CN117116384B - Targeted induction type medical molecular structure generation method - Google Patents
Targeted induction type medical molecular structure generation method Download PDFInfo
- Publication number
- CN117116384B CN117116384B CN202311359357.4A CN202311359357A CN117116384B CN 117116384 B CN117116384 B CN 117116384B CN 202311359357 A CN202311359357 A CN 202311359357A CN 117116384 B CN117116384 B CN 117116384B
- Authority
- CN
- China
- Prior art keywords
- medical
- sequence
- molecular
- medical molecular
- molecular structure
- 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.)
- Active
Links
- 238000002910 structure generation Methods 0.000 title claims abstract description 85
- 238000000034 method Methods 0.000 title claims abstract description 42
- 230000006698 induction Effects 0.000 title claims abstract description 28
- 239000003814 drug Substances 0.000 claims abstract description 127
- 239000011159 matrix material Substances 0.000 claims abstract description 79
- 238000000605 extraction Methods 0.000 claims abstract description 41
- 238000005457 optimization Methods 0.000 claims abstract description 28
- 230000009471 action Effects 0.000 claims abstract description 26
- 230000006870 function Effects 0.000 claims description 34
- 238000004364 calculation method Methods 0.000 claims description 26
- 238000004220 aggregation Methods 0.000 claims description 25
- 230000002776 aggregation Effects 0.000 claims description 22
- 229940079593 drug Drugs 0.000 claims description 19
- 238000012549 training Methods 0.000 claims description 18
- 230000004927 fusion Effects 0.000 claims description 15
- 238000012545 processing Methods 0.000 claims description 11
- 230000004913 activation Effects 0.000 claims description 9
- 230000008447 perception Effects 0.000 claims description 7
- 230000007246 mechanism Effects 0.000 claims description 6
- 238000011176 pooling Methods 0.000 claims description 4
- 239000000284 extract Substances 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 239000000203 mixture Substances 0.000 claims 1
- 230000008685 targeting Effects 0.000 abstract description 17
- 238000004891 communication Methods 0.000 description 8
- 238000003860 storage Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 6
- 238000009826 distribution Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 239000002547 new drug Substances 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 238000012827 research and development Methods 0.000 description 2
- 108090000790 Enzymes Proteins 0.000 description 1
- 102000004190 Enzymes Human genes 0.000 description 1
- 102000004310 Ion Channels Human genes 0.000 description 1
- 238000004378 air conditioning Methods 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000802 evaporation-induced self-assembly Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000010534 mechanism of action Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 108090000623 proteins and genes Proteins 0.000 description 1
- 238000003041 virtual screening Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/50—Molecular design, e.g. of drugs
Abstract
The invention relates to the technical field of medical structure generation, and discloses a method for generating a medical molecular structure by targeted induction, which comprises the following steps: constructing a medical molecular sequence remote dependency information extraction model to extract local characteristic information of the medical molecular sequence; constructing a medicine molecular site preference matrix by combining targeted induction; carrying out optimization solution on the constructed medical molecular structure generation model; and generating the medical molecular structure based on the optimal medical molecular structure generation model. The invention constructs the corresponding site location information representing the medicine molecular structure and the medicine action target based on the targeting induction mode of the medicine action target, generates a medicine molecular site preference matrix, takes the extracted medicine molecular sequence local characteristic information, medicine molecular sequence structure and medicine molecular site preference matrix as model input, generates the medicine molecular structure related to the medicine action target and the medicine molecular sequence local characteristic information, and realizes the automatic generation of the medicine molecular structure of targeting induction.
Description
Technical Field
The invention relates to the technical field of generation of medical molecular structures, in particular to a method for generating a medical molecular structure by targeted induction.
Background
Biological medicine molecules play an extremely important role in the course of life activities. To study the mechanism of action and the exact function of these biomedical molecules, knowledge of their three-dimensional structure is required. However, in the research and development process of new drugs, the traditional experimental method for structural determination is usually high in cost and more in time consumption, and the pilot medicine analysis structure based on the computer virtual screening is beneficial to utilizing the advantages of high speed and automation of the computer, so that the research and development efficiency of new drugs is improved, and the cost is reduced. In view of the above, the invention provides a method for generating a medical analysis structure by targeted induction, which realizes intelligent generation of a medical molecular structure by medical molecular node preference and graph convolution neural network technology and provides convenience for medical preparation.
Disclosure of Invention
In view of the above, the present invention provides a method for generating a targeting-induced pharmaceutical molecular structure, which aims at: 1) Extracting local characteristic information of a medicine molecular sequence by combining attention weights of medicine molecular sequences in the same medicine category, constructing corresponding position point information representing the medicine molecular structure and the medicine action target point based on a targeting induction mode of the medicine action target point, generating a medicine molecular position preference matrix, inputting the extracted medicine molecular sequence local characteristic information, medicine molecular sequence structure and medicine molecular position preference matrix serving as models, carrying out correlation operation on the local characteristic information, sequence structure and position preference matrix of each medicine molecular sequence serving as point information of a graph convolution neural network, generating a medicine molecular structure related to the medicine action target point and the medicine molecular sequence local characteristic information, and realizing targeting induction medicine molecular structure automatic generation; 2) And constructing an objective function for model parameter training by combining the probability distribution of the generated medical molecular structure and the probability distribution of the medical molecular structure of the same medical category, and carrying out iterative calculation on the objective function by adopting a mode of combining gradient descent of the objective function and linear parameter correction to quickly obtain effective model parameters and construct and obtain a model for generating the medical molecular structure.
The invention provides a targeting-induced medical molecular structure generation method, which comprises the following steps:
s1: constructing a medical molecular sequence set, and constructing a medical molecular sequence remote dependency information extraction model to extract medical molecular sequence local characteristic information, wherein the model takes a medical molecular sequence as input and takes medical molecular sequence local characteristic information as output;
s2: constructing a medical molecular site preference matrix by combining with targeted induction, wherein the medical molecular site preference matrix comprises site order information of medical molecular sequences in a medical molecular sequence set;
s3: constructing a medical molecular structure generation model, wherein the medical molecular structure generation model takes extracted medical molecular sequence local characteristic information, medical molecular sequence structure and medical molecular site preference matrix as input, and takes generated medical molecular structure as output;
s4: carrying out optimization solution on the constructed medical molecular structure generation model to obtain an optimal medical molecular structure generation model;
s5: and generating the medical molecular structure based on the optimal medical molecular structure generation model.
As a further improvement of the present invention:
optionally, constructing a remote dependency information extraction model of the medical molecule sequence in the step S1 includes:
Constructing a medical molecular sequence remote dependency information extraction model, wherein the medical molecular sequence remote dependency information extraction model takes a medical molecular sequence as input and takes local characteristic information of the medical molecular sequence as output, and the medical molecular sequence remote dependency information extraction model comprises an input layer, a characteristic calculation layer, a local characteristic information fusion layer and an output layer;
the input layer is used for receiving the medicine molecular sequences of the same medicine category, normalizing the medicine molecular sequences, and transmitting the normalized medicine molecular sequences to the feature calculation layer;
the feature computation layer contains 3A convolution kernel of matrix size for performing convolution calculation on the normalized medical molecule sequence to obtain medical moleculeSequence features;
the local feature information fusion layer comprises a full-connection layer and a self-attention network layer, and is respectively used for extracting local feature information of the medical molecular sequence features and carrying out local feature information fusion combined with a self-attention mechanism to form the medical molecular sequence local feature information;
the output layer is used for outputting the local characteristic information of the medical molecule sequence corresponding to the medical molecule sequence.
