start from a single extended conformation, each peptide is simulated once for 10 ns with no solvent and no replicas, and no comparison with explicit-solvent simulations or experimentally observed conformations of the same sequences is reported. Given that the interpretation concerns pre-folded conformations in proteins, the stability measure needs a test of context dependence before the biological claim can be accepted.","section":"Section 2.2"}],"minor_comments":[{"comment":"The list of possible pairs of substituted positions contains a duplicate ('-1 and 1' appears twice) and omits the pair '-2 and +1'; the count of six pairs is correct, but the enumeration should be corrected.","section":"Section 2.1"},{"comment":"The residue name 'ASР' contains a Cyrillic character and should be written as 'ASP'.","section":"Section 3, Cluster 2 description"},{"comment":"The time specification is ambiguous: 'within 10,000 picoseconds' followed by 'after 5,000 picoseconds of relaxation' could be clarified as 5 ns relaxation plus 5 ns production; currently it reads as if the total simulation is 10 ns with the first half discarded.","section":"Section 2.2"},{"comment":"The supplementary table contains Russian column headers and several typographical artifacts (e.g., '127.6.2' and '119.4.'); these should be normalized to English and cleaned.","section":"Supplementary Table 2"},{"comment":"The figure caption refers to red and blue isolines (bold and thin) whose construction and numerical values are not described in the caption or Methods; the reader cannot reproduce the threshold choice without additional detail.","section":"Figure 1"}],"recommendation":"major_revision","confidential_remarks":"The paper is likely to be of interest to a specialized computational biology audience, but the strong completeness claim needs either substantial new evidence or a significant weakening. The authors' own text explicitly promises a proof that is never delivered; this should be addressed head-on in revision rather than only by rewording. I see no evidence of misconduct; the main concern is the gap between the evidence and the claims."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing to know: the paper gives you a concrete, reproducible catalog—1,225 pentapeptide sequences with >80% stable MD conformations, clustered into 54 backbone types—and the supplement actually ships the full lists and the X-PLOR protocol. That is new and useful. The second thing: the headline claim that these 54 types are a complete set of topologies for building protein-like folds is not supported by the evidence presented.\n\nWhat it does well: the enumeration is systematic within its chosen subspace (44,860 sequences from an alanine pentapeptide with at most three substitutions), the protocol is described well enough to re-derive, and the cluster table with phi/psi values is a genuine resource. The RMSD comparison to idealized secondary-structure elements is a nice sanity check, and the observation that most clusters match helix or beta geometry is reasonable.\n\nThe soft spots are real. The completeness claim rests on the assumption that three functional-group positions suffice to generate all topologically distinct stable backbone conformations. Section 3 promises 'Later we shall prove the validity of this assumption,' but no proof appears. The discussion only asserts that a fourth substitution would likely not change the topology set. That is not a demonstration. Any stable conformation requiring interaction among four or five side chains is excluded by construction. The 80% stability threshold and 0.15 clustering distance are read off the observed distribution—post hoc thresholds are fine for defining a catalog, but they cannot be used to justify a universal completeness claim. And the Na,K-ATPase application is the same vacuum-MD protocol and the same fitted thresholds, so it is a consistency check, not external validation. Calling it a 'prediction' is too strong.\n\nNone of this kills the paper as a resource. The catalog stands on its own. But the 'complete set' language needs to be either withdrawn or tested: one straightforward test is to check whether all 54 types (and any additional ones) appear when you enumerate pentapeptides from the PDB or from Protein Blocks, and to report how the cluster count changes with the two thresholds.