RNA (Ribonucleic Acid) design method and device based on complex information and electronic equipment

Abstract

The invention provides an RNA (Ribonucleic Acid) design method and device based on complex information and electronic equipment, and the method comprises the following steps: extracting sequence characteristics of protein by utilizing a pre-trained protein language model, and generating expression of the protein; capturing three-dimensional geometrical characteristics of the ribonucleic acid by using a geometric vector sensor encoder to obtain structure embedding of the ribonucleic acid; the expression of the protein and the structure of the ribonucleic acid are embedded into an input complex perception feature fusion module, key features are reserved based on decoupling perception screening, and information fusion is carried out based on a protein-ribonucleic acid interaction attention mechanism; and inputting the fused representation into a decoder to generate a ribonucleic acid sequence conforming to the target function and structure constraint. By combining an RNA tertiary structure and protein language model information and introducing a cross-modal learning framework, comprehensive modeling of RNA-protein interaction is realized, and the reasonability of RNA sequence generation is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of RNA design, and more particularly, to an RNA design method, apparatus, electronic device, and computer-readable storage medium based on complex information. Background Art

[0002] Existing RNA (ribonucleic acid) design methods often rely only on the sequence or structural information of RNA and fail to fully consider the binding environment of RNA with proteins. Moreover, RNA has a high degree of conformational flexibility during the interaction with proteins, and most existing methods use static structure modeling, making it difficult to adapt to dynamic binding scenarios. In addition, existing RNA design methods lack an efficient screening strategy while optimizing the target function, making it difficult to guarantee the quality of the generated sequences. Summary of the Invention

[0003] To solve the existing technical problems, embodiments of the present invention provide an RNA design method, apparatus, electronic device, and computer-readable storage medium based on complex information.

[0004] In a first aspect, embodiments of the present invention provide an RNA design method based on complex information, including: using a pre-trained protein language model to extract sequence features of a protein and generate a context-dependent representation of the protein; using a geometric vector perceptron encoder to capture three-dimensional geometric features of ribonucleic acid and perform a structured characterization of the ribonucleic acid to obtain a structural embedding of the ribonucleic acid; inputting the representation of the protein and the structural embedding of the ribonucleic acid into a complex-aware feature fusion module, screening and retaining key features based on the decoupled perception in the complex-aware feature fusion module, and performing information fusion based on the protein-ribonucleic acid interaction attention mechanism in the complex-aware feature fusion module; inputting the fused representation into a decoder to generate a ribonucleic acid sequence that meets the target function and structural constraints.

[0005] Optionally, the representation of the protein is defined as: where H Pro is the representation of the protein, L p is the number of protein amino acids, and D is the feature dimension.

[0006] Optionally, using a geometric vector perceptron encoder to capture three-dimensional geometric features of ribonucleic acid and perform a structured characterization of the ribonucleic acid to obtain a structural embedding of the ribonucleic acid includes: the geometric vector perceptron encoder processes the ribonucleic acid backbone atom coordinates, converts the ribonucleic acid backbone atom coordinates into vector and scalar features, and combines direction vector and dihedral angle information to capture the global and local structural information of the ribonucleic acid to obtain the structural embedding of the ribonucleic acid.

[0007] Optionally, the structure embedding of ribonucleic acid is defined as: where H RNA is the structure embedding of the ribonucleic acid, L r is the number of ribonucleic acid nucleotides, and D is the feature dimension.

[0008] Optionally, retaining key features based on decoupled perception screening in the complex-aware feature fusion module includes: calculating the Euclidean distance between protein amino acids and ribonucleic acid nucleotides; selecting several protein amino acids closest to the ribonucleic acid nucleotides and correspondingly screening the representation of the protein to obtain a locally screened protein representation; calculating global average pooling to obtain a globally screened protein representation; concatenating the locally screened protein representation and the globally screened protein representation to obtain a finally screened protein representation.

[0009] Optionally, information fusion is performed based on the protein-ribonucleic acid interaction attention mechanism in the complex-aware feature fusion module, including: concatenating the structure embedding of the ribonucleic acid with the finally screened protein representation to obtain a concatenated result, and using the cross-attention mechanism with the structure embedding of the ribonucleic acid as the query vector and the concatenated result as the key and value to enhance the structure embedding of the ribonucleic acid.

[0010] Optionally, after generating the ribonucleic acid sequence, the method further includes: screening the ribonucleic acid that meets the task requirements using an affinity evaluation mechanism, and using the root mean square deviation and an index for measuring the topological similarity between protein structures to ensure the structural compatibility of the ribonucleic acid in the protein-ribonucleic acid complex; in each iteration, screening candidate sequences in a preset region based on the affinity score and using a structure prediction model for verification to obtain an optimal candidate sequence.

[0011] In a second aspect, an embodiment of the present invention further provides an RNA design device based on complex information, including: an encoding module, a fusion module, and a decoding module; the encoding module is used to extract the sequence features of a protein using a pre-trained protein language model to generate a context-dependent representation of the protein; using a geometric vector perceptron encoder to capture the three-dimensional geometric features of ribonucleic acid and perform a structured characterization of the ribonucleic acid to obtain the structure embedding of the ribonucleic acid; the fusion module is used to input the representation of the protein and the structure embedding of the ribonucleic acid into a complex-aware feature fusion module, retain key features based on decoupled perception screening in the complex-aware feature fusion module, and perform information fusion based on the protein-ribonucleic acid interaction attention mechanism in the complex-aware feature fusion module; the decoding module is used to input the fused characterization into a decoder to generate a ribonucleic acid sequence that meets the target function and structural constraints.

[0012] In a third aspect, an embodiment of the present invention provides an electronic device, including a processor and a memory. The memory stores a computer program, and the processor executes the computer program stored in the memory. When the computer program is executed by the processor, the RNA design method based on complex information described in the first aspect above is implemented.

[0013] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the RNA design method based on complex information described in the first aspect above is implemented.

[0014] In a fifth aspect, the present application further provides a computer program product, including a computer program. When the computer program is executed, the RNA design method based on complex information described in the first aspect or any possible design manner of the first aspect can be implemented.

[0015] The RNA design method, device, electronic device, and computer-readable storage medium provided by the embodiments of the present invention, by combining RNA tertiary structure and protein language model information, introducing a cross-modal learning framework, achieve a comprehensive modeling of RNA-protein interactions, and improve the rationality of RNA sequence generation. Moreover, the embodiments of the present invention adopt a complex-aware transformer to fuse multi-level representations of RNA and protein, thereby capturing key structural constraints and dynamic interactions. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the background art, the drawings required to be used in the embodiments of the present invention or the background art will be described below.

[0017] Figure 1 Shows a flowchart of an RNA design method based on complex information provided by an embodiment of the present invention;

[0018] Figure 2 Shows a schematic diagram of the model framework process for executing the RNA design method based on complex information provided by an embodiment of the present invention;

[0019] Figure 3 Shows a schematic structural diagram of an RNA design device based on complex information provided by an embodiment of the present invention;

[0020] Figure 4 Shows a schematic structural diagram of an electronic device for executing the RNA design method based on complex information provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] The embodiments of the present invention will be described below with reference to the accompanying drawings in the embodiments of the present invention.

[0022] Figure 1 The flowchart of an RNA design method based on complex information provided by the embodiments of the present invention is shown. As Figure 1 shown, the method includes the following steps 101-103.

[0023] Step 101: Use a pre-trained protein language model to extract the sequence features of a protein and generate a context-dependent representation of the protein; use a geometric vector perceptron encoder to capture the three-dimensional geometric features of ribonucleic acid, perform a structural characterization of the ribonucleic acid, and obtain a structural embedding of the ribonucleic acid.

[0024] See Figure 2 shown, Figure 2 The schematic diagram of the model framework process for executing the RNA design method based on complex information provided by the embodiments of the present invention is shown. This framework is used to calculate and screen ribonucleic acid (RNA) sequences that meet functional requirements and structural compatibility. This framework is optimized based on the characteristics of the protein-ribonucleic acid complex to improve the accuracy and adaptability of RNA design.

[0025] In the RNA design model, in view of the modeling requirements of the protein-ribonucleic acid complex, the tertiary structure of ribonucleic acid and the sequence and structural information of the protein are encoded to capture key interaction features. First, for the protein in the protein-ribonucleic acid complex, a pre-trained protein language model (PLM), such as ESM-2 (Evolutionary Scale Modeling 2, a Transformer-based protein language model developed by Meta AI for predicting the structure and function from protein sequences) or ESM-3 (the third-generation protein generative AI model developed by the Evolutionary Scale team, aiming to drive innovation in the fields of biomedicine and biotechnology through advanced protein design), is used to encode the sequence and structure of the protein. The protein language model extracts high-dimensional embeddings through deep learning, can capture the sequence-level features of the protein, including secondary structure, functional patterns, and evolutionary relationships, and obtain a representation of the protein, and this representation of the protein has context dependence.

[0026] Secondly, for the ribonucleic acid in the protein-ribonucleic acid complex, that is, for the tertiary structure representation of ribonucleic acid, a method similar to Rhodesign (a deep learning-based RNA sequence design platform that focuses on generating RNA sequences with specific functions based on three-dimensional structure data) is adopted, and a Geometric Vector Perceptron (GVP) encoder is used for encoding to capture the three-dimensional geometric features of RNA and perform a structured characterization on it, obtaining the structural embedding of ribonucleic acid (which can also be understood as the RNA structure representation).

[0027] Step 102: Input the representation of the protein and the structural embedding of the ribonucleic acid into the complex-aware feature fusion module. Based on the decoupled awareness screening in the complex-aware feature fusion module, retain the key features, and perform information fusion based on the protein-ribonucleic acid interaction attention mechanism in the complex-aware feature fusion module.

[0028] To more effectively perform RNA inverse folding design, the embodiment of the present invention proposes a complex-aware information fusion (Complex-Aware Information Fusion) module, which includes a decoupled awareness screening (Dis-AwareFiltering, or distance awareness screening) and a protein-ribonucleic acid interaction attention (PR Inter Attention) mechanism. The former (i.e., the decoupled awareness screening) is used to filter redundant information to ensure the retention of key features, and the latter (i.e., the protein-ribonucleic acid interaction attention mechanism) models the fine interaction pattern between the protein and RNA through the attention mechanism to further optimize the information fusion process.

[0029] That is to say, after obtaining the RNA structure representation (i.e., the structural embedding of ribonucleic acid) and the representation of the protein, the embodiment of the present invention introduces a complex-aware attention network (CAFormer) to integrate the key interaction information of the protein-ribonucleic acid complex and optimize the inverse folding process of RNA design. The complex-aware attention network (CAFormer) architecture consists of a decoupled awareness screening and a protein-ribonucleic acid interaction attention (PR Inter Attention) mechanism to ensure that the model focuses on the binding region between the protein and RNA and optimizes the fusion of interaction features.

[0030] Step 103: Input the fused representation into the decoder to generate a ribonucleic acid sequence that meets the target function and structural constraints.

[0031] The fused representation is input into the decoder, which is used to generate a ribonucleic acid sequence that meets the target function and structural constraints.

[0032] The RNA design method based on complex information provided by the embodiments of the present invention combines RNA tertiary structure and protein language model information, introduces a cross-modal learning framework, realizes comprehensive modeling of RNA-protein interactions, and improves the rationality of RNA sequence generation. Moreover, the embodiments of the present invention adopt a complex-aware transformer to fuse multi-level representations of RNA and proteins, thereby capturing key structural constraints and dynamic interactions.

[0033] Optionally, the representation of the protein is defined as: where H Pro is the representation of the protein, L p is the number of protein amino acids, and D is the feature dimension. This representation provides rich sequence and structural information for the RNA design task, enabling the model to be optimized in different environments.

[0034] Optionally, in step 101 above, "using a geometric vector perceptron encoder to capture the three-dimensional geometric features of ribonucleic acid, perform a structural representation of ribonucleic acid, and obtain a structural embedding of ribonucleic acid" may include the following steps A.

[0035] Step A: The geometric vector perceptron encoder processes the ribonucleic acid backbone atom coordinates (such as C4', C1', N1), converts the ribonucleic acid backbone atom coordinates into vector and scalar features, and combines direction vector and dihedral angle information to capture the global and local structural information of ribonucleic acid, and obtains a structural embedding of ribonucleic acid. Optionally, the structural embedding of ribonucleic acid generated by this encoding method is defined as: where H RNA is the structural embedding of ribonucleic acid, L r is the number of RNA nucleotides, and D is the feature dimension. This representation is used for subsequent protein-ribonucleic acid feature fusion to optimize the generation of ribonucleic acid sequences.

[0036] Optionally, in step 102 above, "retaining key features based on decoupled perception screening in the complex-aware feature fusion module" may include the following steps B1-B3.

[0037] Step B1: Calculate the Euclidean distance between protein amino acids and ribonucleic acid nucleotides. The corresponding formula is: D ij =||C pi -C rj ||, where C pi and C rj respectively represent the three-dimensional coordinates of protein amino acids and ribonucleic acid nucleotides.

[0038] Step B2: Select several protein amino acids that are closest to ribonucleic acid nucleotides, and correspondingly screen the representations of the protein to obtain a locally screened protein representation; calculate global average pooling to obtain a globally screened protein representation.

[0039] In the embodiment of the present invention, K protein amino acids that are closest to ribonucleic acid nucleotides are selected, and the representation of the protein is correspondingly screened: S = argsort(min(D i,: ))[:K]; where, H Pro,local is the locally screened protein representation. In order to retain the global information of the protein, in the embodiment of the present invention, global average pooling (GAP) also needs to be calculated at the same time, and the formula is: H Pro,global = GAP(H Pro ); where, H Pro,global is the globally screened protein representation.

[0040] Step B3: Concatenate the locally screened protein representation H Pro,local and the globally screened protein representation H Pro,global to obtain the finally screened protein representation H Pro,filtered , H Pro,filtered = Concat(H Pro,local , H Pro,global ). This screening mechanism ensures that the model focuses on the key binding regions of the protein-ribonucleic acid complex and retains the global biological background information.

[0041] Optionally, in the above step 102, "performing information fusion based on the protein-ribonucleic acid interaction attention mechanism in the complex-aware feature fusion module" may include the following step C.

[0042] Step C: Concatenate the structural embedding of the ribonucleic acid with the finally screened protein representation to obtain a concatenated result, and use the cross-attention mechanism, with the structural embedding of the ribonucleic acid as the query vector and the concatenated result as the key and value, to enhance the structural embedding of the ribonucleic acid.

[0043] In the protein-ribonucleic acid interaction attention mechanism, in the embodiment of the present invention, the dynamic interaction of ribonucleic acid in different binding environments is modeled through the attention mechanism, so that the generation process of ribonucleic acid can effectively adapt to the specific binding characteristics of the protein. Specifically, the structural embedding H RNA (that is, the structural representation) of the ribonucleic acid is concatenated with the finally screened protein representation H Pro,filtered to obtain a concatenated result H com , H com = Concat(H RNA , H Pro,filtered)。Then, the embodiments of the present invention utilize the cross-attention mechanism to embed the structure of ribonucleic acid into H RNA as the query vector (Query), and H com as the key (Key) and value (Value) to enhance the structural representation of ribonucleic acid (i.e., structural embedding):

[0044] Q = Proj q (H RNA );

[0045] K, V = Proj k (H com ), Proj k (H com );

[0046]

[0047] Among them, Proj * represents linear projection of the query vector Q, key K, and value V. For simplicity of expression, the layer normalization (LayerNorm) operation is omitted.

[0048] In addition, based on the above inverse folding model, the embodiments of the present invention may also correspond to a high-affinity ribonucleic acid RNA design model framework, which integrates protein-ribonucleic acid complex information and can combine an affinity evaluation mechanism to optimize the functional adaptability of ribonucleic acid sequences. That is to say, in addition to including the above structure-to-sequence design model (i.e., the model architecture corresponding to the RNA design method based on complex information), this design framework also includes an evaluation tool.

[0049] Optionally, after generating the ribonucleic acid sequence, the method may further include the following step D1.

[0050] Step D1: Use the affinity evaluation mechanism to screen ribonucleic acids that meet the task requirements, and use the root mean square deviation and an index for measuring the topological similarity between protein structures to ensure the structural compatibility of ribonucleic acids in the protein-ribonucleic acid complex.

[0051] In the embodiments of the present invention, the RNA design method based on complex information can be divided into two stages: the design stage and the constraint evaluation stage. In the design stage, the complex-aware design model in the embodiments of the present invention generates ribonucleic acid sequences that meet the constraint conditions by integrating RNA tertiary structure information and protein-ribonucleic acid complex information. Using pre-trained protein language models (PLMs) and complex-aware attention networks (CAFormer), the model can effectively capture the complex interactions between ribonucleic acids and proteins and optimize the ribonucleic acid sequences to match specific binding sites and structural features. In the constraint evaluation stage, that is, after generating the ribonucleic acid sequences, the embodiments of the present invention can use an evaluation mechanism to screen ribonucleic acids that meet the task requirements, such as using affinity scores to screen target ribonucleic acids with high affinity. The affinity evaluation is based on a regression model trained on the PRA201 dataset to predict the binding affinity between protein and ribonucleic acid. At the same time, the embodiments of the present invention adopt multiple structure prediction models such as AlphaFold3 (a revolutionary artificial intelligence model released by Google DeepMind and the Isomorphic Labs team in May 2024 for predicting the structures of biomolecules and their interactions), RhoFold (a deep learning-based method for predicting the three-dimensional structure of RNA), and RoseTTAFold2NA (a deep learning-based biomolecular structure prediction tool dedicated to predicting the three-dimensional structure of protein-nucleic acid (RNA and DNA) complexes) to calculate the RMSD (Root Mean Square Deviation) and TM-score (Template Modeling Score, an index used to measure the topological similarity between protein structures) of RNA to ensure the structural compatibility of ribonucleic acids in the protein-ribonucleic acid complex, that is, RMSD and TM-Score are used to verify the ribonucleic acid structural compatibility.

[0052] The present invention combines a machine learning-driven affinity prediction model to construct an iterative optimization framework, which verifies by screening candidate RNA sequences and combining multiple folding models to ensure a high degree of compatibility in structure and function of the generated RNA.

[0053] In addition, in the embodiments of the present invention, the RNA design method based on complex information can be divided into three stages: the design stage, the constraint evaluation, and the iterative screening. That is to say, after generating the ribonucleic acid sequences, the method can further include the following step D2.

[0054] Step D2: In each iteration, candidate sequences in a preset region (e.g., the top 10% - 20%) are screened based on the affinity scores and verified using a structure prediction model to obtain the optimal candidate sequences. By updating the candidate pool through multiple iterations, this framework can efficiently generate RNA sequences with high affinity and structural adaptability, providing a computational alternative for experimental methods such as SELEX (Systematic Evolution of Ligands by Exponential Enrichment).

[0055] The present invention provides a computationally efficient alternative solution. Through multiple rounds of iterative optimization, it improves the accuracy of RNA design and effectively reduces the overhead of experimental screening.

[0056] The method for RNA design based on complex information provided by the embodiments of the present invention is described in detail above. This method can also be implemented by a corresponding device. The device for RNA design based on complex information provided by the embodiments of the present invention is described in detail below.

[0057] Figure 3 The structural schematic diagram of a device for RNA design based on complex information provided by the embodiments of the present invention is shown. As Figure 3 shown, the device for RNA design based on complex information includes a processor. The processor includes: an encoding module 31, a fusion module 32, and a decoding module 33.

[0058] The encoding module 31 is used to extract the sequence features of a protein using a pre-trained protein language model to generate a context-dependent representation of the protein; and use a geometric vector perceptron encoder to capture the three-dimensional geometric features of the ribonucleic acid, perform a structured characterization of the ribonucleic acid, and obtain a structural embedding of the ribonucleic acid.

[0059] The fusion module 32 is used to input the representation of the protein and the structural embedding of the ribonucleic acid into a complex-aware feature fusion module, retain key features based on decoupled perception screening in the complex-aware feature fusion module, and perform information fusion based on the protein-ribonucleic acid interaction attention mechanism in the complex-aware feature fusion module.

[0060] The decoding module 33 is used to input the fused representation into a decoder to generate a ribonucleic acid sequence that meets the target function and structural constraints.

[0061] Optionally, the representation of the protein is defined as: where H Pro is the representation of the protein, L p is the number of protein amino acids, and D is the feature dimension.

[0062] Optionally, the encoding module 31 includes: a geometric vector perceptron encoder;

[0063] The geometric vector perceptron encoder converts the ribonucleic acid backbone atom coordinates into vector and scalar features by processing the ribonucleic acid backbone atom coordinates, and captures the global and local structural information of the ribonucleic acid by combining the direction vector and dihedral angle information, so as to obtain the structural embedding of the ribonucleic acid.

[0064] Optionally, the structural embedding of ribonucleic acid is defined as: where H RNA is the structural embedding of the ribonucleic acid, L r is the number of ribonucleic acid nucleotides, and D is the feature dimension.

[0065] Optionally, the fusion module 32 includes: a distance calculation unit, a screening unit, and a splicing unit.

[0066] The distance calculation unit is used to calculate the Euclidean distance between protein amino acids and ribonucleic acid nucleotides.

[0067] The screening unit is used to select several protein amino acids closest to the ribonucleic acid nucleotides, and correspondingly screen the representation of the protein to obtain a locally screened protein representation; calculate the global average pooling to obtain a globally screened protein representation;

[0068] The splicing unit is used to splice the locally screened protein representation and the globally screened protein representation to obtain a finally screened protein representation.

[0069] Optionally, the fusion module 32 includes: an enhancement unit.

[0070] The enhancement unit is used to splice the structural embedding of the ribonucleic acid and the finally screened protein representation to obtain a splicing result, and use the cross-attention mechanism, with the structural embedding of the ribonucleic acid as the query vector, and the splicing result as the key and value, to enhance the structural embedding of the ribonucleic acid.

[0071] Optionally, after generating the ribonucleic acid sequence, the device further includes: an evaluation module and an iteration module.

[0072] The evaluation module is used to screen the ribonucleic acid that meets the task requirements using an affinity evaluation mechanism to ensure the structural compatibility of the ribonucleic acid in the protein-ribonucleic acid complex;

[0073] The iteration module is used to screen candidate sequences in a preset region based on the affinity score in each round of iteration, and use a structure prediction model for verification to obtain an optimal candidate sequence.

[0074] The device provided by the embodiments of the present invention combines RNA tertiary structure and protein language model information, introduces a cross-modal learning framework, realizes comprehensive modeling of RNA-protein interactions, and improves the rationality of RNA sequence generation. Moreover, the embodiments of the present invention adopt a complex-aware transformer to fuse multi-level representations of RNA and proteins, thereby capturing key structural constraints and dynamic interactions. The present invention combines a machine learning-driven affinity prediction model to construct an iterative optimization framework, and verifies by screening candidate RNA sequences and combining multiple folding models to ensure high compatibility of the generated RNA in terms of structure and function. The present invention provides a computationally efficient alternative, which improves the accuracy of RNA design through multiple rounds of iterative optimization and effectively reduces the overhead of experimental screening.

[0075] It should be noted that when the RNA design device based on complex information provided in the above embodiments realizes the corresponding functions, only the above division of each functional module is used as an example. In actual applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the RNA design device based on complex information provided in the above embodiments and the embodiments of the RNA design method based on complex information belong to the same concept. For the specific implementation process, please refer to the method embodiments, which will not be repeated here.

[0076] According to one aspect of the present application, the embodiments of the present invention further provide a computer program product, which includes a computer program containing program codes for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part. When the computer program is executed by the processor, it executes the RNA design method based on complex information provided by the embodiments of the present application.

[0077] In addition, the embodiments of the present invention further provide an electronic device, including a bus, a transceiver, a memory, a processor, and a computer program stored on the memory and executable on the processor. The transceiver, the memory, and the processor are respectively connected through the bus. When the computer program is executed by the processor, it realizes each process of the above-mentioned embodiments of the RNA design method based on complex information and can achieve the same technical effects. To avoid repetition, it will not be repeated here.

[0078] Specifically, referring to Figure 4 as shown, the electronic device includes a bus 1110, a processor 1120, a transceiver 1130, a bus interface 1140, a memory 1150, and a user interface 1160.

[0079] In an embodiment of the present invention, the electronic device further includes: a computer program stored on the memory 1150 and executable on the processor 1120, and when the computer program is executed by the processor 1120, it implements each process of the above embodiment of the RNA design method based on complex information.

[0080] A transceiver 1130, configured to receive and send data under the control of the processor 1120.

[0081] In an embodiment of the present invention, a bus architecture (represented by the bus 1110), the bus 1110 may include any number of interconnected buses and bridges, and the bus 1110 connects various circuits including one or more processors represented by the processor 1120 and a memory represented by the memory 1150 together.

[0082] The bus 1110 represents one or more of any of several types of bus structures, including a memory bus and a memory controller, a peripheral bus, an Accelerate Graphical Port (AGP), a processor, or a local bus using any bus structure in various bus architectures. By way of example and not limitation, such architectures include: an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MCA) bus, an Extended ISA (EISA) bus, a Video Electronics Standards Association (VESA), a Peripheral Component Interconnect (PCI) bus.

[0083] The processor 1120 may be an integrated circuit chip with signal processing capabilities. In implementation, the steps of the above method embodiments may be completed by the integrated logic circuit in the hardware of the processor or instructions in the form of software. The above-mentioned processor includes: general-purpose processor, central processing unit (CPU), network processor (NP), digital signal processor (DSP), application specific integrated circuit (ASIC), field programmable gate array (FPGA), complex programmable logic device (CPLD), programmable logic array (PLA), microcontroller unit (MCU), or other programmable logic devices, discrete gates, transistor logic devices, discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. For example, the processor may be a single-core processor or a multi-core processor, and the processor may be integrated on a single chip or located on multiple different chips.

[0084] The processor 1120 may be a microprocessor or any conventional processor. The method steps disclosed in combination with the embodiments of the present invention may be directly executed and completed by the hardware decoding processor, or executed and completed by the combination of the hardware and software modules in the decoding processor. The software module may be located in a readable storage medium well-known in the art such as random access memory (RAM), flash memory, read-only memory (ROM), programmable ROM (PROM), erasable programmable ROM (EPROM), registers, etc. The readable storage medium is located in the memory, and the processor reads the information in the memory and combines its hardware to complete the steps of the above method.

[0085] The bus 1110 may also connect together various other circuits such as, for example, peripheral devices, voltage regulators, or power management circuits. The bus interface 1140 provides an interface between the bus 1110 and the transceiver 1130, which are all well-known in the art. Therefore, the embodiments of the present invention will not be further described herein.

[0086] The transceiver 1130 can be a single component or multiple components, such as multiple receivers and transmitters, providing units for communicating with various other devices over a transmission medium. For example: The transceiver 1130 receives external data from other devices, and the transceiver 1130 is used to send the data processed by the processor 1120 to other devices. Depending on the nature of the computer system, a user interface 1160 may also be provided, such as: a touch screen, a physical keyboard, a display, a mouse, speakers, a microphone, a trackball, a joystick, a stylus.

[0087] It should be understood that in the embodiments of the present invention, the memory 1150 may further include memories remotely located relative to the processor 1120, and these remotely located memories can be connected to the server through a network. One or more parts of the above networks can be an ad hoc network, an intranet, an extranet, a virtual private network (VPN), a local area network (LAN), a wireless local area network (WLAN), a wide area network (WAN), a wireless wide area network (WWAN), a metropolitan area network (MAN), the Internet, a public switched telephone network (PSTN), a plain old telephone service network (POTS), a cellular telephone network, a wireless network, a Wi-Fi network, and a combination of two or more of the above networks. For example, the cellular telephone network and the wireless network can be a Global System for Mobile Communications (GSM) system, a Code Division Multiple Access (CDMA) system, a Worldwide Interoperability for Microwave Access (WiMAX) system, a General Packet Radio Service (GPRS) system, a Wideband Code Division Multiple Access (WCDMA) system, a Long Term Evolution (LTE) system, an LTE Frequency Division Duplexing (FDD) system, an LTE Time Division Duplexing (TDD) system, an Advanced Long Term Evolution (LTE-A) system, a Universal Mobile Telecommunications System (UMTS) system, an Enhance Mobile Broadband (eMBB) system, a massive Machine Type of Communication (mMTC) system, an UltraReliable Low Latency Communications (uRLLC) system, etc.

[0088] It should be understood that the memory 1150 in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both a volatile memory and a non-volatile memory. Among them, the non-volatile memory includes: Read-Only Memory (ROM), Programmable ROM (PROM), Erasable PROM (EPROM), Electrically EPROM (EEPROM), or Flash Memory.

[0089] The volatile memory includes: Random Access Memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as: Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced SDRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM). The memory 1150 of the electronic device described in the embodiments of the present invention includes but is not limited to the above and any other suitable types of memory.

[0090] In the embodiments of the present invention, the memory 1150 stores the following elements of the operating system 1151 and the application program 1152: executable modules, data structures, or subsets or extended sets thereof.

[0091] Specifically, the operating system 1151 includes various system programs, such as: framework layer, core library layer, driver layer, etc., for implementing various basic services and processing hardware-based tasks. The application program 1152 includes various application programs, such as: Media Player, Browser, for implementing various application services. The program for implementing the method of the embodiments of the present invention may be included in the application program 1152. The application program 1152 includes: applets, objects, components, logics, data structures, and other computer system executable instructions for performing specific tasks or implementing specific abstract data types.

[0092] In addition, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements each process of the above-described embodiment of the RNA design method based on complex information and can achieve the same technical effects. To avoid repetition, details are not described herein again.

[0093] A computer-readable storage medium includes: permanent and non-permanent, removable and non-removable media, which are tangible devices that can retain and store instructions for use by an instruction execution device. A computer-readable storage medium includes: electronic storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, and any suitable combination of the above. A computer-readable storage medium includes: phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), non-volatile random access memory (NVRAM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette tape storage, magnetic tape disk storage or other magnetic storage devices, memory sticks, mechanical encoding devices (such as punched cards or raised structures in grooves on which instructions are recorded), or any other non-transmission medium that can be used to store information accessible by a computing device. As defined in the embodiment of the present invention, a computer-readable storage medium does not include transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission medium (such as light pulses passing through an optical fiber cable), or electrical signals transmitted through wires.

[0094] In several embodiments provided in the present application, it should be understood that the disclosed apparatus, electronic device, and method can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of the modules or units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed coupling or direct coupling or communication connection between each other can be an indirect coupling or communication connection through some interfaces, devices, or units, and can also be a connection in electrical, mechanical, or other forms.

[0095] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units. They can be located in one position or distributed to multiple network units. Some or all of the units can be selected according to actual needs to solve the problems to be solved by the solution of the embodiment of the present invention.

[0096] In addition, in each embodiment of the present invention, each functional unit may be integrated into a processing unit, may exist separately physically for each unit, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of a software functional unit.

[0097] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it may be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiments of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, may be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (including: a personal computer, a server, a data center, or other network devices) to execute all or part of the steps of the methods described in the embodiments of the present invention. And the above-mentioned storage medium includes various media that can store program codes as listed above.

[0098] In the description of the embodiments of the present invention, those skilled in the art should know that the embodiments of the present invention can be implemented as a method, a device, an electronic device, and a computer-readable storage medium. Therefore, the embodiments of the present invention can be specifically implemented in the following forms: complete hardware, complete software (including firmware, resident software, microcode, etc.), and a combination of hardware and software. In addition, in some embodiments, the embodiments of the present invention can also be implemented in the form of a computer program product in one or more computer-readable storage media, and the computer-readable storage media contains computer program codes.

[0099] The above-mentioned computer-readable storage medium may adopt any combination of one or more computer-readable storage media. The computer-readable storage medium includes: an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or component, or any combination of the above. More specific examples of the computer-readable storage medium include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any combination of the above. In the embodiments of the present invention, the computer-readable storage medium may be any tangible medium that contains or stores a program, and the program can be used by or in combination with an instruction execution system, device, or component.

[0100] The computer program code included in the above computer-readable storage medium can be transmitted by any suitable medium, including: wireless, wire, optical cable, radio frequency (RF), or any suitable combination of the above.

[0101] The computer program code for performing the operations of the embodiments of the present invention can be written in assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, integrated circuit configuration data, or in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as C language or similar programming languages. The computer program code can be executed entirely on the user's computer, partially on the user's computer, executed as an independent software package, partially on the user's computer and partially on a remote computer, and entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including: local area network (LAN) or wide area network (WAN), and can also be connected to an external computer.

[0102] The methods, apparatuses, and electronic devices provided by the embodiments of the present invention are described by flowcharts and / or block diagrams.

[0103] It should be understood that each block of the flowchart and / or block diagram, and the combinations of blocks in the flowchart and / or block diagram, can be implemented by computer-readable program instructions. These computer-readable program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine. These computer-readable program instructions are executed by a computer or other programmable data processing device, resulting in an apparatus that implements the functions / operations specified in the blocks of the flowchart and / or block diagram.

[0104] These computer-readable program instructions can also be stored in a computer-readable storage medium that enables a computer or other programmable data processing device to work in a specific manner. In this way, the instructions stored in the computer-readable storage medium produce an instruction device product that includes the instructions for implementing the functions / operations specified in the blocks of the flowchart and / or block diagram.

[0105] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other devices, causing a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other devices to generate a computer-implemented process, so that the instructions executed on the computer or other programmable data processing apparatus can provide a process for realizing the functions / operations specified in the blocks of the flowchart and / or block diagram.

[0106] As described above, the above are only specific embodiments of the embodiments of the present invention, but the protection scope of the embodiments of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed in the embodiments of the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the embodiments of the present invention. Therefore, the protection scope of the embodiments of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for RNA design based on complex information, characterized in that: include: Use the pre-trained protein language model to extract protein sequence features and generate context-dependent protein representations; Using a geometric vector perceptron encoder to capture the three-dimensional geometric features of RNA, perform structural characterization on the RNA, and obtain a structural embedding of the RNA; Embedding the representation of the protein and the structure of the RNA into an input complex-aware feature fusion module, retaining key features based on decoupling perception screening in the complex-aware feature fusion module, and performing information fusion based on the protein-RNA interactive attention mechanism in the complex-aware feature fusion module; The fused representation is input into the decoder to generate RNA sequences that meet the target functional and structural constraints.

2. The method according to claim 1, characterized in that The protein representation is defined as: Among them, H Pro is the protein expression, L p is the number of protein amino acids, and D is the characteristic dimension.

3. The method according to claim 1, characterized in that The method of capturing the three-dimensional geometric features of RNA using a geometric vector sensor encoder, performing structural characterization on the RNA, and obtaining the structural embedding of the RNA includes: The geometric vector perceptron encoder processes the RNA backbone atomic coordinates, converts the RNA backbone atomic coordinates into vector and scalar features, and combines the direction vector and dihedral angle information to capture the global and local structural information of the RNA, thereby obtaining the structural embedding of the RNA.

4. The method according to claim 1, characterized in that: The structural embedding of the RNA is defined as: Among them, H RNA For the RNA structure embedding, L r is the number of ribonucleic acid nucleotides, and D is the characteristic dimension.

5. The method according to claim 1, characterized in that The decoupling perception screening and retaining key features in the complex perception feature fusion module includes: Calculate the Euclidean distance between protein amino acids and RNA nucleotides; Selecting a number of the protein amino acids closest to the ribonucleic acid nucleotides, and correspondingly screening the protein representations to obtain a locally screened protein representation; calculating a global average pooling to obtain a globally screened protein representation; The local screened protein representation and the global screened protein representation are spliced ​​to obtain a final screened protein representation.

6. The method according to claim 5, characterized in that The information fusion is performed based on the protein-RNA interaction attention mechanism in the complex perception feature fusion module, including: The structural embedding of the RNA is concatenated with the final screened protein representation to obtain a concatenated result, and a cross-attention mechanism is used to enhance the structural embedding of the RNA by using the structural embedding of the RNA as a query vector and the concatenated result as a key and a value.

7. The method according to claim 1, characterized in that After generating the RNA sequence, the method further comprises: Using an affinity evaluation mechanism to screen the RNA that meets the task requirements, using a root mean square deviation and an index for measuring topological similarity between protein structures to ensure the structural compatibility of the RNA in a protein-RNA complex; In each round of iteration, candidate sequences in the preset region are screened based on affinity scores and verified using the structure prediction model to obtain the optimal candidate sequence.

8. An RNA design device based on complex information, characterized in that: include: Encoding module, fusion module and decoding module; The encoding module is used to extract sequence features of proteins using a pre-trained protein language model to generate a representation of the protein with context dependency; Using a geometric vector perceptron encoder to capture the three-dimensional geometric features of RNA, perform structural characterization on the RNA, and obtain a structural embedding of the RNA; The fusion module is used to embed the representation of the protein and the structure of the RNA into the input complex perception feature fusion module, screen and retain key features based on the decoupling perception in the complex perception feature fusion module, and perform information fusion based on the protein-RNA interaction attention mechanism in the complex perception feature fusion module; The decoding module is used to input the fused representation into a decoder to generate a RNA sequence that meets the target function and structural constraints.

9. An electronic device comprising a processor and a memory, wherein the memory stores a computer program, characterized in that: The processor executes the computer program stored in the memory to implement the steps in the RNA design method based on compomer information according to any one of claims 1 to 7.

10. A computer program product, characterized in that The invention comprises a computer program, which, when executed, implements the steps in the RNA design method based on compomer information as claimed in any one of claims 1 to 7.

Images