Optionally, the extracting local characteristic information of the medical molecular sequence in the step S1 includes:
Obtaining the medical molecular sequences of different medical categories to form a medical molecular sequence set:
wherein:
a set of medical molecule sequences representing an nth medical category, N representing the number of medical categories;
representing the set of medical molecule sequences->Middle->Group medical molecule sequence,/->，Representing the set of medical molecule sequences->Total number of Chinese medicinal molecule sequences; in an embodiment of the present invention, the medical molecule sequence represents a medical molecule structural expression of a medical molecule;
extracting local characteristic information of the medical molecular sequence by utilizing a medical molecular sequence remote dependency information extraction model, wherein the medical molecular sequence setThe extraction flow of the local characteristic information of the medical molecular sequence is as follows:
s11: the input layer receives the medicine molecule sequence setAnd normalizing the medicinal molecular sequences, wherein the normalization is single-heat encoding of medicinal molecular sequences, and the medicinal molecular sequences are collected +.>Chinese medicine molecular sequenceNormalized result of (2) is->；
S12: the characteristic calculation layer carries out convolution calculation processing on the normalized medical molecular sequence, whereinThe convolution calculation processing flow of (1) is as follows:
wherein:
weight parameters respectively representing three convolution kernels in feature calculation layer A matrix;
respectively representing bias parameters of three convolution kernels in a feature calculation layer;
representing an activation function; in the embodiment of the invention, the selected activation function is a ReLU activation function;
representation->Corresponding medical molecule sequence->Is characterized by the sequence of the pharmaceutical molecule;
s13: the local feature information fusion layer extracts local feature information of the medical molecular sequence features, wherein the medical molecular sequence featuresThe local feature information extraction result of (a) is:
wherein:
indicating transpose,/->Representing the sequence characteristics of a pharmaceutical molecule>Is a local feature information of (1);
s14: according to the local feature information extraction results of different medicine molecular sequence features, local features combined with self-attention mechanism are carried outInformation fusion, forming local characteristic information of medicine molecular sequence, whereinThe corresponding medical molecular sequence local characteristic information has the following formula:
wherein:
an exponential function that is based on a natural constant;
representing the sequence of a pharmaceutical molecule->Local characteristic information of the corresponding medical molecule sequence;
s15: the output layer outputs the local characteristic information of the medicine molecule sequence corresponding to the medicine molecule sequence.
Optionally, the constructing a medical molecular site preference matrix by combining the targeted induction in the step S2 includes:
Constructing a medical molecular site preference matrix by combining targeted induction, wherein the medical molecular site preference matrix comprises site order information of medical molecular sequences in a medical molecular sequence set, and the medical molecular sequence set combined with the nth medical categoryThe medical molecular site preference matrix of (2) is:
wherein:
representing the set of medical molecule sequences->Is a matrix of preference for the medical molecular site of (2)>Representing the sequence of a pharmaceutical molecule->Is a site preference matrix of (a);
representing the sequence of a pharmaceutical molecule->Middle->Preference value of segment molecular structure and K-th drug action target point, wherein the preference value is +.>A preference value of 1 indicates that the segment of molecular structure is conducive to drug action with a drug action target, and a preference value of 0 indicates that the segment of molecular structure has no obvious relationship with the drug action target. In the embodiment of the invention, the drug action target point is a binding site in the same body of the drug, the binding site comprises enzyme, gene locus, ion channel and the like, the medicine molecular sequence represents a medicine molecular structure expression of medicine molecules, and each section of molecular structure is obtained by segmenting the medicine molecular structure expression.
Optionally, constructing a medical molecular structure generating model in the step S3 includes:
Constructing a medical molecular structure generation model, wherein the medical molecular structure generation model takes extracted medical molecular sequence local characteristic information, medical molecular sequence structures and medical molecular site preference matrixes of the same medical category as inputs, and takes generated medical molecular structures as outputs, and the medical molecular structure generation model comprises an input layer, a coding layer, a decoding layer and an output layer;
the input layer is used for receiving the local characteristic information, the medical molecular sequence structure and the medical molecular site preference matrix of the medical molecular sequence of the same medical category;
the coding layer is used for sequentially carrying out aggregation updating, global information extraction and multi-channel perception coding on the receiving result to generate a medicine molecular structure coding representation;
the decoding layer is used for decoding the coded representation of the medical molecular structure to obtain the medical molecular structure;
the output layer is used for outputting the medical molecular structure. In the embodiment of the invention, the generated medical category corresponding to the medical molecular structure is the same as the medical category corresponding to the input medical molecular sequence local characteristic information and the medical molecular site preference matrix.
Optionally, in the step S4, performing optimization solution on the constructed medical molecular structure generating model, including:
Carrying out optimization solving on the constructed medical molecular structure generating model, wherein model parameters to be optimized solved in the medical molecular structure generating model comprise aggregation matrix parameters in a coding layer, matrix parameters in a decoding layer and bias parameters;
the method comprises the steps of obtaining medical molecular sequences of group D and medical category to form medical molecular sequence local characteristic information, medical molecular sequence structure and medical molecular site preference matrix to form training data set data, and constructing an optimization objective function of a medical molecular structure generation model:
wherein:
an optimized objective function representing a model of the generation of a medical molecular structure, < >>Model parameters to be optimally solved for representing a medical molecular structure generation model;
the medical molecular structure set generated by taking the medical molecular sequence local characteristic information, the medical molecular sequence structure and the medical molecular site preference matrix in the training data set data as the input of a medical molecular structure generation model is represented, wherein the model parameters of the medical molecular structure generation model are->；
Representing the set of medical molecular structures->The single thermal code of the traditional Chinese medicine molecular structure d represents the maximum value of cosine similarity represented by the single thermal code of any medicine molecular sequence structure in the training data set data;
Representing a probability sequence consisting of probability values of each segment of the medical molecular structure d output by the model, +.>Representing a probability sequence formed by the occurrence probability of each section of molecular structure in the medical molecular structure d in the training data set data;
and carrying out optimization solving on an optimization objective function of the medical molecular structure generation model to obtain model parameters, wherein the solving flow of the model parameters is as follows:
s41: random generation of initial model parametersSetting the current iteration number of the model parameters as s and the maximum iteration numberThe generation number is Max, and the s-th iteration result of the model parameter is +.>The initial value of s is;
s42: if it isLess than the preset iteration threshold, will +.>Model parameters obtained as final iteration and based on model parameters +.>Constructing to obtain an optimal medical molecular structure generation model, otherwise turning to step S43;
s43: calculating to obtain the iteration step of the s-th iteration of the model parameter：
Wherein:
representing the step size +.>Iteration parameters of (a);
s44: performing iterative optimization of model parameters based on iterative step length, wherein an iterative optimization formula is as follows:
let s=s+1, return to step S42.