\n\nI would send this to peer review, not desk-reject it—the dataset is reproducible and the question is worth asking—but I would insist on major revision focused on the completeness claim. The right framing would be 'a catalog of stable pentapeptides under a defined protocol' rather than 'a complete set of pre-folded building blocks.'","headline":"The paper's real contribution is the reproducible catalog of stable pentapeptides; the 'complete set of topologies' claim outruns the evidence.","tokens_in":24527,"tokens_out":3533,"would_cite":false,"duration_ms":31279,"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":"The paper claims that 54 stable pentapeptide backbone shapes form a complete set of local folds for protein-like structures.","keywords":["pentapeptide","protein folding","molecular dynamics","conformational stability","backbone topology","secondary structure","rational protein design","structural alphabet"],"falsifier":"Enumerate pentapeptides with four or five substituted positions beyond the alanine matrix, up to the full 3.2 million sequence space, cluster their molecular dynamics trajectories with the same 80% stability threshold, and look for any backbone topology outside the 54; one new stable cluster would refute the completeness claim. A weaker but still decisive check is to rerun the same 44,860 simulations with explicit solvent and see whether the stable set changes.","tokens_in":23501,"feed_emoji":"🧬","tokens_out":11488,"duration_ms":106225,"temperature":0.7,"pith_summary":"This paper sets out to show that a small, complete set of locally stable five-residue backbone shapes exists and can serve as the building blocks for designing polypeptide chains with a desired 3D structure. The authors scan 44,860 pentapeptide sequences generated from the alanine pentapeptide by changing at most three positions, run molecular dynamics on each, and find that only 1,225 of them (2.73%) hold one conformation for more than 80% of the simulation time. Clustering those stable conformations yields 54 backbone topologies, and the paper argues that these 54 types form a topologically complete set covering the local structural elements a protein chain needs before folding. If true, this gives rational design a finite dictionary of stable pieces, with sequence rules for each piece, instead of searching over continuous backbone space.","feed_headline":"54 stable peptide blocks may build any protein fold","feed_subtitle":"A scan of 44,860 pentapeptides yields a complete set of local backbone shapes for rational design.","key_machinery":"The load-bearing object is the 54-cluster library of stable pentapeptide backbone conformations, each cluster represented by one central structure and described by the eight backbone dihedral angles $\\varphi$ and $\\psi$ of its five residues. The argument is carried by a three-step pipeline: sequence generation from an alanine matrix with at most three substituted positions, giving 44,860 unique pentapeptides; molecular dynamics in vacuum with the AMBER/OPLS force field, saving 5,000 snapshots per peptide and declaring a peptide stable when a single cluster holds more than 80% of the trajectory; and a second clustering of the 1,225 stable representatives that yields the 54 topologies. The terminal tripeptide fragments of each topology are then compared with idealized secondary-structure coordinates, which lets the authors assign each block a functional role as initiator, maintainer, or terminator of $\\alpha$-helix or $\\beta$-structure, including transition blocks that redirect the chain from one secondary-structure element to another.","core_discovery":"The central claim is that a minimal library of 54 backbone topologies, each realized by a conformationally stable pentapeptide, is sufficient to account for the local structural units (the paper calls them pre-folded conformations) that a polypeptide chain uses on the way to its native three-dimensional structure. Starting from the alanine pentapeptide AAAAA and allowing no more than three of the five positions to be replaced by any of the 20 canonical amino acids, the authors generated 44,860 unique sequences, simulated each for 10,000 picoseconds with molecular dynamics in vacuum using the AMBER/OPLS force field, and kept the 1,225 peptides whose largest conformational cluster occupied more than 80% of the trajectory. Clustering those representatives by their $(\\varphi,\\psi)$ torsion angles gave 54 topological types. Comparison of the terminal $C_\\alpha$ atoms of each type with idealized $\\alpha$-helix and parallel/antiparallel $\\beta$-structure shows that 53 of the 54 clusters match one of these elements at one terminus with RMSD below 0.2 Å, and the authors classify each cluster as initiating, maintaining, or terminating secondary-structure elements. They further report that in human Na,K-ATPase, 79 of the 1,017 overlapping pentapeptides are stable and cover 32.4% of the sequence, which they read as evidence that stable blocks occupy a substantial part of real proteins and can guide rational design.","pith_inferences":["An untested corollary of the completeness claim is that four- or five-position substitutions should not create new stable backbone topologies; enumerating the full 3.2 million sequence space would settle that directly.","A possible extension is to repeat the scan with explicit solvent, since the vacuum protocol omits water-mediated contacts; new topologies there would make the 54 blocks a baseline rather than a complete set.","A design-oriented extension is to treat the 54 topologies as a structural alphabet with attached sequence rules and build chain backbones by concatenation, turning local stability into a planning problem.","The paper suggests two-state pentapeptides as a follow-up; such peptides could act as conformational switches at hinges of moving protein parts."],"forward_implications":["A scan of any protein sequence for the sequence patterns behind the 54 blocks would locate pre-folded regions; in the Na,K-ATPase example, 79 stable pentapeptides cover 32.4% of the chain.","Rational design could assemble a desired backbone path by concatenating blocks, using the initiator, maintainer, and terminator roles to control where secondary structures start, persist, and end.","The cluster-specific residue patterns (such as lysine or arginine at position +2 in cluster 1 and aspartate or glutamate at position +2 in cluster 2) give sequence rules for stabilizing each topology.","If the set is complete as claimed, local backbone prediction reduces to choosing among 54 states instead of searching over continuous dihedral angles.","The 54 types include both secondary-structure-maintaining blocks and transition blocks linking one secondary element to another, so the library covers direction changes of the chain."],"supporting_citations":[{"why":"shows that short peptide fragments adopt specific folded conformations in a confined cavity, motivating the search for stable pentapeptides.","marker":"[1]"},{"why":"defines a five-residue structural alphabet of local backbone shapes that frames the use of pentapeptides as structural units.","marker":"[4]"},{"why":"supplies the entropy-based rationale that local sequence-structure relations allow a minimal set of fragments to represent protein structure.","marker":"[6]"},{"why":"provides the AMBER force field used in the molecular dynamics simulations of all pentapeptides.","marker":"[17]"},{"why":"documents the OPLS force field combined with AMBER for the simulations.","marker":"[18]"},{"why":"justifies molecular dynamics of short peptides without explicit solvent, the protocol adopted in this study.","marker":"[19]"},{"why":"supplies the UPGMA clustering method used to split trajectories into conformational clusters.","marker":"[21]"},{"why":"provides the idealized dihedral angles for alpha-helix and beta-structures used to classify the terminal regions of each cluster.","marker":"[22]"}],"fun_headline_variants":["54 stable pentapeptides form a complete toolkit for protein folds","Minimal set of 54 peptide blocks can shape any protein structure","A library of 54 stable pentapeptides covers all protein backbone shapes","54 stable peptide topologies: the alphabet for protein design","Small set of 54 stable peptides may encode all protein folds"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The completeness claim rests on the assumption that replacing only three of the five alanine positions is enough to generate every topologically distinct stable pentapeptide shape; the paper states this assumption and says a proof will follow, but the proof is not given.","fun_headline_variants_meta":{"raw":{"variants":["54 stable pentapeptides form a complete toolkit for protein folds","Minimal set of 54 peptide blocks can shape any protein structure","A library of 54 stable pentapeptides covers all protein backbone shapes","54 stable peptide topologies: the alphabet for protein design","Small set of 54 stable peptides may encode all protein folds"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000177,"raw_usage":{"total_tokens":1321,"prompt_tokens":1003,"completion_tokens":318,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":619,"completion_tokens_details":{"reasoning_tokens":230}},"tokens_in":619,"tokens_out":318,"duration_ms":3392,"temperature":1.0,"reasoning_tokens":230,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:22:38.111506+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Enumerate pentapeptides with four or five substituted positions beyond the alanine matrix, up to the full 3.2 million sequence space, cluster their molecular dynamics trajectories with the same 80% stability threshold, and look for any backbone topology outside the 54; one new stable cluster would refute the completeness claim. A weaker but still decisive check is to rerun the same 44,860 simulations with explicit solvent and see whether the stable set changes.","supporting_citations":[{"cited_title":"It is believed that interacting with proteins, peptides take a certain conformation","cited_arxiv_id":null,"evidence_quote":"shows that short peptide fragments adopt specific folded conformations in a confined cavity, motivating the search for stable pentapeptides."},{"cited_title":"Magnetic ordering in mackinawite (tetragonal FeS): evidence for strong itinerant spin fluctuations","cited_arxiv_id":"1102.2002","evidence_quote":"defines a five-residue structural alphabet of local backbone shapes that frames the use of pentapeptides as structural units."},{"cited_title":"Case, T.E","cited_arxiv_id":null,"evidence_quote":"provides the AMBER force field used in the molecular dynamics simulations of all pentapeptides."}],"review_version":1}