Design method and system of cross-species universal paramyxovirus RdRp

Abstract

The invention belongs to the field of computational biology and protein engineering, and discloses a cross-species universal paramyxovirus RdRp design method and system, the design system fuses multi-level information such as sequences, protein language characterization and structures, an RdRp candidate sequence with a cross-species function is mined through a multi-mode and multi-model cooperation strategy, and the cross-species universal paramyxovirus RdRp is obtained. The method mainly comprises a candidate mining stage, a preliminary design and wet experiment verification stage and a structure optimization stage, BLAST and other methods are used for sequence similarity analysis, deep representation of a protein sequence is extracted in combination with ESMC and other protein language models, three-dimensional structure prediction is carried out with the assistance of Alphafold3, and the three-dimensional structure is obtained. RdRp candidate molecules with a potential cross-species universality function are screened out by combining multi-dimensional similarity indexes, after preliminary design, structural correction and optimization are performed by means of a protein generation model such as RFdiffusion, and the molecular stability, universality and target binding capacity of the RdRp candidate molecules are emphatically improved.

Description

Technical Field

[0001] The present invention relates to the fields of computational biology and protein engineering, and is directly related to vaccines and public health. Specifically, it relates to a design method and system for a cross-species universal paramyxovirus RdRp. Background Art

[0002] Viruses in the Paramyxoviridae family contain a variety of important human and animal pathogens, such as measles virus, mumps virus, respiratory syncytial virus, etc. These viruses pose a major threat to global public health. The RNA-dependent RNA polymerase (RdRp) of paramyxoviruses is a key enzyme in the virus replication process, responsible for the replication and transcription of the viral genome, and is a core component in the virus life cycle. Developing a universal RdRp can more efficiently achieve species and cross-species virus recognition, which is of great significance in aspects such as detection and virus prevention and control.

[0003] Traditional research on viral RdRp has mainly been limited to comparisons between single virus species or closely related viruses, lacking a systematic exploration of common characteristics among distantly related viruses. Existing methods for mining universal RdRp mainly rely on extensive wet-lab screening and attempts. The research often does not delve deep into the molecular and mechanistic levels. The sequence similarity differences among paramyxovirus RdRps are relatively large, making it difficult for traditional homology modeling-based methods to achieve cross-species design.

[0004] Traditional protein design methods mainly rely on manual discrimination in the initial screening, and then further screening mainly relies on physics-based molecular dynamics simulations or evolutionary analysis based on sequence-drawn phylogenetic trees. The computational cost is high and the accuracy is limited. For example, when using molecular dynamics simulations for optimization, the simulation of a single candidate structure takes 3 - 5 days, and the experimental verification success rate is only about 20%. In addition, after the screening and wet-lab verification are completed, traditional methods often have no subsequent optimization process, or only rely on site-directed mutagenesis or random mutagenesis library screening, which is time-consuming, laborious, and has limited optimization space.

[0005] With the development of artificial intelligence technology, especially the breakthroughs in protein language models such as the ESM series of models and protein structure prediction models such as AlphaFold3, new opportunities have been brought to the field of protein design. However, there is currently a lack of a systematic method to integrate and apply these advanced technologies to the design and optimization of cross-species universal RdRp. Summary of the Invention

[0006] The object of the present invention is to provide a design method and system for a cross-species universal paramyxovirus RdRp. This system integrates multi-level information and uses a multi-modal and multi-model collaborative strategy to mine and design RdRp candidate sequences with cross-species functions, providing a key tool for protein design and subsequent fields such as virus detection, vaccine development, and antiviral drug screening.

[0007] The object of the present invention can be achieved through the following technical solutions:

[0008] A design system for a cross-species universal paramyxovirus RdRp, which includes a candidate mining stage, a preliminary design and wet experiment verification stage, and a structure optimization stage;

[0009] Specifically:

[0010] The candidate mining stage includes the following steps:

[0011] Sequence similarity analysis: Using the BLAST sequence alignment tool, screen RdRp sequences related to the target function from the database, and set the threshold parameters E-value < 1e-5 and sequence coverage > 70%;

[0012] Characterization similarity extraction: Use the pre-trained protein language model ESMC to perform deep characterization extraction on the screened sequences, obtain the 1168-dimensional characterization vector output by the model, and take the average or sequence characterization based on amino acid characterization;

[0013] Calculate the Euclidean distance or cosine similarity between sequences, and select sequences with similarity > 0.85 or Euclidean distance < 0.15;

[0014] Structure similarity analysis: Use Alphafold3 to perform three-dimensional structure prediction on the candidate sequences, require the average pLDDT score > 80, and perform structure similarity alignment through Foldseek, and select structures with TM-score > 0.7;

[0015] Multi-dimensional comprehensive scoring: Integrate sequence similarity, characterization similarity, and structure similarity, construct a scoring function Score = 0.3×Seq_sim + 0.3×Rep_sim + 0.4×Str_sim, and select candidate sequences with scores > 0.8 to enter the next stage.

[0016] The preliminary design and wet experiment verification stage includes the following steps:

[0017] Functional region identification: Based on multiple sequence alignment and evolutionary analysis, identify the conserved functional domains and variable regions in RdRp;

[0018] Preliminary sequence design: According to the results of functional region analysis, design a preliminary RdRp sequence containing key conserved regions;

[0019] In vitro activity verification: Express and purify the designed RdRp protein and test its in vitro RNA synthesis activity;

[0020] Cross-species adaptability verification: Evaluate the recognition ability and replication efficiency of the designed RdRp for RNA templates of different virus species.

[0021] The structure optimization stage includes the following steps:

[0022] Diffusion model optimization: Use the RFdiffusion protein structure generation model to correct and optimize the structure of the preliminarily designed RdRp;

[0023] Stability assessment: Calculate the Rosetta energy score of the optimized structure, requiring an energy reduction > 10%;

[0024] Binding site optimization: Use molecular docking and hot spot residue analysis methods to optimize the binding interface between RdRp and the RNA template;

[0025] Sequence reverse design: Based on the optimized three-dimensional structure, use the LigandMPNN model for sequence reverse design to generate the final RdRp sequence.

[0026] Furthermore, during the sequence similarity analysis process, iterative searches are performed using different combinations of BLAST parameters, including blastp, psi-blast, and delta-blast, and phylogenetic tree analysis is used to ensure the diversity of candidate sequences.

[0027] Furthermore, during the characterization similarity extraction process, the ESMC model is first fine-tuned on RdRp-related sequences, and the fine-tuning parameters include a learning rate of 1e-5, 10 training epochs, and a batch size of 32.

[0028] Furthermore, during the structure similarity analysis process, in addition to using Alphafold3 to predict the three-dimensional structure of candidate sequences, RoseTTAFold and ESMFold are also used as complementary methods for structure prediction, and the consensus structure of the three prediction results is taken.

[0029] Furthermore, the weights in the multi-dimensional comprehensive scoring function can be dynamically adjusted according to specific application scenarios. For scenarios with higher requirements for cross-species recognition ability, the weight of characterization similarity can be increased to 0.4, while the weight of sequence similarity is reduced to 0.2.

[0030] Furthermore, during the functional region recognition process, in addition to conservative region analysis, it also includes analysis of protein surface hydrophobicity and secondary structure prediction to ensure that the designed RdRp has the correct folded structure.

[0031] Furthermore, during the structural optimization stage, in addition to using the RFdiffusion protein structure generation model, models such as TorchProteinLibrary and ProteinMPNN can also be selected for complementary optimization to synthesize the prediction results of multiple models.

[0032] A design method for a cross-species universal paramyxovirus RdRp includes the following steps:

[0033] Data collection and preprocessing: Collect paramyxovirus RdRp-related sequence data from the UniProt and PDB databases, perform sequence redundancy removal, and remove redundant sequences with sequence similarity > 95%;

[0034] Perform candidate mining on the obtained sequences to screen RdRp candidate sequences with potential cross-species functions;

[0035] Among them, candidate mining includes sequence similarity analysis, characterization similarity extraction, structure similarity analysis, and multi-dimensional comprehensive scoring;

[0036] Perform preliminary design on the candidate sequences. The design uses Cedar virus RdRp as a template, retains its key catalytic sites and RNA binding domains, and simultaneously integrates cross-species recognition elements of other paramyxovirus RdRps;

[0037] Synthesize and purify the preliminarily designed RdRp protein, and perform in vitro activity testing and cross-species adaptability verification;

[0038] Based on the results of wet experiments, use the RFdiffusion model to optimize the structure of the preliminarily designed RdRp;

[0039] Perform sequence reverse design on the optimized RdRp to generate the final cross-species universal RdRp sequence;

[0040] Finally, express and purify the optimized RdRp protein and perform comprehensive performance evaluation;

[0041] Comprehensive performance evaluation includes stability, activity, and cross-species adaptability verification.

[0042] Furthermore, during the data collection and preprocessing process, for sequence data, use the CD-HIT algorithm for redundancy removal; for structure data, use tools such as PDBeFold / Foldseek for structure clustering analysis.

[0043] Furthermore, during the preliminary design process, use the SCHEMA algorithm to assist in determining the recombination sites of cross-species recognition elements to ensure the stability of the protein structure after recombination.

[0044] Further, the in vitro activity of the RdRp was detected to examine its driving ability for different paramyxovirus-specific promoters. The detection indicators were that the product showed linear growth within 60 minutes, the activity of Mg 2+ (5 mM) was > 2-fold that of Mn 2+ (1 mM), and there was a linear relationship for 0.5 - 5 μg of the enzyme (slope > 200 cpm / μg).

[0045] Advantages of the present invention: The cross-species universal paramyxovirus RdRp of the present invention has the ability to recognize RNA across paramyxovirus family species, has a binding affinity of more than 75% for RNA templates of different types of paramyxoviruses, has good stability at 37°C, and a half-life exceeding 48 hours;

[0046] The present invention uses methods such as BLAST for sequence similarity analysis, combines protein language models such as ESMC to extract the deep characteristics of protein sequences, and is supplemented by Alphafold3 for three-dimensional structure prediction. Comprehensive multi-dimensional similarity indicators are used to screen RdRp candidate molecules with potential cross-species universal functions. After preliminary design, protein generation models such as RFdiffusion are used for structure correction and optimization, with a focus on improving its molecular stability, universality, and target binding ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The present invention will be further described below with reference to the accompanying drawings.

[0048] Figure 1 is a flowchart of a design method for a cross-species universal paramyxovirus RdRp of the present invention;

[0049] Figure 2 is a three-dimensional protein structure prediction diagram of the RdRp alignment region of the Cedar virus preliminarily screened by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0050] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0051] Example 1

[0052] Please refer to Figure 1 as shown. The present invention is a design method for a cross-species universal paramyxovirus RdRp. The design method for the cross-species universal paramyxovirus RdRp includes the following steps:

[0053] Collect paramyxovirus RdRp sequence data from the UniProt and PDB databases, and perform redundancy removal on the paramyxovirus RdRp sequence data;

[0054] Redundancy removal of paramyxovirus RdRp sequence data includes removing redundant sequences with sequence similarity > 95%.

[0055] Perform candidate mining on the redundancy-processed paramyxovirus RdRp sequence data to screen RdRp candidate sequences with potential cross-species functions;

[0056] Candidate mining includes the following steps:

[0057] Sequence similarity analysis: Use the BLAST sequence alignment tool to screen RdRp sequences related to the target function from the database, and set the threshold parameters E-value < 1e-5 and sequence coverage > 70%;

[0058] Characterization similarity extraction: Use the pre-trained protein language model ESMC to perform deep characterization extraction on the screened sequences, obtain the 1168-dimensional characterization vectors output by the model, and take the average or sequence characterization based on amino acid characterization of the sequences;

[0059] Among them, calculate the Euclidean distance or cosine similarity between sequences, and select sequences with similarity > 0.85 or Euclidean distance < 0.15;

[0060] Structural similarity analysis: Use Alphafold3 to perform three-dimensional structure prediction on the candidate sequences, make the average pLDDT score > 80, and perform structural similarity alignment through Foldseek, and select structures with TM-score > 0.7;

[0061] Multi-dimensional comprehensive scoring: Integrate sequence similarity, characterization similarity, and structural similarity to construct a scoring function Score = 0.3×Seq_sim + 0.3×Rep_sim + 0.4×Str_sim, and select candidate sequences with scores > 0.8.

[0062] Perform preliminary design on the candidate sequences. Using cedivirus RdRp as a template, retain its key catalytic sites and RNA binding domains, and at the same time fuse the cross-species recognition elements of other paramyxovirus RdRps;

[0063] Synthesize and purify the preliminarily designed RdRp protein, and perform in vitro activity testing and cross-species adaptability verification;

[0064] When performing preliminary design on the candidate sequences, functional region recognition is required. Based on multiple sequence alignment and analysis, identify the conserved functional regions and variable regions in paramyxovirus RdRp, and according to the results of functional region recognition, design the preliminary RdRp sequence containing the key conserved regions;

[0065] The in vitro activity verification is to express and purify the designed RdRp protein and test its in vitro RNA synthesis activity;

[0066] The cross-species adaptability verification is to evaluate the recognition ability and replication efficiency of the designed RdRp for RNA templates of different virus species.

[0067] Use the RFdiffusion model to optimize the structure of the preliminarily designed RdRp, perform sequence reverse design on the optimized RdRp, and generate the final cross-species universal RdRp sequence;

[0068] The structural optimization of the preliminarily designed RdRp includes stability evaluation;

[0069] Calculate the Rosetta energy score of the optimized structure to reduce the energy by > 10%.

[0070] The structural optimization of the preliminarily designed RdRp includes binding site optimization;

[0071] Use molecular docking and hot spot residue analysis methods to optimize the binding interface between RdRp and the RNA template;

[0072] Finally, express and purify the optimized RdRp protein and conduct comprehensive performance evaluation, which includes stability, activity, and cross-species adaptability verification.

[0073] Exemplary: For designing a cross-species universal RNA polymerase based on the RdRp of CedPV, such as Figure 2 shown:

[0074] Data collection and preprocessing

[0075] Collect all retrievable RdRp sequences and their paired RNA sequences of members of the Paramyxoviridae family from the NCBI database, including CedPV, measles virus, mumps virus, Hendra virus, Nipah virus, Sendai virus, respiratory syncytial virus, human parainfluenza virus type 1, avian paramyxovirus type 1, and bovine paramyxovirus type 3;

[0076] Use the CD-HIT algorithm (parameter setting: sequence similarity threshold 95%) for redundancy removal, and finally obtain 496 non-redundant sequences;

[0077] At the same time, collect relevant RdRp structures from the PDB database, and after PDBeFold clustering analysis (parameter setting: RMSD threshold ), obtain 1 representative structure, indicating that the protein structure accumulation in the current field is less;

[0078] Candidate mining stage

[0079] Sequence similarity analysis: Using the cedar virus RdRp as the reference sequence, PSI-BLAST was used for searching (parameter settings: E-value < 1e-5, number of iterations 3 times), and 42 related sequences were obtained;

[0080] Characterization similarity extraction: Using the ESMC model pre-trained on 280 million protein sequences (600M parameters, 1168-dimensional characterization), and performing Lora fine-tuning on the RdRp-related sequences (learning rate 1e-5, training for 10 rounds), extracting the 1168-dimensional characterization vectors of 42 sequences, calculating the cosine similarity with the cedar virus RdRp, and screening out 25 sequences with similarity > 0.85;

[0081] Structure similarity analysis: Using Alphafold3 to perform structure prediction on 25 sequences, requiring the average pLDDT score > 80, performing structure alignment through Foldseek, and selecting 18 sequences with TM-score > 0.7 with the cedar virus RdRp;

[0082] Multi-dimensional comprehensive scoring: Based on the formula Score = 0.3×Seq_sim + 0.3×Rep_sim + 0.4×Str_sim, calculate the comprehensive scores of 18 sequences, and select 6 sequences with scores > 0.8 as candidates.

[0083] Preliminary design and wet experiment verification

[0084] Functional region identification: Perform multiple sequence alignment on 6 candidate sequences and the cedar virus RdRp to ensure the conservation of motifs A-F and the core catalytic domain GDN;

[0085] Preliminary sequence design: Using the cedar virus RdRp as the backbone, retaining its highly conserved catalytic domain, and at the same time introducing the RNA binding domain of Hendra virus RdRp (residue positions 430 - 550), design a chimeric sequence CeHe-RdRp;

[0086] In vitro activity verification:;

[0087] Cross-species adaptability verification: Detect the rescue ability of RdRp for the minigenomes of different paramyxoviruses (HiV, NDV, HeV, SeV, etc.). Synthesize the L protein based on the modified RdRp, co-transfect with the helper plasmid, and detect the rescue of recombinant viruses containing different viral leader, trailer sequences and EGFP tags to preliminarily verify the construction of a universal paramyxovirus rescue platform.

[0088] Structure optimization stage

[0089] Diffusion Model Optimization: The RFdiffusion model was used to optimize the structure of the RNA-binding domain of CeHe-RdRp. The parameters were set as 1000 steps, a temperature of 0.1, and the optimization region was residues 430 - 550. After optimization, the RMSD of the RNA-binding domain decreased by The structure became more compact;

[0090] Stability Assessment: The Rosetta energy of the structure before and after optimization was calculated. The energy of the optimized structure decreased by 15.3%, indicating a significant improvement in structural stability;

[0091] Binding Site Optimization: Through molecular docking and hot spot residue analysis, it was found that residues Arg445, Lys472, and Arg490 were crucial for RNA binding. The LigandMPNN model was used to optimize the amino acid environment around these residues, and substitution schemes that could enhance RNA binding were predicted;

[0092] Sequence Reverse Design: Based on the optimized three-dimensional structure, the LigandMPNN model (parameter settings: temperature 0.1, sampling number 100) was used for sequence reverse design to generate the final optimized RdRp sequence CeHe-Opt-RdRp.

[0093] Performance Evaluation

[0094] The rescue ability of RdRp for minigenomes of different paramyxoviruses (HiV, NDV, HeV, SeV, etc.) was detected. The L protein was synthesized based on the modified RdRp, and after co-transfection with the helper plasmid, the rescue of recombinant viruses containing different viral leader, trailer sequences, and EGFP tags was detected to preliminarily verify the construction of a universal paramyxovirus rescue platform. The results showed that the universal reverse genetic platform for paramyxoviruses constructed based on this recombinant RdRp could be compatible with the promoters of HiV, NDV, HeV, and SeV, and the corresponding minigenomes could effectively express EGFP green fluorescence.

[0095] The above examples show that the cross-species universal paramyxovirus RNA-dependent RNA polymerase CeHe-Opt-RdRp designed by the present invention has excellent stability, catalytic activity, and cross-species adaptability, and can be used as an important tool for paramyxovirus research. This design system successfully integrates and applies sequence analysis, protein language model characterization, and structure optimization technologies, achieving a methodological breakthrough from traditional single-dimensional analysis to multi-dimensional integration, providing a reliable approach for the development of universal RdRps with broad application prospects.

[0096] Example 2

[0097] A design system for a cross-species universal paramyxovirus RdRp, comprising:

[0098] A data collection and processing module, which is used to collect paramyxovirus RdRp sequence data from the UniProt and PDB databases and perform redundancy removal processing on the paramyxovirus RdRp sequence data;

[0099] A sequence candidate mining module, which is used to perform candidate mining processing on the redundancy-processed paramyxovirus RdRp sequence data and screen RdRp candidate sequences with potential cross-species functions;

[0100] A candidate sequence design module, which is used to perform preliminary design on the candidate sequences. Using cedivirus RdRp as a template, retain its key catalytic sites and RNA binding domains, and at the same time fuse cross-species recognition elements of other paramyxovirus RdRps;

[0101] Synthesize and purify the preliminarily designed RdRp protein, and perform in vitro activity testing and cross-species adaptability verification

[0102] A structure design and optimization module, which is used to optimize the structure of the preliminarily designed RdRp using the RFdiffusion model, perform sequence reverse design on the optimized RdRp, and generate the final cross-species universal RdRp sequence;

[0103] Finally, express and purify the optimized RdRp protein and perform comprehensive performance evaluation.

[0104] The above has described an embodiment of the present invention in detail, but the content described is only a preferred embodiment of the present invention and cannot be considered as limiting the scope of implementation of the present invention. All equivalent changes and improvements made according to the scope of the present invention application should still fall within the patent coverage scope of the present invention.

Claims

1. A design method for a cross-species universal paramyxovirus RdRp, characterized in that, It includes the following steps: Collect paramyxovirus RdRp sequence data from UniProt and PDB databases, and perform redundancy removal on the paramyxovirus RdRp sequence data; Perform candidate mining on the redundantly processed paramyxovirus RdRp sequence data to screen for RdRp candidate sequences with potential cross-species functions; Perform preliminary design on the candidate sequences. Using Cedar virus RdRp as a template, retain its key catalytic sites and RNA binding domains, and at the same time fuse cross-species recognition elements of other paramyxovirus RdRps; Synthesize and purify the preliminarily designed RdRp protein, and perform in vitro activity testing and cross-species adaptability verification; Use the RFdiffusion model to optimize the structure of the preliminarily designed RdRp, perform sequence reverse design on the optimized RdRp, and generate the final cross-species universal RdRp sequence; Finally, express and purify the optimized RdRp protein and conduct comprehensive performance evaluation.

2. The design method of a cross-species universal paramyxovirus RdRp according to claim 1, characterized in that, The redundancy removal of paramyxovirus RdRp sequence data includes removing redundant sequences with sequence similarity > 95%.

3. The design method of a cross-species universal paramyxovirus RdRp according to claim 1, characterized in that The candidate mining process includes the following steps: Sequence similarity analysis: Use the BLAST sequence alignment tool to screen for RdRp sequences related to the target function from the database, and set the threshold parameters E-value < 1e-5 and sequence coverage > 70%; Characterization similarity extraction: Use the pre-trained protein language model ESMC to perform deep characterization extraction on the screened sequences, obtain the 1168-dimensional characterization vectors output by the model, and take the average based on amino acid characterization or sequence characterization of the sequences; Among them, calculate the Euclidean distance or cosine similarity between sequences, and select sequences with similarity > 0.85 or Euclidean distance < 0.15; Structure similarity analysis: Use Alphafold3 to perform three-dimensional structure prediction on the candidate sequences, make the average pLDDT score > 80, and perform structure similarity alignment through Foldseek, and select structures with TM-score > 0.7; Multi-dimensional comprehensive scoring: Integrate sequence similarity, characterization similarity, and structure similarity, construct a scoring function Score = 0.3×Seq_sim + 0.3×Rep_sim + 0.4×Str_sim, and select candidate sequences with scores > 0.

8.

4. A design method for a cross-species universal paramyxovirus RdRp according to claim 1, characterized in that, When performing preliminary design on the candidate sequences, functional region identification is required. Based on multiple sequence alignment and analysis, identify the conserved functional regions and variable regions in paramyxovirus RdRp.

5. The design method of a cross-species universal paramyxovirus RdRp according to claim 4, wherein According to the results of functional region identification, design a preliminary RdRp sequence containing key conserved regions 6. The design method of a cross-species universal paramyxovirus RdRp according to claim 5, wherein, The in vitro activity verification is to express and purify the designed RdRp protein and test its in vitro RNA synthesis activity; The cross-species adaptability verification is to evaluate the recognition ability and replication efficiency of the designed RdRp for RNA templates of different virus species.

7. A design method of a cross-species universal paramyxovirus RdRp according to claim 1, characterized in that The structural optimization of the preliminarily designed RdRp includes stability evaluation. Calculate the Rosetta energy score of the optimized structure to make the energy reduction > 10%.

8. A design method of a cross-species universal paramyxovirus RdRp according to claim 1, characterized in that The structural optimization of the preliminarily designed RdRp includes binding site optimization; Use molecular docking and hot spot residue analysis methods to optimize the binding interface between RdRp and the RNA template.

9. The design method of a cross-species universal paramyxovirus RdRp according to claim 1, characterized in that The comprehensive performance evaluation includes stability, activity, and cross-species adaptability verification.

10. A design system for a cross-species universal paramyxovirus RdRp, which is used to implement the design method of the cross-species universal paramyxovirus RdRp described in any one of claims 1-9, characterized in that, Including: A data collection and processing module, which is used to collect paramyxovirus RdRp sequence data from the UniProt and PDB databases and perform redundancy removal processing on the paramyxovirus RdRp sequence data; A sequence candidate mining module, which is used to perform candidate mining processing on the redundant paramyxovirus RdRp sequence data and screen RdRp candidate sequences with potential cross-species functions; A candidate sequence design module, which is used to perform preliminary design on the candidate sequences. Taking Cedar virus RdRp as a template, retaining its key catalytic sites and RNA binding domains, and at the same time fusing cross-species recognition elements of other paramyxovirus RdRps; Synthesize and purify the preliminarily designed RdRp protein, and conduct in vitro activity tests and cross-species adaptability verification. A structural design optimization module, which is used to optimize the structure of the preliminarily designed RdRp using the RFdiffusion model, perform sequence reverse design on the optimized RdRp, and generate the final cross-species universal RdRp sequence; Finally, express and purify the optimized RdRp protein and conduct a comprehensive performance evaluation.

Images