Optionally, in the step S5, the medical molecular structure generation is performed based on an optimal medical molecular structure generation model, including:
And generating a medical molecular structure by utilizing an optimal medical molecular structure generation model, wherein the medical molecular structure generation flow of the nth medical category is as follows:
s51: the input layer receives the local characteristic information of the medicine molecular sequence of the nth medicine categoryMedical molecular sequence Structure->And a medical molecular site preference matrix->：
Wherein:
representing the structure of the medical molecular sequence of the nth medical category,/->Representing the sequence of a pharmaceutical molecule->Is of the structure of->Representing the sequence of a pharmaceutical molecule->Is>A single thermal coded representation of the segment molecular structure;
medical molecule sequence->Local characteristic information of the corresponding medical molecule sequence;
s52: local characteristic information of coding layer on medical molecular sequenceMedical molecular sequence Structure->And a medical molecular site preference matrix->And performing aggregation update, wherein an aggregation update formula is as follows:
wherein:
representing an activation function;
representing matrix splicing treatment;
w represents an aggregation matrix parameter;
information representing local characteristics of a sequence of a pharmaceutical molecule>Medical molecular sequence Structure->And a medical molecular site preference matrix->Is updated by aggregation of the (a);
s53: updating results for aggregationGlobal information extraction is carried out:
wherein:
representing aggregate update results->Global information extraction results of (a);
Representing extracting a maximum value of each row of elements;
s54: the coding layer carries out multi-channel perception coding to generate medicine molecular structure coding representation：
Wherein:
representing a global pooling operation;
s55: the decoding layer encodes and represents the medicine molecular structureDecoding to obtain medicinal molecular structure ∈>Wherein the decoding formula is:
wherein:
the probability value of each segment of molecular structure is corresponding to the probability matrix form;
representing matrix parameters in the decoding layer, +.>Representing bias parameters in the decoding layer;
s56: output layer from medical molecular structureThe R section molecular structure with the highest probability value is selected to form the medical molecule.
In order to solve the above-described problems, the present invention provides an electronic apparatus including:
a memory storing at least one instruction;
the communication interface is used for realizing the communication of the electronic equipment; a kind of electronic device with high-pressure air-conditioning system
And the processor executes the instructions stored in the memory to realize the targeting-induced medical molecular structure generation method.
In order to solve the above-mentioned problems, the present invention also provides a computer-readable storage medium having stored therein at least one instruction that is executed by a processor in an electronic device to implement the above-mentioned targeting-induced medical molecular structure generation method.
Compared with the prior art, the invention provides a targeted induction medical molecular structure generation method, which has the following advantages:
firstly, the scheme provides a targeting-induced medical molecular structure generation algorithm, and combines the targeting induction to construct a medical molecular site preference matrix, wherein the medical molecular site preference matrix comprises site order information of medical molecular sequences in a medical molecular sequence set, and the medical molecular sequence set is combined with a medical molecular sequence set of an nth medical categoryThe medical molecular site preference matrix of (2) is:
wherein:representing the set of medical molecule sequences->Is a matrix of preference for the medical molecular site of (2)>Representing the sequence of a pharmaceutical molecule->Is a site preference matrix of (a); />Representing the sequence of a pharmaceutical molecule->Middle (f)Preference value of segment molecular structure and K-th drug action target point, wherein the preference value is +.>The preference value of 1 indicates that the segment of molecular structure is beneficial to the drug action with the drug action target, the preference value of 0 indicates that the segment of molecular structure has no obvious relation with the drug action target, and the optimal medicine molecular structure generation model is utilized to generate medicine molecular structure, wherein the medicine molecular structure generation flow of the nth medicine category is as follows: the input layer receives the local characteristic information of the medicine molecular sequence of the nth medicine category >Medical molecular sequence Structure->And a medical molecular site preference matrix->：
Wherein:representing the structure of the medical molecular sequence of the nth medical category,/->Representing the sequence of a pharmaceutical moleculeIs of the structure of->Representing the sequence of a pharmaceutical molecule->Is>A single thermal coded representation of the segment molecular structure; />Medical molecule sequence->Local characteristic information of the corresponding medical molecule sequence; partial characteristic information of coding layer on medical molecule sequence>Medical molecular sequence Structure->And a medical molecular site preference matrix->And performing aggregation update, wherein an aggregation update formula is as follows:
wherein: />Representing an activation function; />Representing matrix splicing treatment; w represents an aggregation matrix parameter; />Information representing local characteristics of a sequence of a pharmaceutical molecule>Medical molecular sequence Structure->And a medical molecular site preference matrix->Is updated by aggregation of the (a); updating the results for aggregation->Global information extraction is carried out:
wherein: />Representing aggregate update results->Global information extraction results of (a); />Representing extracting a maximum value of each row of elements; the coding layer carries out multi-channel perception coding to generate a medicine molecular structure coding representation +.>：
Wherein:representing a global pooling operation; the decoding layer encodes the pharmaceutical molecular structure to express +.>Decoding to obtain medicinal molecular structure ∈ >Wherein the decoding formula is:
wherein: />The probability value of each segment of molecular structure is corresponding to the probability matrix form; />Representing matrix parameters in the decoding layer, +.>Representing bias parameters in the decoding layer; the output layer is from the medical molecular structure->The R section molecular structure with the highest probability value is selected to form the medical molecule. The method comprises the steps of extracting local characteristic information of a medicine molecular sequence by combining attention weights of medicine molecular sequences in the same medicine category, constructing corresponding position point information representing the medicine molecular structure and the medicine action target point based on a targeting induction mode of the medicine action target point, generating a medicine molecular position preference matrix, inputting the extracted medicine molecular sequence local characteristic information, the medicine molecular sequence structure and the medicine molecular position preference matrix serving as models, generating a medicine molecular structure related to the medicine action target point and the medicine molecular sequence local characteristic information, and realizing automatic generation of the medicine molecular structure of targeting induction.
