CN116189776A - Antibody structure generation method based on deep learning - Google Patents

Antibody structure generation method based on deep learning Download PDF

Info

Publication number
CN116189776A
CN116189776A CN202211640445.7A CN202211640445A CN116189776A CN 116189776 A CN116189776 A CN 116189776A CN 202211640445 A CN202211640445 A CN 202211640445A CN 116189776 A CN116189776 A CN 116189776A
Authority
CN
China
Prior art keywords
antibody
model
sequence
dimensional
convolution
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202211640445.7A
Other languages
Chinese (zh)
Inventor
周文
舒坤贤
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chongqing University of Post and Telecommunications
Original Assignee
Chongqing University of Post and Telecommunications
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Chongqing University of Post and Telecommunications filed Critical Chongqing University of Post and Telecommunications
Priority to CN202211640445.7A priority Critical patent/CN116189776A/en
Publication of CN116189776A publication Critical patent/CN116189776A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B40/00—ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B30/00—ICT specially adapted for sequence analysis involving nucleotides or amino acids
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • Medical Informatics (AREA)
  • General Health & Medical Sciences (AREA)
  • Biophysics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Evolutionary Computation (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Biology (AREA)
  • Software Systems (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Biotechnology (AREA)
  • Biomedical Technology (AREA)
  • Analytical Chemistry (AREA)
  • Bioethics (AREA)
  • Epidemiology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Chemical & Material Sciences (AREA)
  • Public Health (AREA)
  • Proteomics, Peptides & Aminoacids (AREA)
  • Databases & Information Systems (AREA)
  • Computational Linguistics (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention belongs to the technical field of biological information, and particularly relates to an antibody structure generation method based on deep learning, which comprises the following steps: acquiring sequence data to be predicted, and preprocessing the data; performing decomposition dimension reduction treatment on the preprocessed feature sequence by adopting a PCA method, wherein the process of decomposition dimension reduction treatment comprises the steps of calculating a covariance matrix of the sequence feature, and calculating a feature value of the covariance matrix; sorting the feature values, screening out corresponding feature vectors according to the sorted feature values, and taking the screened features as sequence features after decomposing and reducing the dimension; inputting the decomposed and dimension-reduced sequence data into a trained improved neural network model to obtain an antibody structure prediction result; according to the invention, aiming at the antibody structure, all heavy atom information is extracted from each amino acid and used as a label, so that the space structure of the antibody can be expressed more accurately.

