76 results about "Sequence design" patented technology

The invention belongs to the technical field of essential gene prediction, and particularly relates to an essential gene prediction method based on DNA large model and time-frequency domain deep learning fusion, and the method comprises the steps: taking a domain DNA large model as a core representation layer, and obtaining special gene representation through cross-species corpus pre-training and task fine tuning; a T-Block and F-Block dual-channel time-frequency fusion structure is adopted, and the local dependence and long-range regulation relation of a gene sequence is synchronously captured by expanding DFT (Discrete Fourier Transform), complex value attention and iDFT (Initial Discrete Fourier Transform) conversion; designing an efficient modeling reasoning scheme of sliding window slices and gene-level aggregation aiming at an ultra-long sequence; in combination with class imbalance and a noise robust training strategy, cross-cell line / cross-platform transferable threshold output is realized through temperature scaling calibration, an uncertainty quantization and structured interface is matched, and drug target screening and experimental design decision are supported. The system supports the realization of multiple programming languages, and can complete low-delay end-to-end reasoning in a conventional hardware environment.

Disclosed are an auxiliary pilot adaptive optimization method, apparatus and device, and a medium, which relate to the technical field of wireless communications, and solve the problems of static configuration, resource waste and single function of the existing auxiliary pilot. An auxiliary pilot frequency sequence is designed into a synchronous prefix and an estimation core section, and the synchronous prefix adopts a pseudo-random sequence with high autocorrelation and is used for timing synchronization (positioning through a sequence autocorrelation peak value) and frequency synchronization (calculating frequency offset through a phase difference) in a demodulation process; the estimation core section adopts an orthogonal sequence with a low peak-to-average ratio and is used for accurate channel estimation; the double-segment sequence is connected through a preset code pattern, and it is ensured that a demodulation end can perform synchronous analysis. By sensing the channel state in real time, parameters of the auxiliary pilot frequency are dynamically adjusted, and a synchronization function is fused, so that the demodulation performance and the resource utilization rate are improved.

The invention discloses an autoclave temperature offline simulation and online prediction method based on Hybrid KAN feature fusion. The autoclave temperature offline simulation and online prediction method comprises the steps that historical temperature time sequence data and static process parameter data of different batches are collected; performing feature screening and preprocessing, and constructing a time sequence data input sequence and a static process parameter input sequence; designing a Hybrid KAN neural network model, taking the time sequence data input sequence and the static process parameter input sequence as input, and outputting a predicted value of the temperature of the autoclave at the next moment through feature extraction and fusion; taking minimization of a loss function as a target, training the model, and obtaining an optimal model by using a structured pruning strategy; and deploying an optimal model, and realizing off-line simulation and on-line prediction of the temperature of the autoclave through rolling time window iteration multi-step prediction. The method is high in accuracy, wide in universality, small in model parameter scale, high in reasoning efficiency and suitable for autoclave temperature prediction in complex scenes.

A sequence transmission method and an apparatus are provided, which may be applied to a downlink synchronization scenario, a random access scenario, a sensing scenario, a radar scenario, an integrated sensing and communication scenario, or the like, to increase sequence design diversity, and improve sequence design flexibility and target detection accuracy. The method includes: A transmit end apparatus determines N first sequences, and sends the N first sequences. An nth first sequence in the N first sequences is determined based on an nth second sequence in N second sequences, formula (I), am is a prime number, M is a positive integer greater than 1, and n=0,1, . . . , N−1. Each second sequence is a sequence in a Golay complementary pair GCP. The N second sequences include formula (II) first sub-sequence sets, each first sub-sequence set includes am second sub-sequence sets, each second sub-sequence set includes formula (III) second sequences, m=0,1, . . . , M−1, and a−1=1.

The invention discloses a diffusion model reasoning acceleration method based on optimal time step sequence search and knowledge distillation, which constructs a unified search space covering a time step sequence and corresponding model architecture configuration on the premise of not performing overall fine adjustment on a pre-trained diffusion model, and searches an optimal time step sequence. And designing a knowledge distillation training method based on the optimal diffusion time step sequence obtained by searching, constructing a main distillation loss function in a discrete time step subspace limited by the optimal diffusion time step sequence, and introducing adjacent time step consistency loss to relieve discrete errors introduced by a non-uniform time step span. In the reasoning stage, a teacher model is frozen, a distilled student model is combined with an efficient sampler ACDMS based on UniPC and fused with flow matching dynamic correction, finite step sampling is carried out on an optimal time step sequence, and reasoning acceleration under generation quality maintenance is achieved.

The invention discloses a cross-graph modeling three-dimensional RNA inverse folding method based on multi-structure comparison, and the method comprises the steps: carrying out the multi-structure comparison of a target RNA structure and an RNA structure database at an offline stage, obtaining a closest similar structure, carrying out the alignment processing, extracting node and edge features from the target RNA structure, and generating a training set through cross-structure coding; the training module is used for training a sequence decoder; in the online stage, through a trained sequence decoder, real-time input and cross-structure coded features are decoded, and after sequence probability distribution is obtained, a design sequence is obtained through temperature coefficient sampling. According to the method, the multi-structure comparison technology and the graph neural network are combined, and the accuracy and diversity of RNA sequence design are remarkably improved by utilizing the evolutionary conservative property of the target RNA structure and the homologous structure of the target RNA structure.

Techniques pertaining to mark sequence designs for enhanced long range (ELR) physical-layer protocol data unit (PPDU) in wireless communications are described. A wireless communication apparatus (e.g., a station (STA) ) generates an ELR PPDU and transmits the ELR PPDU in a wireless communication. The ELR PPDU may include a legacy preamble part, an ELR preamble part, and an ELR data symbol part. The ELR preamble part may include an ELR mark field (ELR-MARK) having two mark symbols, an ELR short training field (ELR-STF), an ELR long training field (ELR-LTF), and an ELR signal field (ELR-SIG), with the ELR-MARK transmitted after the legacy preamble part.

The invention discloses a time series prediction method based on a multi-modal enhanced large language model, which comprises the following steps: acquiring historical time series data, constructing a standardized input matrix and setting core task parameters; dividing data blocks through a sliding window mechanism, and converting the data blocks into uniform dimension features through linear embedding; a semantic prototype is generated based on a large language model vocabulary, and time sequence features and semantic features are fused through multi-head cross-attention; alternately splicing the data blocks and the corresponding semantic information, and constructing a self-multi-modal input sequence; designing a three-level structured prompt including context, task target and modal guidance, and fusing the three-level structured prompt with a multi-modal sequence; and training the lightweight model by adopting a frozen training strategy, and outputting a prediction result of a specified time step in the future. According to the method, through single-source data enhancement and prompt guidance, the time sequence reasoning capability of a small-parameter large language model is activated, high-precision prediction is guaranteed, efficient deployment is achieved, and the method adapts to long-term and short-term prediction tasks in the fields of electric power, traffic, meteorology and the like.

The application discloses an mRNA vaccine sequence design system and method based on a large language model intelligent agent, and the mRNA vaccine sequence design system based on the large language model intelligent agent comprises a user interaction and task analysis module, a task planning and intelligent agent scheduling module, an RNA design skill library and skill description module, a multi-target evaluation and closed-loop optimization module and a data and model management module. The application integrates a large language model and various mRNA sequence analysis and optimization tools, automatically completes mRNA vaccine sequence generation, evaluation and optimization for a target antigen in a computer environment, is suitable for candidate sequence design and screening of a preventive or therapeutic mRNA vaccine, improves design efficiency, and reduces the degree of dependence on manual experience and manual script arrangement.

This invention relates to an RNA sequence design method based on isovariant geometry and graph neural networks, comprising: first, RNA three-dimensional structure data and corresponding sequences; then, after dataset encapsulation and feature extraction, vector embeddings that can be processed by graph neural networks are generated; next, the data is fed into the Equiformer+GVP model for modeling, and the model outputs the probability distribution of four bases at each nucleotide position; finally, the bases are selected according to the probability distribution to form the predicted RNA sequence. Among these methods, E(3) isovariant constraints are introduced to achieve geometric consistency expression, and multi-scale geometric feature encoding is combined to enhance the model's understanding of local and global structures, significantly improving the model's physical rationality, prediction accuracy and universality; and the three-dimensional geometric skeleton of the RNA molecule is used as the only input, eliminating the dependence on secondary structures or external contact diagrams, and learning and generation are performed directly at the three-dimensional spatial coordinate level, thereby achieving significant improvements in physical consistency, geometric completeness and generalization ability.

The invention belongs to the technical field of multidisciplinary design, and discloses a data-driven AUV (Autonomous Underwater Vehicle) multidisciplinary design optimization method considering time consumption difference of simulation units. The method comprises the steps of generating an initial sample for an optimization object and segmenting the initial sample into design samples of each subject sequence; analyzing the high-fidelity evaluation average time consumption of the design samples of each subject sequence, calculating the amplification multiplying power of the design samples of other non-time-consuming subject sequences on the basis of a total time consumption consistency principle by taking the design sample of the time-consuming subject sequence as a benchmark to construct the design samples, and performing high-fidelity evaluation on the constructed design samples to construct a subject sequence library; establishing an agent model of each subject sequence based on the subject sequence library; and segmenting the multidisciplinary system of the optimization object into each disciplinary sequence, and carrying out rapid evaluation based on the corresponding agent model. According to the method, the problem of time isomerism evaluation in the multidisciplinary design optimization process is solved, and the optimization efficiency is remarkably improved.

The invention relates to an efficient frequency hopping sequence construction method based on a national cryptographic algorithm SM4, belongs to the technical field of information security, and solves the problems that in the prior art, frequency hopping sequence design is low in security and insufficient in anti-interference performance and cannot meet the domestic application requirement. Comprising the following steps: generating 32 rounds of round keys based on a user key Key; receiving system clock information TOD and the round key, and generating a pseudo-random sequence of frequency hopping communication through 32 rounds of iterative operation; and converting the pseudo-random sequence to obtain a frequency hopping sequence with an adjustable frequency slot number. The frequency hopping sequence with the adjustable frequency gap number is generated, the safety is high, the efficiency is high, the anti-interference capacity is high, and the domestic application requirement is met.

The application relates to an antibody sequence design method based on a given antigen-antibody complex structure. The method comprises a. a modeling stage, which uses a deep neural network to characterize the three-dimensional local structure environment of antibody amino acid residues and the amino acid sequence, and establishes a model; b. a sequence design stage, which uses the model to iteratively optimize the amino acid sites of each CDR region by inputting the main chain structure of the target antibody. In each iteration process, the model adjusts the amino acid type according to the previous update result, gradually converges the CDR region amino acid sequence that meets the target structure and functional requirements, and completes the sequence design of the antibody. The application combines the advantages of main chain design and protein language model, focuses on the amino acid sequence design of the antibody CDR region, and significantly improves the design accuracy and functionality by introducing antigen epitope information and an iterative optimization process.

The application belongs to the technical field of multidisciplinary design, and discloses a data-driven AUV multidisciplinary design optimization method considering time consumption difference of simulation units. The method comprises the following steps: generating initial samples for an optimization object and dividing the initial samples into design samples of each discipline sequence; analyzing high-fidelity evaluation average time consumption of the design samples of each discipline sequence, calculating expansion multiples of the design samples of other non-most time-consuming discipline sequences based on the design samples of the most time-consuming discipline sequence as a benchmark to construct design samples according to a total time consumption consistency principle, performing high-fidelity evaluation on the constructed design samples to construct a discipline sequence library; establishing a proxy model of each discipline sequence based on the discipline sequence library; and dividing a multidisciplinary system of the optimization object into each discipline sequence and performing rapid evaluation based on the corresponding proxy model. The method of the application significantly improves optimization efficiency by solving the problem of evaluation time heterogeneity in the process of multidisciplinary design optimization.

Various schemes are described for long training field (LTF) sequence design for distributed tone resource units (DRUs) for wider bandwidth in wireless communications. A device (e.g., a working station (STA)) generates a DRU LTF sequence of a DRU using a predefined DRU LTF base sequence. The device transmits the DRU LTF sequence of the DRU in wireless communications of a bandwidth of 160 MHz or more.

The invention discloses a communication system of an improved parallel cascade LDPC code structure based on a single encoder, and belongs to the technical field of aviation platform laser communication. Aiming at the bottlenecks that the traditional PCGC depends on double encoders, the hardware overhead is large, the decoding delay is high and the complementarity requirement on component codes is harsh, the system adopts a single encoder architecture design: firstly, a basic verification sequence is generated through a single LDPC encoder, and differential verification constraints are constructed through interleaving and repeating operations; meanwhile, a zero syndrome and non-zero syndrome dual-check sequence design is introduced, so that the minimum distance of a code word is increased; a decoding end fuses a normalized minimum sum algorithm and a hierarchical message updating and dynamic early stop mechanism, combines channel sensing soft information construction and a self-adaptive interleaving and redundancy allocation strategy, and adapts to composite channel characteristics of an aviation platform.

The application provides a method and system for generating and predicting the function of mRNA untranslated region sequence based on coding sequence. The method comprises: constructing a pre-training data set and a downstream task data set; using the pre-training data set to perform autoregressive training on the constructed conditional generation model to obtain a UTR sequence generation model; using the downstream task data set to fine-tune the UTR sequence generation model to obtain a downstream task function prediction model; generating candidate UTR sequences based on the UTR sequence generation model, and evaluating the generation ability of the UTR sequence generation model in the non-coding region sequence generation task; and predicting the function attribute of the UTR sequence based on the downstream task function prediction model, and evaluating the prediction ability of the downstream task function prediction model in multiple UTR related downstream tasks. The application considers the synergistic relationship between UTR and CDS in the UTR sequence generation process, and improves the efficiency and rationality of UTR sequence design.

The invention discloses an mRNA vaccine sequence design system and method based on large language model intelligent proxy. The mRNA vaccine sequence design system based on the large language model intelligent agent comprises a user interaction and task analysis module, a task planning and intelligent agent scheduling module, an RNA design skill library and skill description module, a multi-target evaluation and closed-loop optimization module and a data and model management module. According to the method, the large language model and various mRNA sequence analysis and optimization tools are integrated, mRNA vaccine sequence generation, evaluation and optimization aiming at the target antigen are automatically completed in a computer environment, the method is suitable for candidate sequence design and screening of prophylactic or therapeutic mRNA vaccines, the design efficiency is improved, and the design cost is reduced. And the degree of relying on artificial experience and manual script arrangement is reduced.

Various schemes pertaining to four times (4×) long-training field (LTF) sequence design for wide bandwidths in wireless communications are described. A processor of an apparatus generates an LTF of a physical-layer protocol data unit (PPDU) with a 78.125 kHz subcarrier spacing by using a predefined LTF base sequence. The apparatus then performs a wireless communication in a 240 MHz, 480 MHz or 640 MHz bandwidth with the PPDU.

The invention discloses a quantitative prediction method of anti-candida peptide activity and a sequence design method of high-activity anti-candida peptide, and relates to the technical field of bioinformatics and artificial intelligence crossing, the quantitative prediction method comprises the following steps: inputting an antibacterial peptide sequence of which the activity is to be predicted into a pre-trained activity quantitative prediction model to obtain an anti-candida MIC (Minimum Inhibitory Concentration) value; carrying out fusion processing on anti-candida multisource heterogeneous information in the antibacterial peptide sequence by utilizing a feature preprocessing module to obtain a preprocessed antibacterial peptide sequence; the core reasoning module comprises a random forest sub-model and a CatBoost sub-model, a first MIC predicted value is output through the random forest sub-model, and a second MIC predicted value is generated through the CatBoost sub-model; and converting the second MIC predicted value into an anti-candida MIC value in the antibacterial peptide by using a result conversion module. According to the method, the activity characteristics of the antibacterial peptide can be automatically learned from a large amount of sequence data, the activity of the antibacterial peptide is predicted, and the prediction accuracy and reliability are improved.

The embodiment of the invention relates to a communication method, equipment, a device and a medium. In accordance with an example embodiment of the present disclosure, a first device determines a first frequency from a first set of frequencies and a second frequency from a second set of frequencies. The first device also generates a complex exponential sequence based on the first frequency and the second frequency. The complex exponential sequence includes a first complex exponential subsequence associated with a first frequency and a second complex exponential subsequence associated with a second frequency, the first frequency being different from the second frequency. The first device also sends the complex exponential sequence to the second device in response to the first device being in an active state. In this way, more accurate, rapid and extensible user activity detection can be constructed through the leader sequence design disclosed by the invention.

The invention discloses an RNA sequence design method and device based on reinforcement learning and a submerged space diffusion model, equipment and a medium, and relates to the technical field of artificial intelligence. A mixed attention mechanism in the submerged space diffusion model is used for processing an initial RNA sequence and an RNA three-level structure; performing rotation position coding, activation function optimization and mask processing on the processed data to obtain a first RNA sequence; carrying out reward value calculation on the second RNA sequence by utilizing a reinforcement learning optimization engine and based on a preset reward mechanism to obtain each reward value, carrying out fusion and structural analysis on each reward value to obtain a target reward value, and determining a third RNA sequence corresponding to the target reward value from the second RNA sequence; and carrying out multi-objective optimization on the third RNA sequence by utilizing a Pareto frontier method to construct a target RNA, so that the model training time and resource consumption are reduced, and the RNA sequence design based on reinforcement learning and the submerged space diffusion model is realized.

This disclosure provides methods, components, devices and systems for tone mapping, enhanced long range (ELR)-mark sequence designs, and signal field designs for ELR transmissions. Some aspects more specifically relate to ELR-signal (ELR-SIG) field designs within a physical layer (PHY) protocol data unit (PPDU) associated with an ELR format to facilitate a parsing of ELR-data that follows an ELR-SIG field. Some further aspects more specifically relate to mechanisms according to which devices may support a set of ELR sequences associated with relatively low peak-to-average-power ratios (PAPRs) and usable to convey information that a receiving device may use to determine whether to perform an early drop of a received ELR PPDU. In some examples, various devices may support a set of ELR sequences with each ELR sequence of the set indicative of a respective basic service set (BSS) information value.

The invention discloses a robustness verification method and system based on a source code pre-training model and a storage medium. The method comprises the steps of obtaining a model and extracting code data features to obtain a token sequence; performing slicing processing on the token sequence; sampling token sequence slices and recombining to obtain a random token sequence; designing a classification loss function and finely adjusting the large pre-training model; taking the random token sequence as the input of a large pre-training model to obtain a classification label predicted by the model; calculating self-confident degree of classification according to classification labels obtained by multiple times of model prediction, and obtaining output prediction of the large pre-training model; constructing a new token sequence based on the editing distance; and calculating the robust radius according to the boundary of the confidence value of the predicted label and a dichotomy. According to the method, the problem that the robustness is difficult to guarantee due to the fact that an existing random smoothing method is difficult to perform unified modeling due to complex disturbance in a source code scene in an existing method and a large pre-training language model cannot perform fine processing on various features is solved.

Methods, systems, and devices for wireless communications are described. A user equipment (UE) may receive a signal that identifies a set of sequences for low-power (LP) signal transmissions from respective network entities in a set of network entities, each sequence defining a unique sequence of samples for the respective unique network entities when modulating the LP signal transmissions. The UE may receive the LP signal transmissions from the set of network entities in accordance with the set of sequences, a first set of samples of a first sequence from a serving network entity in the set of network entities comprises a different set of samples relative to a second set of samples of a second sequence from a non-serving network entity in the set of network entities. The UE may perform an operation with the serving network entity in accordance with the first set of samples of the first sequence.

This invention discloses a meme island optimization method and related apparatus for complementary sequence sets, belonging to the field of complementary sequence design. The method includes: constructing multiple island subpopulations that evolve in parallel; initializing a global archive for recording the history of search space region visits; performing population evolution in parallel on multiple islands; evaluating each candidate complementary sequence set according to a fitness function that incorporates region guidance; exchanging elite individuals among multiple islands when a preset migration condition is met; and outputting the optimal complementary sequence set found among all islands when an iteration condition is satisfied. By introducing region guidance based on the global archive into the fitness function, the search is actively guided to unexplored regions, systematically avoiding the search process from stagnating in the same basin, avoiding getting trapped in local optima, significantly improving global exploration capability, and reducing the risk of premature convergence.

The invention discloses a deep learning-based de novo enzyme design and catalytic activity prediction system. The system comprises a sequence generation module, a three-dimensional structure construction module, a molecular feature coding module, a catalytic activity prediction module, an iterative optimization module, a data enhancement module, a structural stability correction module and an experimental verification feedback module. The method comprises the following steps: a sequence generation module outputs an amino acid sequence; a construction module generates three-dimensional coordinates; a molecular feature coding module tensions coding structure information; a catalytic activity prediction module calculates an activity score; and the experimental verification feedback module forms closed-loop updating. The method has the beneficial effects that the de novo enzyme sequence design is accelerated, the prediction accuracy of the catalytic activity is improved, the structural stability is ensured, and the high efficiency and controllability of enzyme design are realized.

The application discloses a siRNA sequence design method and an activity prediction device based on deep learning, and relates to the field of artificial intelligence assisted biotechnology. The method comprises the following steps: performing global scanning on a target mRNA, intercepting the target mRNA into RNA sequence fragments according to a predetermined step and a predetermined window width, and enumerating all available candidate siRNA sequence fragments; filtering out sequence fragments with GC content outside a predetermined range in the candidate siRNA sequence fragments, and excluding 15-mer sequences with a frequency greater than a preset threshold in a transcriptome; inputting siRNA sequence data, mRNA sequence and structure data and experimental information into a deep learning model, outputting an activity prediction value, analyzing siRNA sequences with high activity ranking, and outputting potential siRNA sequences. The application can more accurately predict the activity of siRNA in a real experimental scenario and efficiently design siRNA sequences.