Meanwhile, the scheme provides a model parameter solving method, wherein the model parameter solving method is used for obtaining local characteristic information of medical molecular sequences formed by the medical molecular sequences of the group D and the medical category, forming a training data set data by a medical molecular sequence structure and a medical molecular site preference matrix, and constructing an optimization objective function of a medical molecular structure generation model:
Wherein: />Representing medicineOptimized objective function of molecular structure generation model, +.>Model parameters to be optimally solved for representing a medical molecular structure generation model; />The medical molecular structure set generated by taking the medical molecular sequence local characteristic information, the medical molecular sequence structure and the medical molecular site preference matrix in the training data set data as the input of a medical molecular structure generation model is represented, wherein the model parameters of the medical molecular structure generation model are->；/>Representing the set of medical molecular structures->The single thermal code of the traditional Chinese medicine molecular structure d represents the maximum value of cosine similarity represented by the single thermal code of any medicine molecular sequence structure in the training data set data; />Representing a probability sequence consisting of probability values of each segment of the medical molecular structure d output by the model, +.>Representing a probability sequence formed by the occurrence probability of each section of molecular structure in the medical molecular structure d in the training data set data; and carrying out optimization solving on an optimization objective function of the medical molecular structure generation model to obtain model parameters, wherein the solving flow of the model parameters is as follows: randomly generated initial model parameters->Setting the current iteration number of the model parameter as s and the maximum iteration number as Max, wherein the result of the s-th iteration of the model parameter is S has an initial value of 0; if->Less than the preset iteration threshold, will +.>Model parameters obtained as final iteration and based on model parameters +.>Constructing and obtaining an optimal medical molecular structure generation model; calculating to obtain the iteration step length of the s-th iteration of the model parameter>：
Wherein:representing the step size +.>Iteration parameters of (a); iterative optimization of model parameters is carried out based on iteration step length, an objective function for model parameter training is built by combining the generated probability distribution of the medicine molecular structure and the similarity degree of the probability distribution of the medicine molecular structure of the same medicine category, iterative calculation is carried out on the objective function by combining the gradient descent of the objective function and the linear parameter correction, effective model parameters are quickly obtained, and the model parameters are built and used for medicine moleculesAnd (5) a model generated by the structure.
Drawings
FIG. 1 is a schematic flow chart of a method for generating a targeting-induced molecular structure of a drug according to an embodiment of the invention;
fig. 2 is a schematic structural diagram of an electronic device for implementing a method for generating a medical molecular structure for targeted induction according to an embodiment of the present invention.
In the figure: 1 an electronic device, 10 a processor, 11 a memory, 12 a program, 13 a communication interface.
The achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
The embodiment of the application provides a targeting induced medical molecular structure generation method. The main execution body of the targeting-induced medical molecular structure generation method includes, but is not limited to, at least one of a server, a terminal and the like, which can be configured to execute the method provided by the embodiment of the application. In other words, the targeting-induced medical molecular structure generation method may be performed by software or hardware installed in a terminal device or a server device, and the software may be a blockchain platform. The service end includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, and the like.
Example 1:
s1: and constructing a medical molecular sequence set, and constructing a medical molecular sequence remote dependency information extraction model to extract medical molecular sequence local characteristic information, wherein the model takes the medical molecular sequence as input and takes the medical molecular sequence local characteristic information as output.
In the step S1, a medical molecular sequence remote dependency information extraction model is constructed, and the method comprises the following steps:
constructing a medical molecular sequence remote dependency information extraction model, wherein the medical molecular sequence remote dependency information extraction model takes a medical molecular sequence as input and takes local characteristic information of the medical molecular sequence as output, and the medical molecular sequence remote dependency information extraction model comprises an input layer, a characteristic calculation layer, a local characteristic information fusion layer and an output layer;
the input layer is used for receiving the medicine molecular sequences of the same medicine category, normalizing the medicine molecular sequences, and transmitting the normalized medicine molecular sequences to the feature calculation layer;
the feature computation layer contains 3The convolution kernel of the matrix size is used for carrying out convolution calculation processing on the normalized medical molecular sequence to obtain the medical molecular sequence characteristics;
the local feature information fusion layer comprises a full-connection layer and a self-attention network layer, and is respectively used for extracting local feature information of the medical molecular sequence features and carrying out local feature information fusion combined with a self-attention mechanism to form the medical molecular sequence local feature information;
the output layer is used for outputting the local characteristic information of the medical molecule sequence corresponding to the medical molecule sequence.
In the S1 step, the local characteristic information of the medical molecular sequence is extracted by utilizing a medical molecular sequence remote dependent information extraction model, and the method comprises the following steps:
obtaining the medical molecular sequences of different medical categories to form a medical molecular sequence set:
wherein:
a set of medical molecule sequences representing an nth medical category, N representing the number of medical categories;
representing the set of medical molecule sequences->Middle->Group medical molecule sequence,/->，Representing the set of medical molecule sequences->Total number of Chinese medicinal molecule sequences;
extracting local characteristic information of the medical molecular sequence by utilizing a medical molecular sequence remote dependency information extraction model, wherein the medical molecular sequence setThe extraction flow of the local characteristic information of the medical molecular sequence is as follows:
s11: the input layer receives the medicine molecule sequence setAnd normalizing the medicinal molecular sequences, wherein the normalization is single-heat encoding of medicinal molecular sequences, and the medicinal molecular sequences are collected +.>Chinese medicine molecular sequenceNormalized result of (2) is->；
S12: the characteristic calculation layer carries out convolution calculation processing on the normalized medical molecular sequence, whereinThe convolution calculation processing flow of (1) is as follows:
wherein:
respectively representing weight parameter matrixes of three convolution kernels in the feature calculation layer;
Respectively representing bias parameters of three convolution kernels in a feature calculation layer; />
Representing an activation function;
representation->Corresponding medical molecule sequence->Is characterized by the sequence of the pharmaceutical molecule;
s13: the local feature information fusion layer extracts local feature information of the medical molecular sequence features, wherein the medical molecular sequence featuresThe local feature information extraction result of (a) is:
wherein:
indicating transpose,/->Representing the sequence characteristics of a pharmaceutical molecule>Is a local feature information of (1);
s14: according to the local feature information extraction results of different medical molecular sequence features, carrying out local feature information fusion combining with a self-attention mechanism to form medical molecular sequence local feature information, whereinThe corresponding medical molecular sequence local characteristic information has the following formula:
wherein:
an exponential function that is based on a natural constant;
representing the sequence of a pharmaceutical molecule->Local characteristic information of the corresponding medical molecule sequence;
s15: the output layer outputs the local characteristic information of the medicine molecule sequence corresponding to the medicine molecule sequence.
S2: and constructing a medical molecular site preference matrix by combining the targeting induction, wherein the medical molecular site preference matrix comprises site order information of medical molecular sequences in a medical molecular sequence set.
Constructing a medicine molecular site preference matrix by combining targeted induction in the step S2, which comprises the following steps:
constructing a medical molecular site preference matrix by combining targeted induction, wherein the medical molecular site preference matrix comprises site order information of medical molecular sequences in a medical molecular sequence set, and the medical molecular sequence set combined with the nth medical categoryThe medical molecular site preference matrix of (2) is:
wherein:
representing the set of medical molecule sequences->Is a matrix of preference for the medical molecular site of (2)>Representing the sequence of a pharmaceutical molecule->Is a site preference matrix of (a);
representing the sequence of a pharmaceutical molecule->Middle->Preference value of segment molecular structure and K-th drug action target point, wherein the preference value is +.>A preference value of 1 indicates that the segment of molecular structure is conducive to drug action with a drug action target, and a preference value of 0 indicates that the segment of molecular structure has no obvious relationship with the drug action target.
S3: and constructing a medical molecular structure generation model, wherein the medical molecular structure generation model takes the extracted medical molecular sequence local characteristic information, the medical molecular sequence structure and the medical molecular site preference matrix as input and takes the generated medical molecular structure as output.