Description

Antibody structure generation method based on deep learning
Technical Field
The invention belongs to the technical field of biological information, and particularly relates to an antibody structure generation method based on deep learning.
Background
Antibodies (antibodies) are proteins that play a critical role in humoral immunity. The structure of an antibody consists of two heavy (H) and two light (L) polypeptide chains. Their primary function is to bind with high affinity and specificity to foreign invaders, and antibodies are generated to mediate immune responses against foreign pathogens as part of adaptive immunity. With the development of artificial intelligence (Artificial Intelligence, AI) technology, particularly Deep Learning (DL) technology, and the accumulation of antibody structure data, deep Learning-based antibody structure prediction has made tremendous progress. In drug discovery and protein engineering, a primary goal is to design an antibody that can perform useful functions as a therapeutic drug, which requires the analysis of antibody structure, which is time consuming, laborious, and very expensive in traditional methods. In recent years, deep learning technology has been widely used in the antibody field, under the push of achievements in the fields of computer vision, natural language processing, and the like.
An antibody monomer consists of two heavy chains (H) and two light chains (L), which are joined together to form a "Y" shaped structure. They have an NH2 terminal variable region or antigen binding fragment (Fab) and a COOH constant region or crystallizable fragment (Fc), with different parts of the structure having different functions. The variable regions determine the idiotype of the antibody and have affinity for pathogen antigens. The constant region performs other immune-related functions such as complement fixation, macrophage binding, and the antibodies can be labeled with isotopes in the study.
The antigen binding site of an antibody is present in the Fab region, consisting essentially of six Complementarity Determining Regions (CDRs): heavy and light chains are three each. Antibodies can recognize millions of different antigens due to the diversity generated during antibody synthesis. The heavy and light chains undergo genetic recombination in which different combinations of genes occur to increase the diversity of the final gene product. Five of the six antibody CDR regions typically fold into one of several classical conformations, which can be predicted by existing methods, while the remaining loop (CDR H3) is not well predicted by conventional methods due to the highly variable structure observed experimentally. In order to computationally analyze or predict the effectiveness of an antibody, it is often necessary to generate a three-dimensional model. Conventional structural determination methods due to X-ray crystallography, nuclear Magnetic Resonance (NMR), and low temperature electron microscopy (CryoEM) are laborious, time-consuming, and expensive.
Existing machine learning methods for protein structure prediction have focused on co-evolution based methods. The accuracy of these methods depends on the number of homologous protein sequences available in the database. For many proteins, especially those that do not have sufficient sequence homologs, such as antibodies. The method based on the co-evolution has low machine learning effect and poor predicted structural accuracy.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides an antibody structure generation method based on deep learning, which comprises the following steps: acquiring sequence data to be predicted, and preprocessing the data; inputting the preprocessed sequence data into a trained improved neural network model to obtain a predicted antibody structure; constructing an antibody according to the predicted antibody structure; an improved neural network model dimension improved ResNet-RCCA model;
the process of training the improved neural network model includes:
s1: acquiring an original antibody data set, wherein the original antibody data is sequence information of an antibody, and comprises an antibody primary structure, an antibody secondary structure and an antibody tertiary structure;
s2: converting antibody sequence information in an original antibody data set into matrix data, and taking the matrix data as sequence characteristics; acquiring atomic coordinate information of an antibody structure, and taking the information as a structure label;
s3: adopting a PCA method to decompose and dimension-reduce the sequence characteristics;
s4: inputting the structural tag and the sequence characteristics after decomposing and dimension reducing into an improved ResNet-RCCA model to obtain an antibody structure prediction result;
s5: and calculating a loss function of the model according to the antibody structure prediction result, continuously adjusting model parameters, and finishing model training when the square loss function is minimum.
Preferably, the process of converting antibody sequence information in the original antibody dataset into matrix data comprises: and inputting the antibody sequence information into a pre-trained protein semantic model ESM-1B to obtain matrix data with the antibody sequence information, wherein the protein semantic model ESM-1B is a high-model-capacity transducer model which takes a protein sequence as input and is subjected to super-parameter optimization training.
Preferably, the process of decomposing and dimension-reducing the sequence features by adopting a PCA method comprises the following steps: calculating a covariance matrix of the sequence characteristics, and calculating characteristic values of the covariance matrix; and sorting the feature values, screening out corresponding feature vectors according to the sorted feature values, and taking the screened features as sequence features after decomposition and dimension reduction.
Preferably, the improved ResNet-RCCA model comprises a ResNet model and 6 RCCA modules; the ResNet model consists of a one-dimensional residual error convolution network and a two-dimensional residual error convolution network, each one-dimensional convolution block consists of two convolution layers and two pooling layers, and the convolution kernel size of the convolution layers is 5*5; the dimension residual error convolution network consists of a two-dimensional convolution layer, a pooling layer and 25 two-dimensional convolution blocks, wherein the two-dimensional convolution block consists of two convolution layers and two pooling layers, the convolution kernel size of the convolution layer is 3 x 3, and the RCCA module is arranged behind the two-dimensional convolution layer.
Preferably, the processing of the structural tag and the decomposed reduced-dimension sequence features using the modified ResNet-RCCA model includes: inputting the decomposed and dimension-reduced sequence features into a one-dimensional residual convolution network to obtain a one-dimensional feature map; splicing the one-dimensional feature map with the input sequence features, and inputting the spliced feature map into a two-dimensional residual convolution network to obtain a two-dimensional feature map; inputting the two-dimensional feature map into an RCCA module, learning time sequence information of the two-dimensional feature map through a residual convolution network, and learning spatial information of the two-dimensional feature map by adopting a cross attention mechanism; and predicting the antibody structure according to the time sequence information and the space information of the two-dimensional characteristic diagram to obtain a predicted antibody structure.
In order to achieve the above object, the present invention further provides a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor, and the computer program when executed by the processor implements any one of the above methods for generating an antibody structure based on deep learning.
In order to achieve the above object, the present invention also provides an antibody structure generating device based on deep learning, comprising a processor and a memory; the memory is used for storing a computer program; the processor is connected with the memory and is used for executing the computer program stored in the memory so as to enable the antibody structure generating device based on the deep learning to execute any one of the antibody structure generating methods based on the deep learning.
The invention has the following beneficial effects:
the invention fully utilizes the sequence information and the amino acid characteristic information of the antibody, and is not applicable to the co-evolution information to predict the structure of the antibody; according to the invention, aiming at the antibody structure, all heavy atom information is extracted from each amino acid and used as a label, so that the space structure of the antibody can be expressed more accurately; the invention provides a novel hybrid neural network, which fuses ResNet and a cross attention mechanism, and the fused network can more effectively extract and learn protein sequence information, so that the prediction accuracy is improved, and the hybrid neural network has good generalization performance for different data sets.
Drawings
FIG. 1 is a flow chart of a method for predicting antibody structure based on deep learning according to the present invention;
FIG. 2 is a diagram of a model framework structure of the present invention;
fig. 3 is a block diagram of an improved neural network of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
An antibody structure generation method based on deep learning, as shown in fig. 1-2, comprises the following steps: acquiring sequence data to be predicted, and preprocessing the data; inputting the preprocessed sequence data into a trained improved neural network model to obtain a predicted antibody structure; constructing an antibody according to the predicted antibody structure; improved neural network model the improved ResNet-RCCA model.
The process of training the improved neural network model includes:
s1: acquiring an original antibody data set, wherein the original antibody data is sequence information of an antibody, and comprises an antibody primary structure, an antibody secondary structure and an antibody tertiary structure;
s2: converting antibody sequence information in an original antibody data set into matrix data, and taking the matrix data as sequence characteristics; acquiring atomic coordinate information of an antibody structure, and taking the information as a structure label;
s3: adopting a PCA method to decompose and dimension-reduce the sequence characteristics;
s4: inputting the structural tag and the sequence characteristics after decomposing and dimension reducing into an improved ResNet-RCCA model to obtain an antibody structure prediction result;
s5: and calculating a loss function of the model according to the antibody structure prediction result, continuously adjusting model parameters, and finishing model training when the square loss function is minimum.
The process of decomposing and dimension-reducing processing of the sequence features by adopting the PCA method comprises the following steps: calculating a covariance matrix of the sequence features, performing feature value decomposition on the covariance matrix, obtaining feature values and feature vectors of the covariance matrix, and sequencing the feature values from large to small; and screening out corresponding feature vectors according to the sorted feature values, and taking the screened features as sequence features after decomposing and reducing the dimension. Screening out the corresponding feature vectors is to screen out the first 100 feature vectors.
The antibody primary structure data is converted into matrix data by using a protein semantic model as a sequence characteristic, and heavy atom coordinate information (C alpha atom, C beta atom, N atom and O atom) is extracted from the antibody tertiary structure data to be used as a structure label.
In this example, the SAbdab database is used, and the database has corresponding sound data and image data, and the process of converting the data in the training set includes:
the antibody structure data contains information on the primary structure, secondary structure, and tertiary structure of the antibody. Processing the antibody structure data, extracting antibody primary structure information and antibody tertiary structure information, adopting a pre-trained ESM-1b model for the antibody primary structure, converting the sequence data into matrix data, performing dimension reduction processing on the matrix data by using a PAC dimension reduction method, and then transposing the matrix data to fix the dimension of the matrix data.
For each amino acid in the antibody tertiary structure, heavy atom coordinate information (C alpha atom, C beta atom, N atom, O atom) is extracted as a structural tag, representing the antibody tertiary structure.
A pre-trained protein semantic model ESM-1b was used. ESM-1b is actually a high-model-capacity transducer trained by super parameter (hyper parameter) optimization with protein sequences as inputs. After training, the model outputs information of secondary tertiary structure, function, homology and the like of the implicit protein in the characteristic representation (presentation), and the information can be displayed through linear projection (linear projection).
As shown in FIG. 3, the modified ResNet-RCCA model includes a ResNet model and 6 RCCA modules. The ResNet model is composed of a one-dimensional residual convolution network and a two-dimensional residual convolution network. The one-dimensional residual convolution network consists of one-dimensional convolution layer, pooled layers and 3 one-dimensional convolution blocks, each one-dimensional convolution block consists of two convolution layers and two pooled layers, and the convolution kernel size of the convolution layers is 5*5. The two-dimensional residual error convolution network consists of a two-dimensional convolution layer, a pooling layer and 25 two-dimensional convolution blocks, wherein the two-dimensional convolution block consists of two convolution layers and two pooling layers, the convolution kernel size of the convolution layer is 3 x 3, and the RCCA module is arranged behind the two-dimensional convolution layer.
The improved ResNet-RCCA model comprises a ResNet model and 6 RCCA modules: and processing the primary structure data of the antibody through a protein language model to obtain sequence information, inputting the sequence information as characteristics, learning time sequence information by using a residual convolution network and a cross attention mechanism, predicting the structure of the antibody, and finally obtaining a prediction result.
The process of processing the structural labels and the sequence features after decomposing and dimension reduction by adopting the improved ResNet-RCCA model comprises the following steps: inputting the decomposed and dimension-reduced sequence features into a one-dimensional residual convolution network to obtain a one-dimensional feature map; splicing the one-dimensional feature map with the input sequence features, and inputting the spliced feature map into a two-dimensional residual convolution network to obtain a two-dimensional feature map; inputting the two-dimensional feature map into an RCCA module, learning time sequence information of the two-dimensional feature map through a residual convolution network, and learning spatial information of the two-dimensional feature map by adopting a cross attention mechanism; and predicting the antibody structure according to the time sequence information and the space information of the two-dimensional characteristic diagram to obtain a predicted antibody structure.
When the neural network is constructed, the process of calculating the loss function of the model comprises the steps of making cross entropy of the input joint vector characteristics and the actual labels of the vectors, and using the cross entropy as the loss function of the model; the loss function expression of the model is:
Figure BDA0004008666090000061
wherein y' i Is the actual label; y is i Training a sample for the current model; i predicted tag probability.
In an embodiment of the present invention, the present invention further includes a computer readable storage medium having a computer program stored thereon, which when executed by a processor, implements any of the above-described deep learning antibody structure generation methods.
An antibody structure generating device based on deep learning comprises a processor and a memory; the memory is used for storing a computer program; the processor is connected with the memory and is used for executing the computer program stored in the memory so that the antibody structure generating device based on deep learning can execute any one of the antibody structure generating methods based on deep learning
Specifically, the memory includes: various media capable of storing program codes, such as ROM, RAM, magnetic disk, U-disk, memory card, or optical disk.
Preferably, the processor may be a general-purpose processor, including a central processing unit (Central Processing Unit, abbreviated as CPU), a network processor (Network Processor, abbreviated as NP), etc.; but also digital signal processors (Digital Signal Processor, DSP for short), application specific integrated circuits (Application Specific Integrated Circuit, ASIC for short), field programmable gate arrays (Field Programmable Gate Array, FPGA for short) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.
While the foregoing is directed to embodiments, aspects and advantages of the present invention, other and further details of the invention may be had by the foregoing description, it will be understood that the foregoing embodiments are merely exemplary of the invention, and that any changes, substitutions, alterations, etc. which may be made herein without departing from the spirit and principles of the invention.

Claims (8)

1. An antibody structure generation method based on deep learning, which is characterized by comprising the following steps: acquiring sequence data to be predicted, and preprocessing the data; inputting the preprocessed sequence data into a trained improved neural network model to obtain a predicted antibody structure; constructing an antibody according to the predicted antibody structure; an improved neural network model dimension improved ResNet-RCCA model;
the process of training the improved neural network model includes:
s1: acquiring an original antibody data set, wherein the original antibody data is sequence information of an antibody, and comprises an antibody primary structure, an antibody secondary structure and an antibody tertiary structure;
s2: converting antibody sequence information in an original antibody data set into matrix data, and taking the matrix data as sequence characteristics; acquiring atomic coordinate information of an antibody structure, and taking the information as a structure label;
s3: adopting a PCA method to decompose and dimension-reduce the sequence characteristics;
s4: inputting the structural tag and the sequence characteristics after decomposing and dimension reducing into an improved ResNet-RCCA model to obtain an antibody structure prediction result;
s5: and calculating a loss function of the model according to the antibody structure prediction result, continuously adjusting model parameters, and finishing model training when the square loss function is minimum.
2. The method of claim 1, wherein the step of converting the antibody sequence information in the original antibody dataset into matrix data comprises: and inputting the antibody sequence information into a pre-trained protein semantic model ESM-1B to obtain matrix data with the antibody sequence information, wherein the protein semantic model ESM-1B is a high-model-capacity transducer model which takes a protein sequence as input and is subjected to super-parameter optimization training.
3. The method for generating the antibody structure based on deep learning according to claim 1, wherein the process of performing decomposition and dimension reduction on the sequence features by adopting a PCA method comprises the following steps: calculating a covariance matrix of the sequence features, performing feature value decomposition on the covariance matrix, obtaining feature values and feature vectors of the covariance matrix, and sequencing the feature values from large to small; and screening out corresponding feature vectors according to the sorted feature values, and taking the screened features as sequence features after decomposing and reducing the dimension.
4. The method for generating the antibody structure based on deep learning according to claim 1, wherein the improved ResNet-RCCA model comprises a ResNet model and 6 RCCA modules; the ResNet model consists of a one-dimensional residual error convolution network and a two-dimensional residual error convolution network, each one-dimensional convolution block consists of two convolution layers and two pooling layers, and the convolution kernel size of the convolution layers is 5*5; the two-dimensional residual error convolution network consists of a two-dimensional convolution layer, a pooling layer and 25 two-dimensional convolution blocks, wherein the two-dimensional convolution block consists of two convolution layers and two pooling layers, the convolution kernel size of the convolution layer is 3 x 3, and the RCCA module is arranged behind the two-dimensional convolution layer.
5. A method of generating an antibody structure based on deep learning according to claim 3, wherein the processing of the structural tag and the decomposed and dimension reduced sequence features using the modified res net-RCCA model comprises: inputting the decomposed and dimension-reduced sequence features into a one-dimensional residual convolution network to obtain a one-dimensional feature map; splicing the one-dimensional feature map with the input sequence features, and inputting the spliced feature map into a two-dimensional residual convolution network to obtain a two-dimensional feature map; inputting the two-dimensional feature map into an RCCA module, learning time sequence information of the two-dimensional feature map through a residual convolution network, and learning spatial information of the two-dimensional feature map by adopting a cross attention mechanism; and predicting the antibody structure according to the time sequence information and the space information of the two-dimensional characteristic diagram to obtain a predicted antibody structure.
6. The method for generating an antibody structure based on deep learning according to claim 1, wherein the model has a loss function expression of:
Figure FDA0004008666080000021
wherein y' i Is the actual label; y is i Training a sample for the current model; i predicted tag probability.
7. A computer-readable storage medium having stored thereon a computer program, characterized in that the computer program is executed by a processor to implement the deep learning based antibody structure generating method of any one of claims 1 to 6.
8. An antibody structure generating device based on deep learning is characterized by comprising a processor and a memory; the memory is used for storing a computer program; the processor is connected to the memory for executing the computer program stored in the memory, so that the deep learning-based antibody structure generating device executes the deep learning-based antibody structure generating method of any one of claims 1 to 6.
CN202211640445.7A 2022-12-20 2022-12-20 Antibody structure generation method based on deep learning Pending CN116189776A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211640445.7A CN116189776A (en) 2022-12-20 2022-12-20 Antibody structure generation method based on deep learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211640445.7A CN116189776A (en) 2022-12-20 2022-12-20 Antibody structure generation method based on deep learning

Publications (1)

Publication Number Publication Date
CN116189776A true CN116189776A (en) 2023-05-30

Family

ID=86431774

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211640445.7A Pending CN116189776A (en) 2022-12-20 2022-12-20 Antibody structure generation method based on deep learning

Country Status (1)

Country Link
CN (1) CN116189776A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116844632A (en) * 2023-07-07 2023-10-03 北京分子之心科技有限公司 Method and device for determining antibody sequence structure

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116844632A (en) * 2023-07-07 2023-10-03 北京分子之心科技有限公司 Method and device for determining antibody sequence structure
CN116844632B (en) * 2023-07-07 2024-02-09 北京分子之心科技有限公司 Method and device for determining antibody sequence structure

Similar Documents

Publication Publication Date Title
Jin et al. Predicting organic reaction outcomes with weisfeiler-lehman network
CN110910951B (en) Method for predicting free energy of protein and ligand binding based on progressive neural network
Zhang et al. A survey on graph diffusion models: Generative ai in science for molecule, protein and material
Yu et al. Cyclic differentiable architecture search
CN114503203A (en) Protein structure prediction from amino acid sequences using self-attention neural networks
Ketata et al. Diffdock-pp: Rigid protein-protein docking with diffusion models
Li et al. Protein loop modeling using deep generative adversarial network
CN114464247A (en) Method and device for predicting binding affinity based on antigen and antibody sequences
CN116189776A (en) Antibody structure generation method based on deep learning
CN114913917B (en) Drug target affinity prediction method based on digital twin and distillation BERT
CN112420125A (en) Molecular attribute prediction method and device, intelligent equipment and terminal
Yu et al. Learning protein multi-view features in complex space
CN113257357B (en) Protein residue contact map prediction method
Jin et al. Prediction of protein secondary structure based on an improved channel attention and multiscale convolution module
CN113611354B (en) Protein torsion angle prediction method based on lightweight deep convolutional network
Chen et al. Diversified multiscale graph learning with graph self-correction
Papamarkou et al. Position Paper: Challenges and Opportunities in Topological Deep Learning
Li et al. ctP 2 ISP: Protein–Protein Interaction Sites Prediction Using Convolution and Transformer With Data Augmentation
Einav et al. Quantitatively visualizing bipartite datasets
Hahanov et al. Similarity–Difference Analysis and Matrix Fault Diagnosis of SoC-components
Sunny et al. Deepbindppi: epitope-paratope prediction using attention based graph convolutional network
CN115458048B (en) Antibody humanization method based on sequence coding and decoding
Kalimeris et al. Deep Learning on Point Clouds for 3D Protein Classification Based on Secondary Structure
Antony et al. Towards Protein Tertiary Structure Prediction Using LSTM/BLSTM
Aenuga et al. Identification of Protein Structure Using Hybrid Deep Learning Model

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination