A complex peptide traceable structure encoding method

Abstract

This invention discloses a method for encoding the traceable structure of complex peptides, comprising: parsing the original input data to construct an intermediate representation object independent of the input format; mapping the intermediate representation object based on a rule base to construct a standardized molecular graph; serializing the standardized molecular graph to generate a standardized structural string representation; and generating a traceable structure encoding result aligned with the standardized structural string representation based on character position intervals, node attributes, edge attributes of the standardized molecular graph, and the intermediate representation object. By constructing an intermediate representation object independent of the input format, the method solves the problem of inconsistent parsing caused by differences in input formats in traditional methods. The obtained encoding result can be directly used for input of artificial intelligence models, structure screening, source auditing, and data archiving, improving the computability, traceability, and reusability of complex peptide data.

Description

Technical Field

[0001] This invention relates to the field of structural data processing technology for chemical information, and in particular to a method for encoding the traceable structure of complex peptides. Background Technology

[0002] Complex peptide molecules typically exhibit one or more structural features, such as chiral differences, extended residues, terminal modifications, side chain modifications, head-to-tail cyclization, side chain-to-side chain linkages, or disulfide bond linkages. Compared to standard natural peptides, these molecules are more complex in terms of input form, structural constraints, and source information, and are widely used in drug design, sequence management, structural archiving, and automated processing scenarios.

[0003] Existing complex peptide data typically originates from various heterogeneous input formats, such as linear sequence descriptions, FASTA files, sequences with extended residue labels, sequence descriptions with modification annotations, and monomer-linkage description data. These different sources exhibit variations in residue nomenclature, chiral representation, linkage expression, and modification representation, leading to potentially different representations of the same complex peptide across different data sources. This hinders standardized parsing, structural archiving, and data reuse.

[0004] Existing technologies include hierarchical representation schemes for macromolecules or modified peptides, such as HELM (Hierarchical Editing Language for Macromolecules), which describes macromolecules containing non-natural monomers, complex linkages, and modification information. There are also sequence representation schemes for general molecular diagrams, such as SMILES (Simplified Molecular Linear Input Specification), which convert molecular diagrams into linear string representations. However, these schemes typically lack integrated processing for unified parsing of complex peptide multi-source inputs, source preservation during structural string generation, and traceable encoding outputs that can be reverse-directed to original residue identifiers, linkage relationships, and modification origins.

[0005] On the other hand, the following problems still exist in the process of generating structure codes for complex peptides: First, the existing process lacks a unified mechanism that can simultaneously handle mixed inputs of extended residue identifiers, chiral differences, modification markers, and linkage constraints; Second, during the representation conversion process, chiral information, cross-residue linkage information, modification information, and extended residue source information are prone to being missing or inconsistent; Third, the existing output usually only provides a structural string representation, which is difficult to directly support reverse localization, structure retrieval, structure screening, input of artificial intelligence models, and automated processing. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for encoding the traceable structure of complex peptides, which realizes unified parsing of multi-source inputs, standardized processing of complex structures, and aligned output of structural strings and source information.

[0007] The objective of this invention is achieved through the following technical solution:

[0008] This application discloses a method for encoding the traceable structure of complex peptides, comprising: acquiring raw input data of the complex peptide to be processed, wherein the raw input data is one of multiple heterogeneous input formats; parsing the raw input data to construct an intermediate representation object independent of the input format, wherein the intermediate representation object includes at least the original residue identifiers, chiral information, linkage information, and modification information parsed from the raw input data; and mapping the intermediate representation object based on a rule base to construct a standardized molecular graph, wherein the rule base includes standard residue mapping rules and extended residue mapping rules, wherein the chiral information, linkage information, and modification information are retained in the mapping process, and the portion processed by the extended residue mapping rules also retains its chiral information, linkage information, and modification information. The standardization process involves: serializing the standardized molecular graph to generate a standardized structural string representation; synchronously recording the character position intervals in the standardized structural string corresponding to the node attributes and edge attributes of the molecular graph; generating a traceable structural encoding result aligned with the standardized structural string representation based on the recorded character position intervals and the correspondence between the node attributes, edge attributes, and the original residue identifiers, chiral information, connection relationship information, and modification information in the intermediate representation object; and including the index relationship between the character position intervals of the standardized structural string and the original residue identifiers, connection relationship information, and modification information.

[0009] The benefits are as follows: By constructing an intermediate representation object independent of the input format, it normalizes multi-source heterogeneous complex peptide inputs such as linear sequences, FASTA format, cyclization constraints, and modification annotations, solving the problem of inconsistent parsing caused by differences in input formats in traditional methods. Simultaneously, during the construction and sequencing of the standardized molecular map, it synchronously records the mapping relationship between character position intervals and original residue identifiers, chiral information, linkage relationships, and modification information, generating traceable structural encoding results that are perfectly aligned with the standardized SMILES strings. This avoids the loss of key structural information during format conversion and achieves precise directional traceability from the standardized representation to the original input. The resulting encoding results can be directly used for input to artificial intelligence models, structure screening, source auditing, and data archiving, improving the computability, traceability, and reusability of complex peptide data. Attached Figure Description

[0010] Figure 1 This is a schematic diagram of the traceable structure encoding generation method provided in an embodiment of the present invention.

[0011] Figure 2 This is a schematic diagram of the intermediate representation object structure provided in an embodiment of the present invention.

[0012] Figure 3 This is a schematic diagram illustrating the rule base mapping and standardized structure encoding generation provided in an embodiment of the present invention.

[0013] Figure 4 This is a schematic diagram illustrating the correspondence between the standardized SMILES string and the traceable structure encoding result, using Example 1 as an example, provided as an embodiment of the present invention. Detailed Implementation

[0014] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] Since the embodiments of this application involve a large number of technical terms, some technical terms will be explained before the description:

[0016] FASTA format data is a text-based biological sequence representation format, typically consisting of an identifier line beginning with the character ">" followed by a sequence line; in this application, FASTA format data is mainly used to represent the residue sequences of polypeptides or complex peptides and their constraint markers.

[0017] Original residue identifiers refer to residue names, single-letter codes, three-letter codes, or extended residue codes that are directly read and retained from the original input data, such as A, G, D-Ala, X001, etc.; their meaning focuses on "input source identifiers", and are not limited to natural residues, nor are they equivalent to residue numbers in PDB files.

[0018] Extended residues refer to residues or residue identifiers that are outside the basic standard residue set and need to be interpreted through an extended rule base. These include non-natural amino acids, residues with special substituents, artificially defined residues, and other residues that need to be mapped to structural units by template identifiers.

[0019] Cross-residue linkages refer to the linkages formed between two or more different residues, including peptide bond linkages between adjacent residues, head-to-tail cyclization linkages, side-chain-to-side-chain linkages, and disulfide bond linkages.

[0020] Cycling constraints refer to the constraints that allow a linear peptide chain to form a cyclic structure by connecting two residues at different positions in the sequence; for example, head-to-tail cyclic constraints are used to indicate that a head-to-tail connection is formed between the terminal residue and the head residue.

[0021] Monomer-linkage description refers to a structural description method centered on monomer units and the connection methods between monomers; in this application, it can be used to describe the connection relationships between residues, modifying groups, or extended structural units.

[0022] Terminal modification refers to the chemical modification of the terminal groups of a polypeptide molecule chain, mainly including N-terminal modification and C-terminal modification; where N-terminus refers to the amino terminus and C-terminus refers to the carboxyl terminus, examples include N-terminal acetylation and C-terminal amidation.

[0023] Side chain modification refers to the chemical modification or labeling of the side chain groups of amino acid residues, such as introducing modifying groups at the amino group of lysine, the thiol group of cysteine, or other reactive side chain positions.

[0024] SMILES, short for Simplified Molecular Input Line Entry System, is a structural string that uses characters to linearly represent molecular structures. RDKit, a cheminformatics toolkit, is used in this application embodiment to perform validity parsing, normalization, and structural information verification on SMILES strings.

[0025] MolFromSmiles parsing refers to the parsing process of constructing molecular objects based on SMILES strings; CanonSmiles normalization refers to converting parsable SMILES strings into uniform normalized SMILES strings.

[0026] FindMolChiralCenters refers to the process of finding chiral centers in a molecule. The results returned typically include atomic indices and R / S configuration labels.

[0027] RingInfo_NumRings refers to the verification result of the number of rings in the molecular ring information.

[0028] CIP refers to the Cahn-Ingold-Prelog rule used to label chiral configurations R / S.

[0029] Lys, Phe, Gly, Asp, and Ala represent lysine, phenylalanine, glycine, aspartic acid, and alanine, respectively; K, F, G, D, and A are the commonly used single-letter codes for these residues. HT constraint represents head-to-tail looping constraint, which establishes a connection between the terminal residue and the head residue to form a closed loop.

[0030] According to the complex peptide traceability structure encoding method described in the embodiments of this application, refer to... Figure 1 The method, which takes FASTA single-letter sequences, extended residue identifier sequences, and connection constraint / modification markers as input, and outputs normalized structural string representations and traceable structural encoding results, includes the following steps:

[0031] S1: Acquisition of raw input data.

[0032] The raw input data of the complex peptide to be processed is obtained. The raw input data is one of a variety of heterogeneous input formats, specifically including at least one of the following: linear sequence data, FASTA format data, sequence data with extended residue labels, sequence data with cyclization constraints, sequence data with modification annotations, and monomer-linkage description data.

[0033] S2: Unified parsing and intermediate representation object construction.

[0034] The original input data is parsed to construct an intermediate representation object that is independent of the input format, and then combined with... Figure 2 Understood. The intermediate representation object includes at least the original residue identifiers, chiral information, linkage information, and modification information parsed from the original input data. In one embodiment, the intermediate representation object further includes one or more of the following: circularization relationship information, end-group modification information, side-chain modification information, template identifier information, mapping state information, and checksum information.

[0035] In other words, the intermediate representation object refers to a structured intermediate representation, independent of the original input format, constructed after parsing raw input data of any heterogeneous input format. This object explicitly transforms the various structural semantics implicit in the original input into several standardized fields, such as... Figure 2 It includes at least: a raw residue identifier list, used to record residue identifiers appearing in the input, standard residue / extended residue type information of the mapping, sequential linkage information, circular linkage information, chiral information, and linkage relationship information; modification information, used to record modification states such as end-group modifications and side-chain modifications; template identifier information, character position interval information, sequence traversal mapping information (used to represent mapping relationships), and multiple rules source information. Thus, by constructing the above intermediate representation object, complex peptide input data from different sources and in different formats are unified into the same structured object model. This intermediate representation object serves as the data foundation for structure generation and encoding output, enabling subsequent rule mapping, molecular graph construction, and sequence processing to proceed with a unified workflow, without the need to design separate processing logic for each input format.

[0036] In this embodiment, complex modified peptides and extended residue sequences refer to peptide inputs that include at least one of the following features: chiral differences, extended residue identifiers, terminal modifications, side chain modifications, head-to-tail linkage constraints, side chain-to-side chain linkages, or disulfide bond linkages; in this embodiment, intermediate representation objects refer to structured representation objects whose parsed input information is used to characterize the original residue identifiers, mapped structural units, chiral states, linkage relationships, modification information, and rule sources.

[0037] S3: Rule mapping and construction of standardized molecular graphs.

[0038] refer to Figure 3 Understanding is that the intermediate representation object is mapped based on a rule base to construct a standardized molecular graph. The rule base includes standard residue mapping rules and extended residue mapping rules. In the mapping process, chiral information, connectivity information, and modification information are preserved. Furthermore, the portion processed by the extended residue mapping rules also retains the source information of the rules it is based on. Specifically, the basic rule base and the extended rule base jointly participate in rule matching and mapping, standardized molecular graph construction, and structural string serialization output, simultaneously recording character position ranges and establishing serialization traversal mapping relationships.

[0039] Specifically, the rule base includes a basic rule base and an extended rule base. The basic rule base includes at least standard residue mapping rules, chiral mapping rules, linkage mapping rules, and modification mapping rules; the extended rule base includes at least extended mapping terms, template identifiers, and mapped structural units. The mapping process retains at least chiral information, linkage information, modification information, and extended residue source information. The extended residue mapping rules include extended mapping terms, template identifiers, and mapped structural units; during mapping processing, the extended mapping terms are matched according to the extended residue identifiers in the original input to obtain the corresponding template identifiers and mapped structural units, and the template identifiers and mapped structural units are retained in the subsequently generated traceable structure encoding results.

[0040] The chirality mapping rules are configured through a chirality mapping rule example table. For example, as shown in Table 1, when the input residue is Ala (alanine) and the chirality input is L, a structural representation is generated according to the standard residue template, and the output chirality code Chirality=L is output, where Chirality represents chirality; when the input residue is Ala and the chirality input is D, the chirality label is retained and a corresponding structural representation is generated, and the output chirality code Chirality=D is output; when the input residue is Gly (glycine) and the chirality input is none, a structural representation is generated according to the standard residue template, and the output chirality code Chirality=None is output, where None indicates no chirality label.

[0041] Table 1. Example Table of Chiral Mapping Rules

[0042]

[0043] The connection mapping rules are configured through the example table of connection mapping rules. For example, as shown in Table 2, for peptide bond connections, residues i to j are sequentially connected, creating a connection edge in the normalized molecular graph and outputting the connection code Connection=peptide_bond, where Connection represents the connection relationship and peptide_bond represents the peptide bond; for head-tail ring connections, a closed-loop edge is created in the normalized molecular graph, outputting the connection code Connection=head_tail_ring, where head_tail_ring represents head-tail ringing; for side-chain-side-chain connections, a cross-residue connection edge is created in the normalized molecular graph, outputting the connection code Connection=side_chain_link; for disulfide bond connections, a cross-residue connection edge is created in the normalized molecular graph, outputting the connection code Connection=disulfide, where disulfide represents the disulfide bond.

[0044] Table 2 Example Table of Connection Relationship Mapping Rules

[0045]

[0046] Modification rules are configured through a modification rule example table. For example, as shown in Table 3, for the no-modification type, no additional modification template is introduced, and the output modification code Modification=none, where Modification indicates the modification state and none indicates no modification; for the N-terminal acetylation modification type, the input specification is Ac-, and the N-terminal modification template is loaded during normalization, outputting the modification code Modification=Ac, where Ac represents acetyl; for the C-terminal amidation modification type, the input specification is -NH2, and the C-terminal modification template is loaded during normalization, outputting the modification code Modification=NH2, where NH2 represents the amidation end group; for the side-chain modification type, the modification template is loaded at the corresponding residue side-chain position during normalization, outputting the modification code Modification=side_chain, where side_chain indicates side-chain modification.

[0047] Table 3 Example Table of Modification Rules

[0048]

[0049] When the intermediate representation object contains end-group modification information, the mapping process of the intermediate representation object based on the rule base includes: according to the end-group modification rules in the modification mapping rules, incorporating the modification nodes or edges corresponding to N-terminal acetylation modification and C-terminal amidation modification into the normalized molecular graph; recording the character position range corresponding to the modification nodes or edges in the normalized structural string during the serialization process, and outputting the corresponding modification code in the traceable structural encoding result.

[0050] In some implementations, for inputs containing N-terminal acetylation and / or C-terminal amidation annotations, the parsing phase writes the corresponding end-group modifications into the end-group modification field of the intermediate representation object; during the normalization phase, the N-terminal acetylation template and the C-terminal amidation template are loaded according to the modification mapping rules, and the corresponding modification nodes or edges are incorporated into the normalized molecular graph; during the structural string serialization process, the character position ranges corresponding to the modification nodes or edges are recorded, and the corresponding modification codes are output in the traceable structure encoding results. Through this method, the modification mapping rules can work synergistically with residue mapping, chiral mapping, and connection relationship mapping to preserve end-group modification information.

[0051] S4: Standardized structure string generation.

[0052] The standardized molecular graph is serialized to generate a standardized structural string representation. During the serialization process, the character position ranges corresponding to the molecular graph node attributes and edge attributes in the standardized structural string are recorded synchronously. The node attributes include at least the original residue identifier, the mapped structural unit, the chiral state, the modification state, and the rule source. The edge attributes include at least the sequential connection relationship, the circular connection relationship, the cross-residue connection relationship, and the closed loop marker information.

[0053] In one implementation, the standardized structural string is represented as a standardized SMILES string representation; when a closed-loop structure exists, the serialization rule further includes a closed-loop marker generation rule. When the intermediate representation object contains cyclic relationship information, the serialization rule for serializing the standardized molecular graph includes a closed-loop marker generation rule; during the serialization process, closed-loop markers are generated in the standardized structural string according to the closed-loop marker generation rule, and the character position range corresponding to the closed-loop markers is recorded to retain the cyclic connection source information in the traceable structure encoding result. The method further includes: after generating the standardized SMILES string, verifying the validity and normalization of the standardized SMILES string using cheminformatics tools.

[0054] S5: Generation of traceable structure coding results.

[0055] Based on the recorded character position ranges, and based on the correspondence between the node attributes, edge attributes, and the original residue identifiers, chiral information, connection relationship information, and modification information in the intermediate representation object of the standardized molecular graph, a traceable structural encoding result aligned with the standardized structural string representation is generated. The traceable structural encoding result includes the index relationship between the character position ranges of the standardized structural string and the original residue identifiers, connection relationship information, and modification information.

[0056] In this application, the traceable structure encoding result refers to the encoded output result that corresponds to the character segments or structural units in the standardized structure string and can be reverse-located back to the original input source. The traceable structure encoding result is output in at least one of the following forms: character position interval index table, residue type encoding table, chirality encoding table, connection relationship encoding table, modification encoding table, rule source encoding table, template identifier encoding table, and reverse mapping index table. The traceable structure encoding result is not independently inferred after the standardized structure string is generated, but is established synchronously during the serialization output process.

[0057] Specifically, during the serialization of the standardized molecular graph, the character position intervals corresponding to node attributes and edge attributes are recorded. Subsequently, based on the character position intervals and the mapping relationship between the nodes, edge attributes, and intermediate representation objects of the standardized molecular graph established during the traversal process, a traceable structure encoding result aligned with the standardized structure string is generated. The traceable structure encoding result includes a reverse mapping index table; the reverse mapping index table is used to reverse-locate one or more character position intervals in the standardized structure string representation to the corresponding original residue identifier in the intermediate representation object. When the original residue identifier is an extended residue identifier, the reverse mapping index table also associates the rule source identifier and template identifier on which the extended residue identifier is based in the extended residue mapping rule.

[0058] In other words, when the intermediate representation object is mapped based on the rule base to construct the standardized molecular graph, the node attributes of the standardized molecular graph are obtained by mapping the original residue identifiers, chiral information, and modification information in the intermediate representation object, thus there is a direct correspondence between the node attributes and the corresponding information in the intermediate representation object; the edge attributes of the standardized molecular graph are obtained by mapping the connection relationship information in the intermediate representation object, thus there is a direct correspondence between the edge attributes and the connection relationship information in the intermediate representation object.

[0059] Based on the subsequent embodiments, during the serialization of the standardized molecular graph to generate the standardized structural string representation, the character position intervals corresponding to the node attributes and edge attributes of the molecular graph in the standardized structural string are recorded simultaneously. Specifically, the character position interval corresponding to each node attribute and each edge attribute in the standardized structural string is recorded. Based on the correspondence between the character position intervals and the node and edge attributes of the standardized molecular graph, and the correspondence between the node and edge attributes of the standardized molecular graph and the original residue identifiers, chiral information, connectivity information, and modification information in the intermediate representation object, the corresponding node attribute or edge attribute can be determined from any character position interval using the character position interval as an index. This allows for the determination of the original residue identifiers, chiral information, connectivity information, or modification information corresponding to that node or edge attribute in the intermediate representation object, thereby generating a traceable structural encoding result that is completely aligned with the standardized structural string representation.

[0060] Therefore, the complex peptide traceable structure encoding method described above can be compatible with complex peptide inputs containing extended residues, chiral differences, modification markers, and linkage constraints through the same processing flow, improving the consistency of representation results; it can simultaneously retain key information such as chirality, modification, and cyclization during rule mapping and serialization, reducing the risk of losing key structural information; it can automatically output traceable structure encoding results aligned with standardized structural strings, improving the computability, traceability, and reusability of the results; and the obtained results can be directly used for artificial intelligence model input, structure screening, source auditing, and data archiving.

[0061] In specific embodiments of this application, the standardized structure string representation is described using the standardized SMILES string representation. Unless otherwise stated, the character position ranges involved in this specification and drawings all use a closed interval counting method starting from 1.

[0062] The technical solution of the present invention will be further described below with reference to specific embodiments.

[0063] Example 1: Traceable structure encoding of linear complex peptide sequences with chiral differences.

[0064] Combination Figure 4 The illustration shows the alignment between character position intervals and residue types, chirality, connectivity, and modification information. This embodiment is used to illustrate how this application performs unified parsing, rule mapping, and standardization on the original input data when processing linear complex peptide sequences with chiral differences, and generates a standardized structural string representation and a traceable structural encoding result aligned with it.

[0065] S1: Acquisition of raw input data. In this embodiment, the raw input data adopts the FASTA single-letter input syntax supported by the peptide sequence conversion tool, and the input is represented as aG, where lowercase a represents D-Ala and uppercase G represents Gly, corresponding to the semantics of dipeptide sequences containing D-Ala and Gly.

[0066] S2: Unified parsing and intermediate representation object construction. Specifically, the original input data is parsed to construct an intermediate representation object. In this embodiment, the parsing result includes at least: an original residue identifier list, chiral information, linkage information, and modification information. For example, the intermediate representation object can be represented as: residue 1 is Ala, chirality is D; residue 2 is Gly; the linkage relationship is that residue 1 and residue 2 are sequentially connected; and the modification information is none.

[0067] S3: Rule Mapping and Standardized Molecular Graph Construction. The intermediate representation objects are mapped based on a rule base. The rule base includes at least standard residue mapping rules, D / L chiral mapping rules, and linkage mapping rules. The mapping process includes: loading residue templates corresponding to Ala and Gly according to the standard residue mapping rules; preserving chiral information at the α-carbon position of Ala according to the D / L chiral mapping rules; and establishing peptide bond connections between residue 1 and residue 2 according to the linkage mapping rules, thereby constructing a standardized molecular graph.

[0068] S4: Generation of Standardized Structure Strings. In this embodiment, the generation of candidate structure strings is achieved by a publicly available peptide sequence to SMILES conversion tool. In this embodiment, p2smi (a tool for converting peptide sequences or constrained peptide inputs into SMILES format) is used as an example, and RDKit (a cheminformatics toolkit) is used for verification. After the input sequence aG is fed into the conversion tool, the candidate SMILES is obtained as the following string "N[C@H](C)C(=O)NCC(=O)O". Subsequently, RDKit is used to perform MolFromSmiles parsing and CanonSmiles normalization on the candidate result. MolFromSmiles parsing means parsing the SMILES string into a molecular object, and CanonSmiles normalization means generating a unified normalized SMILES string, resulting in the normalized SMILES string "C[C@@H](N)C(=O)NCC(=O)O". Chiral center identification is further performed using FindMolChiralCenters, returning a result of (1, R), where 1 represents the internal atomic index of RDKit, and R represents the R configuration under the CIP rule. This indicates that only one chiral center introduced by D-Ala exists in this embodiment, and the chiral information is preserved in the mapping process and serialization output. It should be noted that the internal atomic index of RDKit is not used as the basis for counting character position intervals in this application; the character position intervals in this application still uniformly adopt the closed interval counting method starting from 1.

[0069] S5: Generation of traceable structure encoding results. Based on the recorded character position range, and according to the correspondence between the node attributes, edge attributes, and the original residue identifiers, chiral information, connection relationship information, and modification information in the intermediate representation object of the standardized molecular graph, a traceable structure encoding result aligned with the standardized structure string representation is generated.

[0070] For example, the encoding result is established synchronously during the serialization process. Referring to Table 4, examples of the traceable structure encoding results are shown in Table 4.

[0071] Table 4. Example of traceable structure coding results in Example 1

[0072]

[0073] In the table above: Character position intervals 1-15 of sequence number 1 correspond to structural unit 1, with a residue type encoding and a value of D-Ala; character position intervals 4-5 of sequence number 2 correspond to the chiral descriptor of structural unit 1, with a chiral encoding and a value of D (corresponding to CIP=R in this embodiment); character position intervals 11-16 of sequence number 3 correspond to the connection between structural unit 1 and structural unit 2, with a connection type encoding and a value of peptide bond; character position intervals 16-23 of sequence number 4 correspond to structural unit 2, with a residue type encoding and a value of Gly; character position intervals 1-23 of sequence number 5 correspond to the whole molecule, with a modification state encoding and a value of none. The chiral-related character position intervals in Table 4 refer to the descriptor character positions carrying chiral information in the structural string, not the atomic coordinate positions.

[0074] For example, in Embodiment 1, the character position interval is recorded during SMILES serialization. The encoding result is generated based on the character position interval and the mapping relationship. It is generated based on the chiral encoding value D (corresponding to CIP=R in this embodiment) and the connection type is peptide bond. The corresponding overall molecular modification information is none.

[0075] Therefore, in conjunction with Embodiment 1, it can be understood that the embodiments of this application perform unified parsing and rule mapping on linear complex peptide sequences with chiral differences, and retain key information such as chirality, linkage and modification during the generation of standardized structural strings, while outputting traceable structural encoding results aligned with the structural representation.

[0076] Example 2: Traceable structure encoding of complex peptide sequences with head-to-tail connection constraints.

[0077] This embodiment illustrates how, when processing complex peptide sequences with start and end-end linkage constraints, this application retains the circular linkage information in the standardized structural string representation and generates a traceable structural encoding result aligned with it.

[0078] S1: Acquisition of raw input data. This embodiment uses the FASTA constraint syntax supported by p2smi. The input header is the character ">ex2|HT", and the input sequence is "KFGDA", where K, F, G, D, and A represent Lys (lysine), Phe (phenylalanine), Gly (glycine), Aspartic acid (aspartic acid), and Ala (alanine), respectively. The HT constraint represents head-tail circularized connections. The connection relationship information recorded during the construction phase includes linear connections 1-2, 2-3, 3-4, 4-5, and head-tail circularized connection 5-1.

[0079] S2: Unified parsing and intermediate representation object construction. After parsing the original input data, an intermediate representation object is constructed. In this example, the intermediate representation object includes residue sequence information, residue type information, sequential connection information, and head-to-tail circularization connection information.

[0080] S3: Rule Mapping and Standardized Molecular Graph Construction. Based on a rule base, the intermediate representation object is mapped. While establishing peptide bond connections between adjacent residues, closed loop edges are established in the standardized molecular graph according to the 5-1 connection relationships corresponding to the HT constraints.

[0081] S4: Generate a standardized structured string representation.

[0082] In this embodiment, candidate structure generation is achieved using a publicly available peptide sequence to SMILES conversion tool, specifically p2smi (version 1.1.1). The original output is: “KFGDA-HT: N2[C@@H](CCCCN)C(=O)N[C@@H](Cc1ccccc1)C(=O)NCC(=O)N[C@@H](CC(=O)O)C(=O)N[C@@H](C)C2(=O)”, where “KFGDA-HT:” represents the sequence identifier line information output by the p2smi tool, and HT indicates head-to-tail circularization constraints; the corresponding candidate SMILES string is:

[0083] "N2[C@@H](CCCCN)C(=O)N[C@@H](Cc1ccccc1)C(=O)NCC(=O)N[C@@H](CC(=O)O)C(=O)N[C@@H](C)C2(=O)".

[0084] Subsequently, RDKit (version 2025.9.6) was used to verify the candidate SMILES. The verification results showed that the MolFromSmiles parsing result was True, meaning the candidate SMILES could be parsed into valid molecule objects; the CanonSmiles normalization process was successful, meaning a unified normalized SMILES string could be obtained. The normalized SMILES string is as follows:

[0085] "C[C@@H]1NC(=O)[C@H](CC(=O)O)NC(=O)CNC(=O)[C@H](Cc2ccccc2)NC(=O)[C@H](CCCCN)NC1=O"

[0086] FindMolChiralCenters returns the following results for chiral center identification: (1, S), (10, S), (25, S), (33, S). The numbers in the returned results represent the internal atomic indices of RDKit, and S represents the S configuration under the CIP rule. RingInfo_NumRings returns 2, indicating that the number of rings obtained from the RDKit ring information verification is 2.

[0087] It should be noted that NumRings=2 indicates that the number of rings is 2, which includes both the aromatic ring in the phenylalanine side chain and the main chain closed ring formed by HT constraint. Therefore, this value is used as the topology verification result in this embodiment, while the source information of the first and last ringization is still directly based on the "HT corresponds to 5-1 connection" recorded in the construction stage.

[0088] S5: Traceable structure encoding result generation. In the traceable structure encoding result generation stage, the encoding result is generated based on the residue order, linear connection relationship, head-to-tail circular connection relationship recorded in the construction stage, and the RDKit verification result, rather than based on the ex-post inference of the final SMILES string.

[0089] When the normalized SMILES are serialized, the character position range of each structural unit in the string can be further recorded, and a correspondence between the character position range and the encoding result can be formed accordingly. This achieves the alignment preservation of the circular connection information in the normalized structural string representation and the traceable encoding layer. An example of the traceable structural encoding result is shown in Table 5.

[0090] Table 5. Example of traceable structure coding results in Example 2

[0091]

[0092] In Table 5: Character position intervals 1-14 of sequence number 1 correspond to the 5th structural unit of the input sequence, with residue type encoding value Ala (alanine); character position intervals 3-6 of sequence number 2 correspond to the 5th chiral descriptor of the input sequence, with chiral encoding value L (corresponding to the S configuration under the CIP rule in this embodiment); character position intervals 15-34 of sequence number 3 correspond to the 4th structural unit of the input sequence, with residue type encoding value Asp (aspartic acid); character position intervals 35-41 of sequence number 4 correspond to the 3rd structural unit of the input sequence, with residue type encoding value Gly (glycine); character position intervals 42-63 of sequence number 5 correspond to the input sequence... The second structural unit in the input sequence corresponds to the residue type code Phe (phenylalanine); the character position interval 64-80 in sequence number 6 corresponds to the structural unit at the first position in the input sequence, with the residue type code Lys (lysine); the character position intervals 8 and 78 in sequence number 7 correspond to the closed loop marker positions, with the linker source code HT corresponding to 5-1 links; the character position intervals 9-14, 29-34, 36-41, and 58-63 in sequence number 8 correspond to the main chain linker fragments, with the linker type code peptide bond; the character position intervals 1-80 in sequence number 9 correspond to the overall molecular topology, with the topology code NumRings=2, i.e., the number of rings is 2. The character position intervals in Table 5 are based on the normalized SMILES string and uniformly adopt the closed interval counting method starting from 1; the corresponding structural units are still identified according to the residue sequence numbers in the original input sequence.

[0093] As demonstrated in Example 2, this application can perform unified parsing and rule mapping on complex peptide sequences with head-tail linkage constraints, and retain the head-tail circular linkage information in the standardized structural string representation and traceable structural encoding results, thereby providing calculable input for subsequent structure retrieval, structure screening, source auditing and automated processing.

[0094] Example 3: Traceable structure encoding of complex peptide sequences containing extended residue markers

[0095] This embodiment illustrates how, when processing extended residue identifiers not pre-included in the basic rule base, the present invention completes mapping processing based on the extended rule base, and while generating a standardized structural string representation, retains the original extended residue identifier and its mapping source information, thereby generating a traceable structural encoding result. It should be noted that in this embodiment, the standardized structural string representation can be generated based on the mapped resolvable structural units, but the original extended residue identifier, mapping source, template identifier, and mapped structural units should be explicitly retained in the traceability coding layer and not lost due to structural mapping.

[0096] S1: Acquisition of raw input data. In this embodiment, the raw input is defined as the characters "X001-GA", using a residue identifier sequence syntax separated by connectors, where X001 represents an extended residue identifier, G represents Gly (glycine), and A represents L-Ala (L-alanine). After parsing the raw input, an intermediate representation object is constructed. The intermediate representation object includes at least the raw residue identifier list, residue order information, adjacent connection relationship information, and rule source identifier information.

[0097] S2: Rule Base Mapping and Standardization. In the rule base mapping stage, the basic rule base and the extended rule base are queried first. For standard residues G and A, they are mapped to Glycine and L-Alanine, respectively, according to the basic rule base. For the extended residue identifier X001, it is mapped to the template item TPL_TLEU_V1 according to the extended mapping item in the extended rule base. TPL_TLEU_V1 is a pre-configured tert-leucine template identifier in the extended rule base, and the corresponding mapped structural unit is Tert-leucine.

[0098] This forms a mapping link from the original input identifier sequence to the rule base mapping result to the standard structural unit sequence. This embodiment demonstrates that X001 is not directly identified by external tools, but is converted into a stable structural unit through rule base mapping, while retaining X001 itself and its mapping source in the traceability coding layer.

[0099] S3: Generation of Standardized Structure String Representation. After the mapping is completed, candidate structure strings are generated based on the mapped structural unit sequences. In this embodiment, the generation of candidate structure strings is achieved by a publicly available peptide sequence to SMILES conversion tool. Taking the linear peptide generation module of p2smi as an example, the generated candidate SMILES strings are as follows:

[0100] The candidate SMILES strings are as follows:

[0101] N[C@@H](C(C)(C)C)C(=O)NCC(=O)N[C@@H](C)C(=O)O.

[0102] Subsequently, the candidate SMILES were validated and normalized, resulting in the following normalized SMILES string:

[0103] C[C@H](NC(=O)CNC(=O)[C@@H](N)C(C)(C)C)C(=O)O.

[0104] The verification results show that the candidate structure can be successfully parsed and a normalized string representation can be generated, and the chiral center identification result is consistent with the composition of the mapped structural units.

[0105] S4: Generation of Traceable Structure Encoding Results. In the stage of generating traceable structure encoding results, the encoding results are not based on the post-hoc inference of the final SMILES string, but are generated jointly based on the mapping link records and the character position ranges recorded during the standardized SMILES serialization process.

[0106] Specifically, during the process of generating the standardized SMILES string representation, the character position range corresponding to each structural unit is recorded; subsequently, based on the correspondence between the character position range and the original residue identifier, rule source, template identifier and mapped structural unit in the intermediate representation object, a traceable structural encoding result aligned with the standardized structural string representation is generated.

[0107] Preferably, it also includes a reverse positioning index, enabling any SMILES fragment to be reverse-positioned to the corresponding original extended residue identifier. Examples of the traceable structure encoding results are shown in Table 6.

[0108] Table 6. Example of traceable structure coding results in Example 3

[0109]

[0110] In Table 6: Character position interval 1-22 of sequence number 1 corresponds to the first structural unit of the input sequence, with the encoding type being the original residue identifier and the encoding value being X001; character position interval 1-22 of sequence number 2 corresponds to the first structural unit of the input sequence, with the encoding type being the rule source and the encoding value being extension_rule_library; character position interval 1-22 of sequence number 3 corresponds to the first structural unit of the input sequence, with the encoding type being the template identifier and the encoding value being TPL_TLEU_V1 (tert-leucine template identifier); sequence number 4... The character position interval 1-22 corresponds to the first structural unit of the input sequence, the encoding type is mapped structural unit, and the encoding value is Tert-leucine; the character position interval 23-29 of sequence number 5 corresponds to the second structural unit of the input sequence, the encoding type is original residue identifier, and the encoding value is G; the character position interval 23-29 of sequence number 6 corresponds to the second structural unit of the input sequence, the encoding type is rule source, and the encoding value is basic_rule_library; the character position interval 23-29 of sequence number 7 corresponds to... The second structural unit corresponding to the input sequence should be encoded as a mapped structural unit with the value "Glycine". The character position range 30-45 of sequence number 8 corresponds to the third structural unit of the input sequence, encoded as a raw residue identifier with the value "A". The character position range 30-45 of sequence number 9 corresponds to the third structural unit of the input sequence, encoded as a rule source with the value "basic_rule_library". The character position range 30-45 of sequence number 10 corresponds to the third structural unit of the input sequence... The type is a mapped structural unit, and the encoding value is L-Alanine (L-type alanine); the character position range 18-24 of sequence number 11 corresponds to the connection between structural unit 1 and structural unit 2, the encoding type is connection type, and the encoding value is peptide bond; the character position range 26-31 of sequence number 12 corresponds to the connection between structural unit 2 and structural unit 3, the encoding type is connection type, and the encoding value is peptide bond; the character position range 1-22 to X001 of sequence number 13 corresponds to the reverse positioning index, the encoding type is reverse mapping, and the encoding value is that it can be reverse-positioned to X001 by the SMILES fragment.

[0111] As can be seen from this embodiment, the present invention can map extended residues to resolvable structural units based on an extended rule base without relying on external tools to directly parse extended residue identifiers. While outputting standardized structural strings, it retains traceability information such as the original extended residue identifiers, rule sources, template identifiers, and mapped structural units, thereby improving the computability, traceability, and reusability of samples containing extended residues. In some embodiments, the results can also be directly used for subsequent structural screening and input to artificial intelligence models.

[0112] It should be noted that this invention is not limited to the specific residue types, linkage types, modification types, or extended mapping templates listed in the above embodiments. Those skilled in the art can adjust the rule base content, intermediate representation object fields, standardized molecular graph construction methods, serialization rules, and encoding output formats according to actual application needs without departing from the inventive concept. These equivalent substitutions or conventional modifications should all be considered to fall within the protection scope of this invention.

[0113] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the concept described herein through the above teachings or related technologies or knowledge. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. A method for encoding the traceable structure of complex peptides, characterized in that, include: Obtain the raw input data of the complex peptide to be processed, wherein the raw input data is one of a variety of heterogeneous input formats; The original input data is parsed to construct an intermediate representation object that is independent of the input format. The intermediate representation object includes at least the original residue identifier, chiral information, connection relationship information and modification information parsed from the original input data. Based on the rule base, the intermediate representation object is mapped to construct a standardized molecular graph. The rule base includes standard residue mapping rules and extended residue mapping rules. In the mapping process, the chiral information, connection relationship information and modification information are retained. The part processed by the extended residue mapping rules also retains the rule source information on which it is based. The standardized molecular graph is serialized to generate a standardized structural string representation. During the serialization process, the character position ranges corresponding to the node attributes and edge attributes of the molecular graph are recorded synchronously in the standardized structural string. The node attributes include: original residue identifier, mapped structural unit, chiral state, modification state, and rule source. The edge attributes include: sequential connection relationship, circular connection relationship, cross-residue connection relationship, and closed-loop marker information. Based on the recorded character position ranges, and based on the correspondence between the node attributes and edge attributes of the standardized molecular graph and the original residue identifiers, chiral information, connection relationship information, and modification information in the intermediate representation object, a traceable structural encoding result aligned with the standardized structural string representation is generated using the character position ranges as indexes; wherein, the traceable structural encoding result includes the index relationship between the character position ranges of the standardized structural string and the original residue identifiers, connection relationship information, and modification information.

2. The method for encoding the traceable structure of complex peptides according to claim 1, characterized in that, The intermediate representation object also includes one or more of the following: circular relationship information, end-base modification information, side-chain modification information, template identification information, mapping status information, and verification mark information.

3. The method for encoding the traceable structure of complex peptides according to claim 2, characterized in that, When the intermediate representation object contains the looping relationship information, the serialization rule for serializing the standardized molecular graph includes the loop closure marker generation rule; During the serialization process, a closed-loop marker is generated in the standardized structure string according to the closed-loop marker generation rule, and the character position range corresponding to the closed-loop marker is recorded so as to retain the source information of the loop connection in the traceable structure encoding result.

4. The method for encoding the traceable structure of complex peptides according to claim 2, characterized in that, When the intermediate representation object contains the end-base modification information, the mapping process of the intermediate representation object based on the rule base includes: According to the end-group modification rules in the modification mapping rules, the modification nodes or edges corresponding to N-terminal acetylation modification and C-terminal amidation modification are incorporated into the normalized molecular graph; during the serialization process, the character position range corresponding to the modification nodes or edges in the normalized structural string is recorded, and the corresponding modification codes are output in the traceable structural encoding results.

5. The method for encoding the traceable structure of complex peptides according to claim 1, characterized in that, The extended residue mapping rule includes extended mapping terms, template identifiers, and mapped structural units; During the mapping process, the extended mapping item is matched according to the extended residue identifier in the original input to obtain the corresponding template identifier and the mapped structural unit, and the template identifier and the mapped structural unit are retained in the traceable structure encoding result.

6. The method for encoding the traceable structure of complex peptides according to claim 1, characterized in that, The traceable structure encoding result is output in at least one of the following forms: character position interval index table, residue type encoding table, chirality encoding table, connection relationship encoding table, modification encoding table, rule source encoding table, template identifier encoding table, and reverse mapping index table.

7. The method for encoding the traceable structure of complex peptides according to claim 6, characterized in that, The traceable structure encoding result includes the reverse mapping index table; The reverse mapping index table is used to reverse the location of one or more character position ranges in the standardized structure string representation to the corresponding original residue identifier in the intermediate representation object.

8. The method for encoding the traceable structure of complex peptides according to claim 7, characterized in that, When the original residue identifier is an extended residue identifier, the reverse mapping index table is also associated with the rule source identifier and template identifier on which the extended residue identifier is based in the extended residue mapping rule.

9. The method for encoding the traceable structure of complex peptides according to claim 1, characterized in that, The character position ranges corresponding to the molecular graph node attributes and edge attributes in the standardized structure string are recorded and established synchronously during the serialization traversal process.

10. The method for encoding the traceable structure of complex peptides according to claim 1, characterized in that, The standardized structure string is represented as a standardized SMILES string; the method further includes: after generating the standardized SMILES string, verifying the validity and standardization of the standardized SMILES string using cheminformatics tools.

Images