In the step S3, a medical molecular structure generation model is constructed, which comprises the following steps:
constructing a medical molecular structure generation model, wherein the medical molecular structure generation model takes extracted medical molecular sequence local characteristic information, medical molecular sequence structures and medical molecular site preference matrixes of the same medical category as inputs, and takes generated medical molecular structures as outputs, and the medical molecular structure generation model comprises an input layer, a coding layer, a decoding layer and an output layer;
the input layer is used for receiving the local characteristic information, the medical molecular sequence structure and the medical molecular site preference matrix of the medical molecular sequence of the same medical category;
the coding layer is used for sequentially carrying out aggregation updating, global information extraction and multi-channel perception coding on the receiving result to generate a medicine molecular structure coding representation;
the decoding layer is used for decoding the coded representation of the medical molecular structure to obtain the medical molecular structure;
the output layer is used for outputting the medical molecular structure. In the embodiment of the invention, the generated medical category corresponding to the medical molecular structure is the same as the medical category corresponding to the input medical molecular sequence local characteristic information and the medical molecular site preference matrix.
S4: and carrying out optimization solution on the constructed medical molecular structure generation model to obtain an optimal medical molecular structure generation model.
And in the step S4, the constructed medical molecular structure generation model is optimized and solved, and the method comprises the following steps:
carrying out optimization solving on the constructed medical molecular structure generating model, wherein model parameters to be optimized solved in the medical molecular structure generating model comprise aggregation matrix parameters in a coding layer, matrix parameters in a decoding layer and bias parameters;
the method comprises the steps of obtaining medical molecular sequences of group D and medical category to form medical molecular sequence local characteristic information, medical molecular sequence structure and medical molecular site preference matrix to form training data set data, and constructing an optimization objective function of a medical molecular structure generation model:
wherein:
an optimized objective function representing a model of the generation of a medical molecular structure, < >>Model parameters to be optimally solved for representing a medical molecular structure generation model; />
The medical molecular structure set generated by taking the medical molecular sequence local characteristic information, the medical molecular sequence structure and the medical molecular site preference matrix in the training data set data as the input of a medical molecular structure generation model is represented, wherein the model parameters of the medical molecular structure generation model are- >；
Representing the set of medical molecular structures->Single thermal coding representation and training number of molecular structure d of Chinese medicineA maximum value of cosine similarity represented by single thermal encoding of any medical molecular sequence structure in the dataset data;
representing a probability sequence consisting of probability values of each segment of the medical molecular structure d output by the model, +.>Representing a probability sequence formed by the occurrence probability of each section of molecular structure in the medical molecular structure d in the training data set data;
and carrying out optimization solving on an optimization objective function of the medical molecular structure generation model to obtain model parameters, wherein the solving flow of the model parameters is as follows:
s41: random generation of initial model parametersSetting the current iteration number of the model parameter as s and the maximum iteration number as Max, and setting the s-th iteration result of the model parameter as +.>S has an initial value of 0;
s42: if it isLess than the preset iteration threshold, will +.>Model parameters obtained as final iteration and based on model parameters +.>Constructing to obtain an optimal medical molecular structure generation model, otherwise turning to step S43;
s43: calculating to obtain the iteration step of the s-th iteration of the model parameter：
Wherein:
representing the step size +.>Iteration parameters of (a);
s44: performing iterative optimization of model parameters based on iterative step length, wherein an iterative optimization formula is as follows:
Let s=s+1, return to step S42.
S5: and generating the medical molecular structure based on the optimal medical molecular structure generation model.
In the step S5, the medical molecular structure generation is performed based on the optimal medical molecular structure generation model, including:
and generating a medical molecular structure by utilizing an optimal medical molecular structure generation model, wherein the medical molecular structure generation flow of the nth medical category is as follows:
s51: the input layer receives the local characteristic information of the medicine molecular sequence of the nth medicine categoryMedical molecular sequence Structure->And a medical molecular site preference matrix->：
Wherein:
representing the structure of the medical molecular sequence of the nth medical category,/->Representing the sequence of a pharmaceutical molecule->Is of the structure of->Representing the sequence of a pharmaceutical molecule->Is>A single thermal coded representation of the segment molecular structure;
medical molecule sequence->Local characteristic information of the corresponding medical molecule sequence;
s52: local characteristic information of coding layer on medical molecular sequenceMedical molecular sequence Structure->And a medical molecular site preference matrix->And performing aggregation update, wherein an aggregation update formula is as follows:
wherein:
representing an activation function;
representing matrix splicing treatment;
w represents an aggregation matrix parameter;
information representing local characteristics of a sequence of a pharmaceutical molecule >Medical molecular sequence Structure->And a medical molecular site preference matrix->Is updated by aggregation of the (a);
s53: updating results for aggregationGlobal information extraction is carried out:
wherein:
representing aggregate update results->Global information extraction results of (a);
representing extracting a maximum value of each row of elements; />
S54: the coding layer carries out multi-channel perception coding to generate medicine molecular structure coding representation：
Wherein:
representing a global pooling operation;
s55: the decoding layer encodes and represents the medicine molecular structureDecoding to obtain medicinal molecular structure ∈>Wherein the decoding formula is:
wherein:
wherein:
is a probability matrixA form corresponding to the probability value of each segment of the molecular structure;
representing matrix parameters in the decoding layer, +.>Representing bias parameters in the decoding layer;
s56: output layer from medical molecular structureThe R section molecular structure with the highest probability value is selected to form the medical molecule.
Example 2:
fig. 2 is a schematic structural diagram of an electronic device for implementing a method for generating a medical molecular structure for targeted induction according to an embodiment of the present invention.
The electronic device 1 may comprise a processor 10, a memory 11, a communication interface 13 and a bus, and may further comprise a computer program, such as program 12, stored in the memory 11 and executable on the processor 10.
The memory 11 includes at least one type of readable storage medium, including flash memory, a mobile hard disk, a multimedia card, a card memory (e.g., SD or DX memory, etc.), a magnetic memory, a magnetic disk, an optical disk, etc. The memory 11 may in some embodiments be an internal storage unit of the electronic device 1, such as a removable hard disk of the electronic device 1. The memory 11 may in other embodiments also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, a Smart Media Card (SMC), a secure digital (SecureDigital, SD) Card, a Flash memory Card (Flash Card) or the like, which are provided on the electronic device 1. Further, the memory 11 may also include both an internal storage unit and an external storage device of the electronic device 1. The memory 11 may be used not only for storing application software installed in the electronic device 1 and various types of data, such as codes of the program 12, but also for temporarily storing data that has been output or is to be output.
The processor 10 may be comprised of integrated circuits in some embodiments, for example, a single packaged integrated circuit, or may be comprised of multiple integrated circuits packaged with the same or different functions, including one or more central processing units (Central Processing unit, CPU), microprocessors, digital processing chips, graphics processors, combinations of various control chips, and the like. The processor 10 is a Control Unit (Control Unit) of the electronic device, connects the respective components of the entire electronic device using various interfaces and lines, executes or executes programs or modules (a program 12 for realizing targeted induction of medical molecular structure generation, etc.) stored in the memory 11, and invokes data stored in the memory 11 to perform various functions of the electronic device 1 and process data.
The communication interface 13 may comprise a wired interface and/or a wireless interface (e.g. WI-FI interface, bluetooth interface, etc.), typically used to establish a communication connection between the electronic device 1 and other electronic devices and to enable connection communication between internal components of the electronic device.
The bus may be a peripheral component interconnect standard (peripheral component interconnect, PCI) bus or an extended industry standard architecture (extended industry standard architecture, EISA) bus, among others. The bus may be classified as an address bus, a data bus, a control bus, etc. The bus is arranged to enable a connection communication between the memory 11 and at least one processor 10 etc.
Fig. 2 shows only an electronic device with components, it being understood by a person skilled in the art that the structure shown in fig. 2 does not constitute a limitation of the electronic device 1, and may comprise fewer or more components than shown, or may combine certain components, or may be arranged in different components.
For example, although not shown, the electronic device 1 may further include a power source (such as a battery) for supplying power to each component, and preferably, the power source may be logically connected to the at least one processor 10 through a power management device, so that functions of charge management, discharge management, power consumption management, and the like are implemented through the power management device. The power supply may also include one or more of any of a direct current or alternating current power supply, recharging device, power failure detection circuit, power converter or inverter, power status indicator, etc. The electronic device 1 may further include various sensors, bluetooth modules, wi-Fi modules, etc., which will not be described herein.
The electronic device 1 may optionally further comprise a user interface, which may be a Display, an input unit, such as a Keyboard (Keyboard), or a standard wired interface, a wireless interface. Alternatively, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch, or the like. The display may also be referred to as a display screen or display unit, as appropriate, for displaying information processed in the electronic device 1 and for displaying a visual user interface.
It should be understood that the embodiments described are for illustrative purposes only and are not limited to this configuration in the scope of the patent application.
The program 12 stored in the memory 11 of the electronic device 1 is a combination of instructions that, when executed in the processor 10, may implement:
constructing a medical molecular sequence set, and constructing a medical molecular sequence remote dependency information extraction model to extract medical molecular sequence local characteristic information;
constructing a medicine molecular site preference matrix by combining targeted induction;
s3: constructing a medical molecular structure generation model;
Carrying out optimization solution on the constructed medical molecular structure generation model to obtain an optimal medical molecular structure generation model;
and generating the medical molecular structure based on the optimal medical molecular structure generation model.
Specifically, the specific implementation method of the above instruction by the processor 10 may refer to descriptions of related steps in the corresponding embodiments of fig. 1 to 2, which are not repeated herein.
It should be noted that, the foregoing reference numerals of the embodiments of the present invention are merely for describing the embodiments, and do not represent the advantages and disadvantages of the embodiments. And the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, apparatus, article, or method. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, apparatus, article or method that comprises the element.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) as described above, comprising instructions for causing a terminal device (which may be a mobile phone, a computer, a server, or a network device, etc.) to perform the method according to the embodiments of the present invention.
The foregoing description is only of the preferred embodiments of the present invention, and is not intended to limit the scope of the invention, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein or in the alternative, which may be employed directly or indirectly in other related arts.
Claims (1)
1. A method of targeted induction of generation of a pharmaceutical molecular structure, the method comprising:
s1: constructing a medical molecular sequence set, and constructing a medical molecular sequence remote dependency information extraction model to extract medical molecular sequence local characteristic information, wherein the model takes a medical molecular sequence as input and takes medical molecular sequence local characteristic information as output;
s2: constructing a medical molecular site preference matrix by combining with targeted induction, wherein the medical molecular site preference matrix comprises site order information of medical molecular sequences in a medical molecular sequence set;
s3: constructing a medical molecular structure generation model, wherein the medical molecular structure generation model takes extracted medical molecular sequence local characteristic information, medical molecular sequence structure and medical molecular site preference matrix as input, and takes generated medical molecular structure as output;
S4: carrying out optimization solution on the constructed medical molecular structure generation model to obtain an optimal medical molecular structure generation model;
s5: generating a medical molecular structure based on the optimal medical molecular structure generation model;
in the step S1, a medical molecular sequence remote dependency information extraction model is constructed, and the method comprises the following steps:
constructing a medical molecular sequence remote dependency information extraction model, wherein the medical molecular sequence remote dependency information extraction model takes a medical molecular sequence as input and takes local characteristic information of the medical molecular sequence as output, and the medical molecular sequence remote dependency information extraction model comprises an input layer, a characteristic calculation layer, a local characteristic information fusion layer and an output layer;
the input layer is used for receiving the medicine molecular sequences of the same medicine category, normalizing the medicine molecular sequences, and transmitting the normalized medicine molecular sequences to the feature calculation layer;
the feature computation layer contains 3The convolution kernel of the matrix size is used for carrying out convolution calculation processing on the normalized medical molecular sequence to obtain the medical molecular sequence characteristics;
the local feature information fusion layer comprises a full-connection layer and a self-attention network layer, and is respectively used for extracting local feature information of the medical molecular sequence features and carrying out local feature information fusion combined with a self-attention mechanism to form the medical molecular sequence local feature information;
The output layer is used for outputting the local characteristic information of the medical molecule sequence corresponding to the medical molecule sequence;
and the step S1 of extracting local characteristic information of the medical molecular sequence comprises the following steps:
obtaining the medical molecular sequences of different medical categories to form a medical molecular sequence set:
wherein:
a set of medical molecule sequences representing an nth medical category, N representing the number of medical categories;
representing the set of medical molecule sequences->Middle->Group medical molecule sequence,/->，/>Representing the set of medical molecule sequences->Total number of Chinese medicinal molecule sequences;
remote information extraction by using medicine molecular sequenceExtracting local characteristic information of medical molecular sequences by taking a model, wherein the medical molecular sequences are gatheredThe extraction flow of the local characteristic information of the medical molecular sequence is as follows:
s11: the input layer receives the medicine molecule sequence setAnd normalizing the medicinal molecular sequences, wherein the normalization is single-heat encoding of medicinal molecular sequences, and the medicinal molecular sequences are collected +.>Chinese medicine molecular sequence->Normalized result of (2) is->；
S12: the characteristic calculation layer carries out convolution calculation processing on the normalized medical molecular sequence, whereinThe convolution calculation processing flow of (1) is as follows:
Wherein:
respectively representing weight parameter matrixes of three convolution kernels in the feature calculation layer;
respectively represent feature calculation layersBias parameters of the three convolution kernels;
representing an activation function;
representation->Corresponding medical molecule sequence->Is characterized by the sequence of the pharmaceutical molecule;
s13: the local feature information fusion layer extracts local feature information of the medical molecular sequence features, wherein the medical molecular sequence featuresThe local feature information extraction result of (a) is:
wherein:
t represents the transpose of the number,representing the sequence characteristics of a pharmaceutical molecule>Is a local feature information of (1);
s14: according to the local feature information extraction results of different medical molecular sequence features, carrying out local feature information fusion combining with a self-attention mechanism to form medical molecular sequence local feature information, whereinCorresponding medical molecule sequence officeThe composition formula of the part characteristic information is as follows:
wherein:
an exponential function that is based on a natural constant;
representing the sequence of a pharmaceutical molecule->Local characteristic information of the corresponding medical molecule sequence;
s15: the output layer outputs the local characteristic information of the medical molecule sequence corresponding to the medical molecule sequence;
constructing a medicine molecular site preference matrix by combining targeted induction in the step S2, which comprises the following steps:
Constructing a medical molecular site preference matrix by combining targeted induction, wherein the medical molecular site preference matrix comprises site order information of medical molecular sequences in a medical molecular sequence set, and the medical molecular sequence set combined with the nth medical categoryThe medical molecular site preference matrix of (2) is:
wherein:
representing the set of medical molecule sequences->Is a matrix of preference for the medical molecular site of (2)>Representing the sequence of a pharmaceutical moleculeIs a site preference matrix of (a);
representing the sequence of a pharmaceutical molecule->Middle->Preference value of segment molecular structure and K-th drug action target point, wherein the preference value is +.>A preference value of 1 indicates that the segment of molecular structure is conducive to drug action with a drug action target, and a preference value of 0 indicates that the segment of molecular structure has no obvious relationship with the drug action target;
in the step S3, a medical molecular structure generation model is constructed, which comprises the following steps:
constructing a medical molecular structure generation model, wherein the medical molecular structure generation model takes extracted medical molecular sequence local characteristic information, medical molecular sequence structures and medical molecular site preference matrixes of the same medical category as inputs, and takes generated medical molecular structures as outputs, and the medical molecular structure generation model comprises an input layer, a coding layer, a decoding layer and an output layer;
The input layer is used for receiving the local characteristic information, the medical molecular sequence structure and the medical molecular site preference matrix of the medical molecular sequence of the same medical category;
the coding layer is used for sequentially carrying out aggregation updating, global information extraction and multi-channel perception coding on the receiving result to generate a medicine molecular structure coding representation;
the decoding layer is used for decoding the coded representation of the medical molecular structure to obtain the medical molecular structure;
the output layer is used for outputting a medical molecular structure;
and in the step S4, the constructed medical molecular structure generation model is optimized and solved, and the method comprises the following steps:
carrying out optimization solving on the constructed medical molecular structure generating model, wherein model parameters to be optimized solved in the medical molecular structure generating model comprise aggregation matrix parameters in a coding layer, matrix parameters in a decoding layer and bias parameters;
the method comprises the steps of obtaining medical molecular sequences of group D and medical category to form medical molecular sequence local characteristic information, medical molecular sequence structure and medical molecular site preference matrix to form training data set data, and constructing an optimization objective function of a medical molecular structure generation model:
wherein:
an optimized objective function representing a model of the generation of a medical molecular structure, < > >Model parameters to be optimally solved for representing a medical molecular structure generation model;
representing a medical molecular structure set generated by taking the medical molecular sequence local characteristic information, the medical molecular sequence structure and the medical molecular site preference matrix in the training data set data as the input of a medical molecular structure generation model, wherein the medical molecular structure generation model is a medical molecular structure set generated by the modelModel parameters are->；
Representing the set of medical molecular structures->The single thermal code of the traditional Chinese medicine molecular structure d represents the maximum value of cosine similarity represented by the single thermal code of any medicine molecular sequence structure in the training data set data;
represents a probability sequence formed by probability values of each section of molecular structure in the medical molecular structure d output by the model,representing a probability sequence formed by the occurrence probability of each section of molecular structure in the medical molecular structure d in the training data set data;
and carrying out optimization solving on an optimization objective function of the medical molecular structure generation model to obtain model parameters, wherein the solving flow of the model parameters is as follows:
s41: random generation of initial model parametersSetting the current iteration number of the model parameter as s and the maximum iteration number as Max, and setting the s-th iteration result of the model parameter as +. >S has an initial value of 0;
s42: if it isLess than the preset iteration threshold, will +.>Model parameters obtained as final iteration and based on model parameters +.>Constructing to obtain an optimal medical molecular structure generation model, otherwise turning to step S43;
s43: calculating to obtain the iteration step of the s-th iteration of the model parameter：
Wherein:
representing the step size +.>Iteration parameters of (a);
s44: performing iterative optimization of model parameters based on iterative step length, wherein an iterative optimization formula is as follows:
let s=s+1, return to step S42;
in the step S5, the medical molecular structure generation is performed based on the optimal medical molecular structure generation model, including:
and generating a medical molecular structure by utilizing an optimal medical molecular structure generation model, wherein the medical molecular structure generation flow of the nth medical category is as follows:
s51: the input layer receives the local characteristic information of the medicine molecular sequence of the nth medicine categoryMolecular sequence structure of medicineAnd a medical molecular site preference matrix->：
Wherein:
representing the structure of the medical molecular sequence of the nth medical category,/->Representing the sequence of a pharmaceutical molecule->Is characterized in that the structure of the (c) is that,representing the sequence of a pharmaceutical molecule->Is>A single thermal coded representation of the segment molecular structure;
medical molecule sequence- >Local characteristic information of the corresponding medical molecule sequence;
s52: local characteristic information of coding layer on medical molecular sequenceMedical molecular sequence Structure->And a medical molecular site preference matrix->And performing aggregation update, wherein an aggregation update formula is as follows:
wherein:
representing an activation function;
representing matrix splicing treatment;
w represents an aggregation matrix parameter;
information representing local characteristics of a sequence of a pharmaceutical molecule>Medical molecular sequence Structure->And a medical molecular site preference matrix->Is updated by aggregation of the (a);
s53: updating results for aggregationGlobal information extraction is carried out:
wherein:
representing aggregate update results->Global information extraction results of (a);
representing extracting a maximum value of each row of elements;
s54: the coding layer carries out multichannel perception coding to generate a medicine molecular structure coding representation:
wherein:
representing a global pooling operation;
s55: the decoding layer encodes and represents the medicine molecular structureDecoding to obtain medicinal molecular structure ∈>Wherein the decoding formula is:
wherein:
the probability value of each segment of molecular structure is corresponding to the probability matrix form;
representing matrix parameters in the decoding layer, +.>Representing bias parameters in the decoding layer;
s56: output layer from medical molecular structure The R section molecular structure with the highest probability value is selected to form the medical molecule.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311359357.4A CN117116384B (en) | 2023-10-20 | 2023-10-20 | Targeted induction type medical molecular structure generation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311359357.4A CN117116384B (en) | 2023-10-20 | 2023-10-20 | Targeted induction type medical molecular structure generation method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117116384A CN117116384A (en) | 2023-11-24 |
| CN117116384B true CN117116384B (en) | 2024-01-09 |
Family
ID=88813125
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311359357.4A Active CN117116384B (en) | 2023-10-20 | 2023-10-20 | Targeted induction type medical molecular structure generation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117116384B (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008091225A1 (en) * | 2007-01-22 | 2008-07-31 | Agency For Science, Technology And Research | Comparative detection of structure patterns in interaction sites of molecules |
| CN110572869A (en) * | 2019-09-11 | 2019-12-13 | 武汉拓宝科技股份有限公司 | equipment awakening method and device based on BLE and Hall switch |
| US11127488B1 (en) * | 2020-09-25 | 2021-09-21 | Accenture Global Solutions Limited | Machine learning systems for automated pharmaceutical molecule screening and scoring |
| CN114220480A (en) * | 2022-02-17 | 2022-03-22 | 武汉宏韧生物医药股份有限公司 | Method and system for analyzing medicine components |
| CN114417986A (en) * | 2022-01-11 | 2022-04-29 | 平安科技（深圳）有限公司 | Artificial intelligence-based medicine characteristic information determination method and device |
| CN114896291A (en) * | 2022-04-28 | 2022-08-12 | 百度在线网络技术（北京）有限公司 | Training method and sequencing method of multi-agent model |
| EP4050612A1 (en) * | 2019-10-21 | 2022-08-31 | Standigm Inc. | Method and device for designing compound |
| CN115101146A (en) * | 2022-07-29 | 2022-09-23 | 郑州大学 | Medicine target prediction method and system based on Weisfeiler-Lehman and deep neural network |
| CN116779061A (en) * | 2023-06-19 | 2023-09-19 | 平安科技（深圳）有限公司 | Interactive drug molecule design method, device, electronic equipment and medium |
| CN116825236A (en) * | 2023-06-30 | 2023-09-29 | 平安科技（深圳）有限公司 | Method, device, equipment and medium for generating drug molecules of protein targets |
| CN116882496A (en) * | 2023-09-07 | 2023-10-13 | 中南大学湘雅医院 | Medical knowledge base construction method for multistage logic reasoning |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020016579A2 (en) * | 2018-07-17 | 2020-01-23 | Gtn Ltd | Machine learning based methods of analysing drug-like molecules |
| US11544535B2 (en) * | 2019-03-08 | 2023-01-03 | Adobe Inc. | Graph convolutional networks with motif-based attention |
| US20200392178A1 (en) * | 2019-05-15 | 2020-12-17 | International Business Machines Corporation | Protein-targeted drug compound identification |
-
2023
- 2023-10-20 CN CN202311359357.4A patent/CN117116384B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008091225A1 (en) * | 2007-01-22 | 2008-07-31 | Agency For Science, Technology And Research | Comparative detection of structure patterns in interaction sites of molecules |
| CN110572869A (en) * | 2019-09-11 | 2019-12-13 | 武汉拓宝科技股份有限公司 | equipment awakening method and device based on BLE and Hall switch |
| EP4050612A1 (en) * | 2019-10-21 | 2022-08-31 | Standigm Inc. | Method and device for designing compound |
| US11127488B1 (en) * | 2020-09-25 | 2021-09-21 | Accenture Global Solutions Limited | Machine learning systems for automated pharmaceutical molecule screening and scoring |
| CN114417986A (en) * | 2022-01-11 | 2022-04-29 | 平安科技（深圳）有限公司 | Artificial intelligence-based medicine characteristic information determination method and device |
| CN114220480A (en) * | 2022-02-17 | 2022-03-22 | 武汉宏韧生物医药股份有限公司 | Method and system for analyzing medicine components |
| CN114896291A (en) * | 2022-04-28 | 2022-08-12 | 百度在线网络技术（北京）有限公司 | Training method and sequencing method of multi-agent model |
| CN115101146A (en) * | 2022-07-29 | 2022-09-23 | 郑州大学 | Medicine target prediction method and system based on Weisfeiler-Lehman and deep neural network |
| CN116779061A (en) * | 2023-06-19 | 2023-09-19 | 平安科技（深圳）有限公司 | Interactive drug molecule design method, device, electronic equipment and medium |
| CN116825236A (en) * | 2023-06-30 | 2023-09-29 | 平安科技（深圳）有限公司 | Method, device, equipment and medium for generating drug molecules of protein targets |
| CN116882496A (en) * | 2023-09-07 | 2023-10-13 | 中南大学湘雅医院 | Medical knowledge base construction method for multistage logic reasoning |
Non-Patent Citations (3)
| Title |
|---|
| 人工智能在药物发现中的应用与挑战;梁礼;邓成龙;张艳敏;滑艺;刘海春;陆涛;陈亚东;;药学进展(第01期);正文 * |
| 基于三元残基组合对的蛋白质相互作用研究;肖薇;何增辉;李诗良;李洪林;;药学学报(第10期);正文 * |
| 计算化学方法在基于受体结构的药物分子设计中的基础理论及应用;曹冉;李伟;孙汉资;周宇;黄牛;;药学学报(第07期);正文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117116384A (en) | 2023-11-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113159147B (en) | Image recognition method and device based on neural network and electronic equipment | |
| CN111932534B (en) | Medical image picture analysis method and device, electronic equipment and readable storage medium | |
| CN113378970B (en) | Sentence similarity detection method and device, electronic equipment and storage medium | |
| CN113868529A (en) | Knowledge recommendation method and device, electronic equipment and readable storage medium | |
| CN116882496B (en) | Medical knowledge base construction method for multistage logic reasoning | |
| CN112651782B (en) | Behavior prediction method, device, equipment and medium based on dot product attention scaling | |
| CN117116384B (en) | Targeted induction type medical molecular structure generation method | |
| CN116383766A (en) | Auxiliary diagnosis method, device, equipment and storage medium based on multi-mode data | |
| CN116702776A (en) | Multi-task semantic division method, device, equipment and medium based on cross-Chinese and western medicine | |
| CN116483976A (en) | Registration department recommendation method, device, equipment and storage medium | |
| CN114596958B (en) | Pathological data classification method, device, equipment and medium based on cascade classification | |
| CN116578690B (en) | Insurance customer service method and system based on artificial intelligence | |
| CN114864032B (en) | Clinical data acquisition method and device based on HIS system | |
| CN112949759B (en) | Text clustering method, text clustering device, electronic equipment and computer readable storage medium | |
| CN115547457B (en) | Recipe intelligent recommendation method, device, equipment and medium based on physical examination data | |
| CN114742423B (en) | Random grouping method, device, equipment and medium | |
| CN116486898A (en) | Method and device for analyzing relationship between medicine and cancer based on graph neural network | |
| CN117290806A (en) | Multi-path parallel medical knowledge integration method | |
| CN116644336A (en) | Method, device, equipment and storage medium for recognizing Chinese medicine terms | |
| CN116881558A (en) | Medicine-based questionnaire pushing method, device, equipment and storage medium | |
| CN116431810A (en) | Pruning paradigm disorder segment extraction method, device, equipment and storage medium | |
| CN116757197A (en) | Text theme segmentation method, device, equipment and storage medium | |
| CN116504317A (en) | Disease and gene relation prediction method, device, electronic equipment and medium | |
| CN116701635A (en) | Training video text classification method, training video text classification device, training video text classification equipment and storage medium | |
| CN116541723A (en) | Image text matching method, device, equipment and storage medium |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |