CN109872780A - A kind of determination method and device of chemical synthesis route - Google Patents
A kind of determination method and device of chemical synthesis route Download PDFInfo
- Publication number
- CN109872780A CN109872780A CN201910194016.3A CN201910194016A CN109872780A CN 109872780 A CN109872780 A CN 109872780A CN 201910194016 A CN201910194016 A CN 201910194016A CN 109872780 A CN109872780 A CN 109872780A
- Authority
- CN
- China
- Prior art keywords
- target compound
- compound
- chemical synthesis
- response rule
- disaggregated model
- 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
Links
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
This application discloses a kind of determination methods of chemical synthesis route, comprising: receives chemical synthesis request, the chemical synthesis request is used for the chemical synthesis route of request target compound；The target compound is input to disaggregated model, obtains response rule corresponding with the target compound；The disaggregated model is with compound for input, and take response rule as the machine learning model of output；The target compound is carried out according to the response rule to decompose raw material needed for retrosynthetic analysis obtains synthesising target compound；Chemical synthesis route corresponding with the target compound is determined according to the raw material.This method realizes chemical synthesis route Automated Design, and disaggregated model accuracy with higher and stronger generalization ability by disaggregated model, can be applied to new reaction, without training again, so improves chemical synthesis route design efficiency.Disclosed herein as well is the training method of disaggregated model, device, equipment and media.
Description
Technical field
This application involves the determination method and devices of chemical field more particularly to a kind of chemical synthesis route.
Background technique
With the rapid development of organic chemistry, the research for synthesizing chemistry is constantly broken through, and the design of synthetic route is changed in chemistry
Work, pharmacy, the status in the multiple fields such as material become most important.One good synthetic route, can not only allow chemical work
Make personnel can quick synthesising target compound, and bigger profit can be brought for industrial sector.
Currently, most of chemical synthesis route design is to rely on the knowledge and experience of chemist itself.And with
New reaction, new reagent continue to bring out, and that many chemists often can not consider many factors in design route is thorough,
Such as yield height, how much is synthesis step, safety and the feature of environmental protection etc..
Related software on the market can only generally provide the design of single step reaction when designing synthetic route, although can be with
The working efficiency of chemists is promoted, but still to carry out screening step by step according to artificial experience.Up to now, not yet
A full automatic complete synthetic route design software moves towards market.
Certainly, industry has had the method for Part Full Automatic complete synthetic route design to be suggested, including is based on similitude
The chemical synthesis route design method of retrieval, the chemical synthesis route design method based on chemical reaction rule either are based on covering
The chemical synthesis route method of special Carlow tree search.Wherein, the chemical synthesis route design method based on similarity retrieval can only
Compound with similar structure is designed, covers not comprehensively, there is biggish limitation；Based on chemical reaction rule
Chemical synthesis route design method is inconsiderate complete, thus accuracy is lower；Chemical synthesis route based on Monte Carlo tree is set
Meter method calculating speed is extremely slow, and time-consuming.
Therefore it provides the chemical synthesis route of a kind of high efficiency, high accuracy determines that method becomes urgent problem to be solved.
Summary of the invention
In view of this, this application provides a kind of determination method of chemical synthesis route, this method is by machine learning application
It is designed in retrosynthetic analysis and chemical synthesis route, the train classification models by way of machine learning, for anti-to chemistry
It answers rule to be predicted, there is stronger accuracy and preferable Generalization Capability, and can be realized full automatic chemical synthesis
Highway route design.Accordingly, present invention also provides a kind of training method of disaggregated model, the determining device of chemical synthesis route,
Training device, equipment, storage medium and the computer program product of disaggregated model.
The application first aspect provides a kind of determination method of chemical synthesis route, which comprises
Chemical synthesis request is received, the chemical synthesis request is used for the chemical synthesis route of request target compound；
The target compound is input to disaggregated model, obtains response rule corresponding with the target compound；Institute
Stating disaggregated model is with compound for input, and take response rule as the machine learning model of output；
Decomposition retrosynthetic analysis is carried out to the target compound according to the response rule to obtain synthesizing the targeted
Raw material needed for closing object；
Chemical synthesis route corresponding with the target compound is determined according to the raw material.
Optionally, it is described according to the response rule to the target compound carry out decompose retrosynthetic analysis synthesized
Raw material needed for the target compound includes:
Decomposition retrosynthetic analysis is carried out to the target compound according to the response rule and obtains decomposition product；
Whether search the decomposition product from raw database with the determination decomposition product is raw material；
Decomposition product is not raw material if it exists, then using the decomposition product as target compound, executes the target compound
It is input to disaggregated model, obtains response rule corresponding with the target compound, and according to the response rule to the mesh
The step of mark compound carries out decomposition retrosynthetic analysis, until the decomposition product is raw material.
Optionally, the method also includes:
It receives and the amendment that the chemical synthesis route is modified is requested；
It is requested in response to the amendment, corrects the chemical synthesis route.
Optionally, the method also includes:
The target compound is converted into corresponding vector using term vector model；
It is then described the target compound is input to disaggregated model to include:
The corresponding vector of the target compound is input to disaggregated model.
Optionally, the machine learning model includes gradient decline decision-tree model, Random Forest model, support vector machines
Any one or more in model and neural network model.
The application second aspect provides a kind of training method of disaggregated model, which comprises
Training sample is obtained, the training sample includes compound and reaction information relevant to the compound, institute
Reaction information is stated including at least response rule；
According to the training sample, using machine learning algorithm train classification models, until meeting training termination condition；
Wherein, the disaggregated model is input with compound, is output with response rule.
Optionally, response rule obtains in the following way in the training sample:
Acquire chemical equation；
Atomic series mark is carried out to the chemical equation, and based in the chemical equation abstraction reaction after mark
The heart；
The reaction center is filtered and clustering processing, response rule is determined according to processing result.
Optionally, the compound in the training sample exists in the form of vectors, and the corresponding vector of the compound passes through
As under type obtains:
The chemical formula of compound each in the chemical equation of acquisition is converted into the minor structure that radius is 1 using root algorithm is rubbed
Sequence；
All minor structure sequences are converted into the corresponding vector of the minor structure sequence using term vector model；
For each compound, the corresponding vector of minor structure sequence of the compound is summed up to obtain the chemical combination
The corresponding vector of object.
Optionally, the machine learning model includes gradient decline decision Tree algorithms, random forests algorithm, support vector machines
Any one or more in algorithm and neural network algorithm.
The application third aspect provides a kind of determining device of chemical synthesis route, and described device includes:
Receiving module, for receiving chemical synthesis request, the chemical synthesis request is used for the change of request target compound
Learn synthetic route；
Module is obtained, for the target compound to be input to disaggregated model, is obtained corresponding with the target compound
Response rule；The disaggregated model is with compound for input, and take response rule as the machine learning model of output；
Decomposing module is closed for carrying out decomposition retrosynthetic analysis to the target compound according to the response rule
At raw material needed for the target compound；
Determining module, for determining chemical synthesis route corresponding with the target compound according to the raw material.
Optionally, the decomposing module is specifically used for；
Decomposition retrosynthetic analysis is carried out to the target compound according to the response rule and obtains decomposition product；
Whether search the decomposition product from raw database with the determination decomposition product is raw material；
Decomposition product is not raw material if it exists, then using the decomposition product as target compound, executes the target compound
It is input to disaggregated model, obtains response rule corresponding with the target compound, and according to the response rule to the mesh
The step of mark compound carries out decomposition retrosynthetic analysis, until the decomposition product is raw material.
Optionally, the receiving module is also used to:
It receives and the amendment that the chemical synthesis route is modified is requested；
Described device further include:
Correction module corrects the chemical synthesis route for requesting in response to the amendment.
Optionally, described device further include:
Conversion module, for the target compound to be converted to corresponding vector using term vector model；
Then the acquisition module is specifically used for when target compound is input to disaggregated model:
The corresponding vector of the target compound is input to disaggregated model.
Optionally, the machine learning model includes gradient decline decision-tree model, Random Forest model, support vector machines
Any one or more in model and neural network model.
The application fourth aspect provides a kind of training device of disaggregated model, and described device includes:
Module is obtained, for obtaining training sample, the training sample includes compound and related to the compound
Reaction information, the reaction information include at least response rule；
Training module, for using machine learning algorithm train classification models, being instructed until meeting according to the training sample
Practice termination condition；
Wherein, the disaggregated model is input with compound, is output with response rule.
Optionally, described device further include:
Acquisition module, for acquiring chemical equation；
Abstraction module, for carrying out atomic series mark to the chemical equation, and based on the chemical reaction after mark
Formula abstraction reaction center；
Processing module determines reaction rule according to processing result for being filtered to the reaction center and clustering processing
Then.
Optionally, the compound in the training sample exists in the form of vectors, described device further include:
First conversion module, for being converted the chemical formula of compound each in the chemical equation of acquisition using root algorithm is rubbed
The minor structure sequence for being 1 for radius；
Second conversion module, for all minor structure sequences to be converted to the minor structure sequence pair using term vector model
The vector answered；
Module is summed it up, for being directed to each compound, the corresponding vector of minor structure sequence of the compound is added
Vector corresponding with the compound is obtained.
Optionally, the machine learning model includes gradient decline decision Tree algorithms, random forests algorithm, support vector machines
Any one or more in algorithm and neural network algorithm.
The 5th aspect of the application provides a kind of equipment, and the equipment includes processor and memory:
Said program code is transferred to the processor for storing program code by the memory；
The processor is closed for the chemistry according to instruction execution the application first aspect in said program code
The training method of disaggregated model described in determination method or the application second aspect at route.
The 5th aspect of the application provides a kind of computer readable storage medium, and the computer readable storage medium is used for
Store program code, said program code be used for execute chemical synthesis route described in the application first aspect determination method or
The training method of the described in any item disaggregated models of person's the application second aspect.
As can be seen from the above technical solutions, the embodiment of the present application has the advantage that
It is for target compound to be synthesized, its is defeated this application provides a kind of determination method of chemical synthesis route
Enter to the disaggregated model by machine learning algorithm training, it is output with response rule which, which is input with compound,
In this way, response rule corresponding with target compound can be obtained, which can characterize to form the target compound
Chemical reaction carries out decomposition retrosynthetic analysis to target compound based on the response rule and obtains original needed for synthesising target compound
Material is then based on the raw material and determines chemical synthesis route corresponding with target compound.It is designed with traditional chemical synthesis route
It compares, this method train classification models by way of machine learning, to be used for the corresponding response rule of predictive compound, passes through
Decomposition retrosynthetic analysis is carried out to classifying rules, to realize chemical synthesis route Automated Design, and the disaggregated model has
Higher accuracy and stronger generalization ability, can be applied to new reaction, without training again, so improve chemical conjunction
At highway route design efficiency.
Detailed description of the invention
In order to illustrate the technical solutions in the embodiments of the present application or in the prior art more clearly, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of application without any creative labor, may be used also for those of ordinary skill in the art
To obtain other drawings based on these drawings.
Fig. 1 is the scene framework figure of the determination method of chemical synthesis route in the embodiment of the present application；
Fig. 2 is the flow chart of the determination method of chemical synthesis route in the embodiment of the present application；
Fig. 3 is the exemplary diagram that chemical synthesis route is determined in the embodiment of the present application；
Fig. 4 is the scene framework figure of the training method of disaggregated model in the embodiment of the present application；
Fig. 5 is the flow chart of the training method of disaggregated model in the embodiment of the present application；
Fig. 6 is the schematic diagram that the minor structure sequence of compound is determined in the embodiment of the present application；
Fig. 7 A, Fig. 7 B, Fig. 7 C are by the determination method of the chemical synthesis route of the embodiment of the present application and by tradition side
The schematic diagram of method progress chemical synthesis route design；
Fig. 8 is a structural schematic diagram of the determining device of chemical synthesis route in the embodiment of the present application；
Fig. 9 is a structural schematic diagram of the training device of disaggregated model in the embodiment of the present application；
Figure 10 is a structural schematic diagram of server in the embodiment of the present application；
Figure 11 is a structural schematic diagram of terminal in the embodiment of the present application.
Specific embodiment
In order to make those skilled in the art more fully understand application scheme, below in conjunction in the embodiment of the present application
Attached drawing, the technical scheme in the embodiment of the application is clearly and completely described, it is clear that described embodiment is only this
Apply for a part of the embodiment, instead of all the embodiments.Based on the embodiment in the application, those of ordinary skill in the art exist
Every other embodiment obtained under the premise of creative work is not made, shall fall in the protection scope of this application.
The description and claims of this application and term " first ", " second ", " third ", " in above-mentioned attached drawing
The (if present)s such as four " are to be used to distinguish similar objects, without being used to describe a particular order or precedence order.It should manage
The data that solution uses in this way are interchangeable under appropriate circumstances, so that embodiments herein described herein for example can be to remove
Sequence other than those of illustrating or describe herein is implemented.In addition, term " includes " and " having " and theirs is any
Deformation, it is intended that cover it is non-exclusive include, for example, containing the process, method of a series of steps or units, system, production
Product or equipment those of are not necessarily limited to be clearly listed step or unit, but may include be not clearly listed or for this
A little process, methods, the other step or units of product or equipment inherently.
It can only be to similar structure for the chemical synthesis route design method in the prior art based on similarity retrieval
Compound be designed, based on chemical reaction rule chemical synthesis route design method it is inconsiderate entirely cause accuracy compared with
It is low；The technical issues of chemical synthesis route design method calculating speed based on Monte Carlo tree is extremely slow, time-consuming, the application mentions
A kind of chemical synthesis route based on machine learning has been supplied to determine method.
This method trains obtained disaggregated model to predict the corresponding response rule of compound using machine learning, base
Decomposition retrosynthetic analysis is carried out in the response rule, realizes full automatic complete synthetic route design.Also, this method is used
Disaggregated model be to be obtained by machine learning training, accuracy with higher and generalization ability, and it can be applied
In new reaction, and it is not limited to the compound of similar structure, without training again, improves chemical synthesis route design effect
Rate.
It is appreciated that the determination method of chemical synthesis route provided by the present application can be applied to processing equipment, the processing
Equipment can be any equipment with data-handling capacity, including terminal or server.In practical application, processing equipment can
Be it is independent, be also possible to multiple equipment and cooperate the cluster to be formed.The determination method of chemical synthesis route can be to apply journey
The form of sequence is stored in processing equipment, and processing equipment realizes chemical synthesis route design by executing the application program.
In order to enable the scheme of the application it is clearer, it can be readily appreciated that below in conjunction with concrete scene to chemical synthesis road
The determination method of line is introduced.Scene framework figure shown in Figure 1 includes terminal 10 and server 20 in the scene, uses
The operation for determining chemical synthesis route is triggered at family by terminal 10, and terminal 10 generates chemical synthesis request in response to user's operation,
And chemical synthesis request is sent to server 20, chemical synthesis request is used for the chemical synthesis route of request target compound,
Server 20 in response to the chemical synthesis request, the target compound is input to disaggregated model, the disaggregated model be with
Compound is input, take response rule as the machine learning model of output, server 20 can obtain corresponding with target compound
Response rule, then according to the response rule to target compound carry out decompose retrosynthetic analysis obtain synthesizing the targeted
Raw material needed for closing object, determines the corresponding chemical synthesis route of target compound according to the raw material, so realizes that complete chemistry closes
At route Automated Design.
Next, by being carried out from determination method of the angle of server to chemical synthesis route provided by the embodiments of the present application
It is discussed in detail.
The determination method of chemical synthesis route shown in Figure 2, this method comprises:
S201: chemical synthesis request is received.
The chemical synthesis request is used for the chemical synthesis route of request target compound.It is carried in chemical synthesis request
There is the mark of target compound, such as the chemical formula of target compound, can so indicate that server is generated for target compound
Chemical synthesis route.
In practical application, user can trigger the operation for determining the chemical synthesis route of target compound by terminal,
Terminal in response to aforesaid operations, request, and sends the chemical synthesis to server by the chemical synthesis generated for target compound
Request.
S202: being input to disaggregated model for the target compound, obtains reaction rule corresponding with the target compound
Then.
The disaggregated model is with compound for input, and take response rule as the machine learning model of output.Server
The mark that target compound is chemically extracted in synthesis request, is input to disaggregated model for target compound, thus acquisition and mesh
Mark the corresponding response rule of compound.
In view of the data-handling capacity of disaggregated model, in practical application, the compound for being input to disaggregated model is general
Exist in the form of vectors.Based on this, target compound is input to disaggregated model and is specifically as follows by server, and targeted is combined into
Corresponding vector is converted to, the corresponding vector of the target compound is then input to disaggregated model.
As a kind of possible implementation, server can realize compound vector by term vector model word2vec
Change.Specifically, the chemical formula of compound is converted to the minor structure sequence that radius is 1 by the server by utilizing root algorithm that rubs, then sharp
All minor structure sequences are converted into the corresponding vector of the minor structure sequence with word2vec, then to the minor structure sequence pair
The vector answered sums up to obtain the corresponding vector of compound.
Compound structure is converted into n-dimensional vector by word2vec model, and n is positive integer.It will be exported from the root algorithm that rubs
The minor structure of compound be considered as " word ", be considered as " sentence " for compound, be a kind of unsupervised method, be initially not mark
Training in the data of note, to obtain the feature vector of minor structure, these feature vectors can be generalized composite vector.Wherein,
When carrying out vectorization to compound, word2vec model can use Skip-gram mode, window size window size
Parameter can be set to 5, and size insertion dimensional embeddings can be set to 300, in practical application, above-mentioned
Parameter can also be selected or be adjusted according to actual needs.It so, it is possible training and obtain the vector of all minor structures, Mei Gehua
The vector for closing object directly can be summed it up to obtain by its all minor structure vector.
In the present embodiment, server is response rule prediction to be realized based on disaggregated model, and then realize chemical synthesis road
Line design, therefore, the accuracy of disaggregated model and generalization ability are particularly important.Specifically, server, which uses, passes through engineering
Algorithm train classification models are practised, the machine learning algorithm includes gradient decline decision Tree algorithms (Gradient Boosting
Decision Tree, GBDT), random forests algorithm, in algorithm of support vector machine and neural network algorithm any one or it is more
Kind namely the disaggregated model can be in GBDT model, Random Forest model, supporting vector machine model and neural network model
Any one or more.The training process of disaggregated model will be introduced in greater detail below, and it will not be described in detail here.
S203: decomposition retrosynthetic analysis is carried out to the target compound according to the response rule and obtains synthesizing the mesh
Raw material needed for marking compound.
Specifically, the chemical reaction of response rule characterization, server can use the response rule and carry out to target compound
It decomposes retrosynthetic analysis and obtains decomposition product, which is the reactant for forming target compound.It should be noted that service
Device carries out decomposing at least one available decomposition product of retrosynthetic analysis, for each decomposition product, server can by
The mode that the decomposition product is searched in raw database determines whether the decomposition product is raw material.
Wherein, raw material refers to the material for further processing, and in the present embodiment, raw material is the material that can be directly obtained
Material, such as can be by being obtained from nature, or the material that can be obtained by way of purchase, be also possible to user from
The material of definition.Server safeguards there is raw database in advance, is stored with resourcing manifest in the raw database, in this way, service
Device can determine whether decomposition product is raw material by the resourcing manifest of inquiry raw database storage.
In some possible implementations, synthesis targeted is can be obtained by the primary retrosynthetic analysis that decomposes in server
Raw material needed for closing object namely server can be looked by the primary retrosynthetic analysis gained decomposition product that decomposes in raw database
It finds, decomposition product is raw material.In other possible implementations, server is obtained by repeatedly decomposing retrosynthetic analysis
To raw material needed for synthesising target compound.In other words, server decompose obtained by decomposition retrosynthetic analysis to target compound
In the presence of the decomposition product for not being raw material in object, in this way, server can be executed using the decomposition product as goal decomposition object by the mesh
Mark compound is input to disaggregated model, obtains response rule corresponding with the target compound, and according to the response rule
The step of decomposition retrosynthetic analysis is carried out to the target compound, until the decomposition product is raw material, namely service
Device is using the decomposition product as goal decomposition object, return step S202.
In specific implementation, server can use Open-Source Tools and realize to the decomposition retrosynthetic analysis of target compound,
Decomposition product corresponding with target compound is obtained, and determines whether the decomposition product is former by searching for the mode of raw database
Material.For example, server can use the RunReactants method in RDkit lib, it can obtain the target compound pair
The decomposition product answered.
S204: chemical synthesis route corresponding with the target compound is determined according to the raw material.
After determining raw material needed for synthesising target compound, server is according to the raw material and decomposes the anti-of retrosynthetic analysis
Chemical synthesis route corresponding with target compound is determined to path.When the raw material is obtained by once decomposing retrosynthetic analysis
When, then server can be designed based on the raw material include a chemical reaction chemical synthesis path with synthesising target compound,
When the raw material is obtained by repeatedly decomposing retrosynthetic analysis, then server can be based on the inverse of the sequencing for determining raw material
Sequence, design include the chemical synthesis path of multiple chemical reactions with synthesising target compound.
In order to make it easy to understand, illustrating below with reference to specific example.
In one example (referring to the left side Fig. 3), the corresponding raw material of target compound C is A and B, wherein A and B is based on one
It is secondary decomposing retrosynthetic analysis and determination, in this way, it is A and B in the first specified requirements that server, which can design chemical synthesis path,
I.e. 1. lower reaction generates C to condition, and wherein first specified requirements includes the ratio of A and B, temperature, pH value, catalyst etc..
In another example (referring to the right side Fig. 3), the corresponding raw material of target compound G includes A, B, D and F, wherein A, B,
D and F is the determination by multiple retrosynthetic analysis, determines raw material F and intermediate E according to target compound G first, secondly
Determine raw material D and intermediate C according to intermediate E, raw material A and B then determined according to intermediate C, in this way, server according to
It determines the backward of raw material sequencing, determines that chemistry synthesis path is, A and B are in the 2. lower reaction life of the second specified requirements, that is, condition
At intermediate C, intermediate C and raw material D, in third specified requirements, that is, condition, 3. lower reaction generates intermediate E, intermediate E and raw material
F generates target compound G four specified requirements, that is, condition is 4. lower, wherein the second specified requirements, third specified requirements and the
Four specified requirements include but is not limited to ratio, reaction temperature, pH value of reactant etc..
It is appreciated that the method for determining chemical synthesis route provided by the embodiments of the present application is in addition to that can automatically generate chemistry
Outside synthetic route, manual intervention can also be received, generate corresponding chemical synthesis route according to actual needs.Specifically, it services
Device can also receive the amendment request being modified to the chemical synthesis route, request then in response to the amendment, amendment
The chemical synthesis route.
From the foregoing, it will be observed that the embodiment of the present application provides a kind of method of determining chemical synthesis route, for mesh to be synthesized
Compound is marked, the disaggregated model by machine learning algorithm training is input to, which is input with compound, with
Response rule is output, in this way, response rule corresponding with target compound can be obtained, which can characterize to be formed
The chemical reaction of the target compound carries out decomposition retrosynthetic analysis to target compound based on the response rule and obtains synthesizing this
Raw material needed for target compound is then based on the raw material and determines chemical synthesis route corresponding with target compound.With it is traditional
Chemical synthesis route design is compared, this method train classification models by way of machine learning, to be used for predictive compound pair
The response rule answered, by carrying out decomposition retrosynthetic analysis to classifying rules, thus realize chemical synthesis route Automated Design, and
And disaggregated model accuracy with higher and stronger generalization ability, it can be applied to new reaction, without training again,
So improve chemical synthesis route design efficiency.
The determination method of chemical synthesis route provided by the embodiments of the present application is realized based on disaggregated model, this is based on,
The embodiment of the present application also provides a kind of training method of disaggregated model, this method can be using arbitrarily with data-handling capacity
Processing equipment, including terminal or server.
Below by taking server as an example, in conjunction with concrete scene being introduced to the training method of the disaggregated model.Referring to fig. 4
Shown in disaggregated model training method scene framework figure, include server 410 in the scene, server 410 is from network
Chemical equation is acquired, the chemical equation based on acquisition generates training sample, then utilizes machine according to the training sample
Algorithm train classification models, until meeting training termination condition.
In order to enable the technical solution of the application it is clearer, it can be readily appreciated that below from the angle of server to classification mould
The training method of type describes in detail.
Referring to Fig. 5, the training method of disaggregated model provided by the embodiments of the present application includes:
S501: training sample is obtained.
The training sample includes compound and reaction information relevant to the compound, and the reaction information is at least
Including response rule.In practical application, server can crawl chemistry instead by crawling tool such as crawler from internet
Formula is answered, response rule and compound are chemically extracted in reaction equation, to generate training sample.
The embodiment of the present application provides a kind of implementation for extracting response rule, and specifically, collection of server chemistry is anti-
Formula is answered, atomic series mark then is carried out to the chemical equation, and based in the chemical equation abstraction reaction after mark
The heart determines response rule according to processing result followed by being filtered to the reaction center and clustering processing.In this way, service
Device can generate training sample based on the response rule.
Wherein, reaction center refers to participation reaction and on the influential atom of reaction and group.In organic chemical reactions
In, there is in a compound the case where there are multiple identical groups, therefore often for the position of clearly specific reactive group,
It needs to carry out the atom and group that atomic series mark (atom-atom mapping) carrys out orienting response to chemical equation, so
Reaction center extraction is carried out based on the chemical equation after mark afterwards.Since many organic chemical reactions have similar reaction class
Type, therefore obtained reaction center needs are filtered and clustering processing, determine response rule according to processing result, it is anti-based on this
Answer rule that can form response rule data set.
In practical application, the compound in the training sample exists in the form of vectors, the compound it is corresponding to
Amount can obtain in the following way: server by utilizing rubs root algorithm for the chemical formula of compound each in the chemical equation of acquisition
The minor structure sequence (referring to Fig. 6) that radius is 1 is converted to, is for certain organic compound, respectively centered on each atom, is found
Stain part indicates central atom in minor structure, Fig. 6, and virtual coil is possible be comprising the atom of only minor structure, in this way, can
To obtain the minor structure sequence that radius is 1, then, server by utilizing term vector model is converted to all minor structure sequences described
Then the corresponding vector of minor structure sequence is directed to each compound, to the corresponding vector of minor structure sequence of the compound into
Row adduction obtains the corresponding vector of the compound.
In the present embodiment, after collecting chemical equation, it is contemplated that a kind of compound can there are many expression way,
The molecular formula of compound can be normalized.For example, same molecular formula may be expressed as CN2C (=O) N (C) C (=O) C1
=C2N=CN1C or CN1C=NC2=C1C (=O) N (C) C (=O)-N2C can be expressed as following standard after normalization
Literary style Cn1cnc2c1c (=O) n (C) c (=O) n2C can substantially reduce data redudancy by so handling, and improve model
Training effectiveness and category of model accuracy.
S502: according to the training sample, using machine learning algorithm train classification models, until meeting training terminates item
Part.
Wherein, the disaggregated model is input with compound, is output with response rule.In train classification models, clothes
Training sample can be grouped by business device at random according to preset ratio, form training set and test set, the centrally stored trained number of training
According to test data in test set.Wherein, preset ratio can be arranged according to actual needs, as an example, preset ratio
It can be 7:3.
Then, the training data in server by utilizing training set, in conjunction with machine learning algorithm train classification models, specially
The classifying quality of disaggregated model is continued to optimize by adjusting parameter, wherein classifying quality can pass through Receiver Operating Characteristics' song
Area (the Area surrounded under line (receiver operating characteristic curve, ROC) with reference axis
Under Curve, AUC) it is characterized, AUC refers at random when selecting a positive sample and negative sample, current sorting algorithm root
Positive sample is come to the probability of negative sample signature according to the score value being calculated, it is clear that AUC value is bigger, and current sorting algorithm is got over
It is possible that positive sample is come before negative sample, can preferably classify.When server train classification models meet training knot
When beam condition, disaggregated model at this time can be used for execute classification task.Wherein, which can be according to reality
Demand and be arranged, such as can be set to the loss function of disaggregated model and be in the loss letter of convergence state or disaggregated model
Number is less than preset threshold.
In some possible implementations, the machine learning algorithm includes gradient decline decision Tree algorithms, random gloomy
Any one or more in woods algorithm, algorithm of support vector machine and neural network algorithm.In order to make it easy to understand, being declined with gradient
For decision Tree algorithms, disaggregated model is established using the method for GBDT, on the training data by adjusting the quantity of regression tree, every
The parameters such as depth, loss function, the learning rate of a lone tree continue to optimize more classifying quality AUC, to obtain preferable mould
Type.
In addition, server can also carry out the feelings such as over-fitting to trained disaggregated model using the test data in test set
Condition verifying, if over-fitting occurs in disaggregated model, further adjusts parameter, to obtain optimal models.
From the foregoing, it will be observed that the embodiment of the present application provides a kind of training method of disaggregated model, this method is to include compound
And the training sample that response rule corresponding with compound is formed in interior information, using machine learning algorithm to disaggregated model into
Row training can be used for response rule prediction, compound is input to this when disaggregated model meets training termination condition
Disaggregated model, the disaggregated model, that is, corresponding response rule of exportable generation chemical combination.Pass through the disaggregated model of this method training
Can the corresponding response rule of predictive compound by carrying out decomposition retrosynthetic analysis to classifying rules chemical conjunction may be implemented
At route Automated Design, and disaggregated model accuracy with higher and stronger generalization ability, it can be applied to new
Reaction, without training again, so improves chemical synthesis route design efficiency.
It should be noted that the embodiment of the present application also utilize the embodiment of the present application training disaggregated model and traditional change
It learns synthetic route design method to test (Fig. 7 A, Fig. 7 B and Fig. 7 C) several unknown compounds, base in the embodiment of the present application
It is significantly shorter than the chemical synthesis route that traditional approach is designed in the chemical synthesis route of disaggregated model design, conjunction is greatly lowered
At the process complexity of target compound.
The above are the determination method of chemical synthesis route provided by the embodiments of the present application, the training method of disaggregated model one
A little specific implementations are based on this, and the embodiment of the present application also provides corresponding devices, below will be from the angle of function modoularization
Above-mentioned apparatus is introduced.
The structural schematic diagram of the determining device of chemical synthesis route shown in Figure 8, the device 800 include:
Receiving module 810, for receiving chemical synthesis request, the chemical synthesis request is for request target compound
Chemical synthesis route；
Module 820 is obtained, for the target compound to be input to disaggregated model, is obtained and the target compound pair
The response rule answered；The disaggregated model is with compound for input, and take response rule as the machine learning model of output；
Decomposing module 830 is obtained for carrying out decomposition retrosynthetic analysis to the target compound according to the response rule
The raw material to needed for synthesizing the target compound；
Determining module 840, for determining chemical synthesis route corresponding with the target compound according to the raw material.
Optionally, the decomposing module 830 is specifically used for:
Decomposition retrosynthetic analysis is carried out to the target compound according to the response rule and obtains decomposition product；
Whether search the decomposition product from raw database with the determination decomposition product is raw material；
Decomposition product is not raw material if it exists, then using the decomposition product as target compound, executes the target compound
It is input to disaggregated model, obtains response rule corresponding with the target compound, and according to the response rule to the mesh
The step of mark compound carries out decomposition retrosynthetic analysis, until the decomposition product is raw material.
Optionally, the receiving module 810 is also used to:
It receives and the amendment that the chemical synthesis route is modified is requested；
Described device 800 further include:
Correction module corrects the chemical synthesis route for requesting in response to the amendment.
Optionally, described device 800 further include:
Conversion module, for the target compound to be converted to corresponding vector using term vector model；
Then the acquisition module 820 is specifically used for when target compound is input to disaggregated model:
The corresponding vector of the target compound is input to disaggregated model.
Optionally, the machine learning model includes gradient decline decision-tree model, Random Forest model, support vector machines
Any one or more in model and neural network model.
The embodiment of the present application also provides a kind of training device of disaggregated model, structural schematic diagram shown in Figure 9 should
Device 900 includes:
Obtain module 910, for obtaining training sample, the training sample include compound and with the compound phase
The reaction information of pass, the reaction information include at least response rule；
Training module 920 is used for according to the training sample, using machine learning algorithm train classification models, until full
Foot training termination condition；
Wherein, the disaggregated model is input with compound, is output with response rule.
Optionally, described device 900 further include:
Acquisition module, for acquiring chemical equation；
Abstraction module, for carrying out atomic series mark to the chemical equation, and based on the chemical reaction after mark
Formula abstraction reaction center；
Processing module determines reaction rule according to processing result for being filtered to the reaction center and clustering processing
Then.
Optionally, the compound in the training sample exists in the form of vectors, described device 900 further include:
First conversion module, for being converted the chemical formula of compound each in the chemical equation of acquisition using root algorithm is rubbed
The minor structure sequence for being 1 for radius；
Second conversion module, for all minor structure sequences to be converted to the minor structure sequence pair using term vector model
The vector answered；
Module is summed it up, for being directed to each compound, the corresponding vector of minor structure sequence of the compound is added
Vector corresponding with the compound is obtained.
Optionally, the machine learning model includes gradient decline decision Tree algorithms, random forests algorithm, support vector machines
Any one or more in algorithm and neural network algorithm.
Fig. 8 to Fig. 9 from the angle of function modoularization to the determining device of chemical synthesis route provided by the embodiments of the present application,
The training device of disaggregated model is introduced, and the embodiment of the present application also provides corresponding equipment, below will be from hardware entities
Angle above-mentioned apparatus provided by the embodiments of the present application is introduced.
Figure 10 is a kind of structural schematic diagram of equipment provided by the embodiments of the present application, which can be server, the clothes
Business device 1000 can generate bigger difference because configuration or performance are different, may include one or more central processing units
(central processing units, CPU) 1022 (for example, one or more processors) and memory 1032, one
The storage medium 1030 of a or more than one storage application program 1042 or data 1044 (such as deposit by one or more magnanimity
Store up equipment).Wherein, memory 1032 and storage medium 1030 can be of short duration storage or persistent storage.It is stored in storage medium
1030 program may include one or more modules (diagram does not mark), and each module may include in server
Series of instructions operation.Further, central processing unit 1022 can be set to communicate with storage medium 1030, in server
The series of instructions operation in storage medium 1030 is executed on 1000.
Server 1000 can also include one or more power supplys 1026, one or more wired or wireless nets
Network interface 1050, one or more input/output interfaces 1058, and/or, one or more operating systems 1041, example
Such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM etc..
The step as performed by server can be based on the server architecture shown in Fig. 10 in above-described embodiment.
Wherein, CPU 1022 is for executing following steps:
Chemical synthesis request is received, the chemical synthesis request is used for the chemical synthesis route of request target compound；
The target compound is input to disaggregated model, obtains response rule corresponding with the target compound；Institute
Stating disaggregated model is with compound for input, and take response rule as the machine learning model of output；
Decomposition retrosynthetic analysis is carried out to the target compound according to the response rule to obtain synthesizing the targeted
Raw material needed for closing object；
Chemical synthesis route corresponding with the target compound is determined according to the raw material.
Optionally, CPU 1022 can be also used for executing times of the determination method of chemical synthesis route in the embodiment of the present application
Anticipate a kind of implementation the step of.
The embodiment of the present application also provides a kind of equipment for train classification models, have and equipment phase shown in Figure 10
Same structure, wherein CPU 1022 is for executing following steps:
Training sample is obtained, the training sample includes compound and reaction information relevant to the compound, institute
Reaction information is stated including at least response rule；
According to the training sample, using machine learning algorithm train classification models, until meeting training termination condition；
Wherein, the disaggregated model is input with compound, is output with response rule.
Optionally, CPU 1022 can be also used for execute the embodiment of the present application in disaggregated model training method it is any one
The step of kind implementation.
The embodiment of the present application also provides another equipment, as shown in figure 11, for ease of description, illustrate only and this Shen
Please the relevant part of embodiment, it is disclosed by specific technical details, please refer to the embodiment of the present application method part.The terminal can be with
Being includes mobile phone, tablet computer, personal digital assistant (full name in English: Personal Digital Assistant, English contracting
Write: PDA), point-of-sale terminal (full name in English: Point of Sales, english abbreviation: POS), any terminal such as vehicle-mounted computer set
It is standby, taking the terminal as an example:
Figure 11 shows the block diagram of the part-structure of mobile phone relevant to terminal provided by the embodiments of the present application.With reference to figure
11, mobile phone includes: radio frequency (full name in English: Radio Frequency, english abbreviation: RF) circuit 1110, memory 1120, defeated
Enter unit 1130, display unit 1140, sensor 1150, voicefrequency circuit 1160, Wireless Fidelity (full name in English: wireless
Fidelity, english abbreviation: WiFi) components such as module 1170, processor 1180 and power supply 1190.Those skilled in the art
It is appreciated that handset structure shown in Figure 11 does not constitute the restriction to mobile phone, it may include more more or fewer than illustrating
Component perhaps combines certain components or different component layouts.
Memory 1120 can be used for storing software program and module, and processor 1180 is stored in memory by operation
1120 software program and module, thereby executing the various function application and data processing of mobile phone.Memory 1120 can be led
It to include storing program area and storage data area, wherein storing program area can be needed for storage program area, at least one function
Application program (such as sound-playing function, image player function etc.) etc.；Storage data area, which can be stored, uses institute according to mobile phone
Data (such as audio data, phone directory etc.) of creation etc..In addition, memory 1120 may include high random access storage
Device, can also include nonvolatile memory, and a for example, at least disk memory, flush memory device or other volatibility are solid
State memory device.
Processor 1180 is the control centre of mobile phone, using the various pieces of various interfaces and connection whole mobile phone,
By running or execute the software program and/or module that are stored in memory 1120, and calls and be stored in memory 1120
Interior data execute the various functions and processing data of mobile phone, to carry out integral monitoring to mobile phone.Optionally, processor
1180 may include one or more processing units；Preferably, processor 1180 can integrate application processor and modulation /demodulation processing
Device, wherein the main processing operation system of application processor, user interface and application program etc., modem processor is mainly located
Reason wireless communication.It is understood that above-mentioned modem processor can not also be integrated into processor 1180.
In the embodiment of the present application, processor 1180 included by the terminal is also with the following functions:
Chemical synthesis request is received, the chemical synthesis request is used for the chemical synthesis route of request target compound；
The target compound is input to disaggregated model, obtains response rule corresponding with the target compound；Institute
Stating disaggregated model is with compound for input, and take response rule as the machine learning model of output；
Decomposition retrosynthetic analysis is carried out to the target compound according to the response rule to obtain synthesizing the targeted
Raw material needed for closing object；
Chemical synthesis route corresponding with the target compound is determined according to the raw material.
Optionally, the processor 1180 is also used to execute the training method of disaggregated model provided by the embodiments of the present application
The step of any one implementation.
The embodiment of the present application also provides a kind of for the abnormal equipment for commenting on detection, the equipment application in service side,
With structure identical with equipment shown in Figure 11, wherein processor 1180 is for executing following steps:
Training sample is obtained, the training sample includes compound and reaction information relevant to the compound, institute
Reaction information is stated including at least response rule；
According to the training sample, using machine learning algorithm train classification models, until meeting training termination condition；
Wherein, the disaggregated model is input with compound, is output with response rule.
Optionally, processor 1180 can be also used for execute the embodiment of the present application in disaggregated model training method it is any
A kind of the step of implementation.
The embodiment of the present application also provides a kind of computer readable storage medium, for storing program code, the program code
For executing in a kind of determination method of chemical synthesis route described in foregoing individual embodiments or the training method of disaggregated model
Any one embodiment.
The embodiment of the present application also provides a kind of computer program product including instruction, when run on a computer,
So that computer executes a kind of determination method of chemical synthesis route described in foregoing individual embodiments or the training of disaggregated model
Any one embodiment in method.
It is apparent to those skilled in the art that for convenience and simplicity of description, the system of foregoing description,
The specific work process of device and unit, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.
In several embodiments provided herein, it should be understood that disclosed device and method can pass through it
Its mode is realized.For example, the apparatus embodiments described above are merely exemplary, for example, the division of the unit, only
Only a kind of logical function partition, there may be another division manner in actual implementation, such as multiple units or components can be tied
Another system is closed or is desirably integrated into, or some features can be ignored or not executed.Another point, it is shown or discussed
Mutual coupling, direct-coupling or communication connection can be through some interfaces, the INDIRECT COUPLING or logical of device or unit
Letter connection can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It should be appreciated that in this application, " at least one (item) " refers to one or more, and " multiple " refer to two or two
More than a."and/or" indicates may exist three kinds of relationships, for example, " A and/or B " for describing the incidence relation of affiliated partner
It can indicate: only exist A, only exist B and exist simultaneously tri- kinds of situations of A and B, wherein A, B can be odd number or plural number.Word
Symbol "/" typicallys represent the relationship that forward-backward correlation object is a kind of "or"." at least one of following (a) " or its similar expression, refers to
Any combination in these, any combination including individual event (a) or complex item (a).At least one of for example, in a, b or c
(a) can indicate: a, b, c, " a and b ", " a and c ", " b and c ", or " a and b and c ", and wherein a, b, c can be individually, can also
To be multiple.
The above, above embodiments are only to illustrate the technical solution of the application, rather than its limitations；Although referring to before
Embodiment is stated the application is described in detail, those skilled in the art should understand that: it still can be to preceding
Technical solution documented by each embodiment is stated to modify or equivalent replacement of some of the technical features；And these
It modifies or replaces, the spirit and scope of each embodiment technical solution of the application that it does not separate the essence of the corresponding technical solution.
Claims (13)
1. a kind of determination method of chemical synthesis route, which is characterized in that the described method includes:
Chemical synthesis request is received, the chemical synthesis request is used for the chemical synthesis route of request target compound；
The target compound is input to disaggregated model, obtains response rule corresponding with the target compound；Described point
Class model is with compound for input, and take response rule as the machine learning model of output；
Decomposition retrosynthetic analysis is carried out to the target compound according to the response rule to obtain synthesizing the target compound
Required raw material；
Chemical synthesis route corresponding with the target compound is determined according to the raw material.
2. the method according to claim 1, wherein it is described according to the response rule to the target compound
It carries out decomposing retrosynthetic analysis and obtains raw material needed for synthesizing the target compound and include:
Decomposition retrosynthetic analysis is carried out to the target compound according to the response rule and obtains decomposition product；
Whether search the decomposition product from raw database with the determination decomposition product is raw material；
Decomposition product is not raw material if it exists, then using the decomposition product as target compound, execution inputs the target compound
To disaggregated model, response rule corresponding with the target compound is obtained, and according to the response rule to the targeted
The step of object carries out decomposition retrosynthetic analysis is closed, until the decomposition product is raw material.
3. the method according to claim 1, wherein the method also includes:
It receives and the amendment that the chemical synthesis route is modified is requested；
It is requested in response to the amendment, corrects the chemical synthesis route.
4. according to claim 1 to method described in 3 any one, which is characterized in that the method also includes:
The target compound is converted into corresponding vector using term vector model；
It is then described the target compound is input to disaggregated model to include:
The corresponding vector of the target compound is input to disaggregated model.
5. according to claim 1 to method described in 3 any one, which is characterized in that the machine learning model includes gradient
Decline any one or more in decision-tree model, Random Forest model, supporting vector machine model and neural network model.
6. a kind of training method of disaggregated model, which is characterized in that the described method includes:
Training sample is obtained, the training sample includes compound and reaction information relevant to the compound, described anti-
Information is answered to include at least response rule；
According to the training sample, using machine learning algorithm train classification models, until meeting training termination condition；
Wherein, the disaggregated model is input with compound, is output with response rule.
7. according to the method described in claim 6, it is characterized in that, response rule obtains in the following way in the training sample
It arrives:
Acquire chemical equation；
Atomic series mark is carried out to the chemical equation, and based on the chemical equation abstraction reaction center after mark；
The reaction center is filtered and clustering processing, response rule is determined according to processing result.
8. according to the method described in claim 6, it is characterized in that, the compound in the training sample is deposited in the form of vectors
The corresponding vector of the compound obtains in the following way:
The chemical formula of compound each in the chemical equation of acquisition is converted into the minor structure sequence that radius is 1 using root algorithm is rubbed
Column；
All minor structure sequences are converted into the corresponding vector of the minor structure sequence using term vector model；
For each compound, the corresponding vector of minor structure sequence of the compound is summed up to obtain the compound pair
The vector answered.
9. according to method described in claim 6 to 8 any one, which is characterized in that the machine learning algorithm includes gradient
Decline any one or more in decision Tree algorithms, random forests algorithm, algorithm of support vector machine and neural network algorithm.
10. a kind of determining device of chemical synthesis route, which is characterized in that described device includes:
Receiving module, for receiving chemical synthesis request, chemistry of the chemical synthesis request for request target compound is closed
At route；
Module is obtained, for the target compound to be input to disaggregated model, is obtained corresponding with the target compound anti-
Answer rule；Institute's disaggregated model is with compound for input, and take response rule as the machine learning model of output；
Decomposing module obtains synthesis institute for carrying out decomposition retrosynthetic analysis to the target compound according to the response rule
State raw material needed for target compound；
Determining module, for determining chemical synthesis route corresponding with the target compound according to the raw material.
11. a kind of training device of disaggregated model, which is characterized in that described device includes:
Module is obtained, for obtaining training sample, the training sample includes compound and relevant to the compound anti-
Information is answered, the reaction information includes at least response rule；
Training module, for using machine learning algorithm train classification models, being tied until meeting training according to the training sample
Beam condition；
Wherein, the disaggregated model is input with compound, is output with response rule.
12. a kind of equipment, which is characterized in that the equipment includes processor and memory:
Said program code is transferred to the processor for storing program code by the memory；
The processor is used to be closed according to the chemistry described in any one of claim 1 to 5 of the instruction execution in said program code
At the training method of the described in any item disaggregated models of determination method or claim 6 to 9 of route.
13. a kind of computer readable storage medium, which is characterized in that the computer readable storage medium is for storing program generation
Code, said program code require the determination method or right of 1 to 5 described in any item chemical synthesis routes for perform claim
It is required that the training method of 6 to 9 described in any item disaggregated models.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910194016.3A CN109872780A (en) | 2019-03-14 | 2019-03-14 | A kind of determination method and device of chemical synthesis route |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910194016.3A CN109872780A (en) | 2019-03-14 | 2019-03-14 | A kind of determination method and device of chemical synthesis route |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109872780A true CN109872780A (en) | 2019-06-11 |
Family
ID=66920313
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910194016.3A Pending CN109872780A (en) | 2019-03-14 | 2019-03-14 | A kind of determination method and device of chemical synthesis route |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109872780A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111524557A (en) * | 2020-04-24 | 2020-08-11 | 腾讯科技（深圳）有限公司 | Inverse synthesis prediction method, device, equipment and storage medium based on artificial intelligence |
| CN112037868A (en) * | 2020-11-04 | 2020-12-04 | 腾讯科技（深圳）有限公司 | Training method and device for neural network for determining molecular reverse synthetic route |
| WO2020254884A1 (en) * | 2019-06-18 | 2020-12-24 | International Business Machines Corporation | Generating control settings for a chemical reactor |
| CN112652365A (en) * | 2020-03-05 | 2021-04-13 | 腾讯科技（深圳）有限公司 | Inverse synthesis processing method, device, electronic equipment and computer readable storage medium |
| CN113053474A (en) * | 2021-04-13 | 2021-06-29 | 南京信息工程大学 | Light chargeable material screening system and operation method thereof |
| CN113140260A (en) * | 2020-01-20 | 2021-07-20 | 腾讯科技（深圳）有限公司 | Method and device for predicting reactant molecular composition data of composition |
| EP3896699A1 (en) * | 2020-03-06 | 2021-10-20 | Accenture Global Solutions Limited | Using machine learning for generating chemical product formulations |
| CN113782109A (en) * | 2021-09-13 | 2021-12-10 | 烟台国工智能科技有限公司 | Reactant derivation method and reverse synthesis derivation method based on Monte Carlo tree |
| WO2022130648A1 (en) * | 2020-12-18 | 2022-06-23 | 富士通株式会社 | Information processing program, information processing method, and information processing device |
| CN115206450A (en) * | 2022-09-15 | 2022-10-18 | 药融云数字科技(成都)有限公司 | Synthetic route recommendation method and terminal |
| CN115312135A (en) * | 2022-07-21 | 2022-11-08 | 苏州沃时数字科技有限公司 | Method, system, device and storage medium for predicting chemical reaction conditions |
| US11675334B2 (en) | 2019-06-18 | 2023-06-13 | International Business Machines Corporation | Controlling a chemical reactor for the production of polymer compounds |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008106080A2 (en) * | 2007-02-27 | 2008-09-04 | University Of Pittsburgh - Of The Commonwealth System Of Higher Education | Synthesis of desacetoxytubulysin h and analogs thereof |
| CN101789047A (en) * | 2010-02-05 | 2010-07-28 | 四川大学 | Method for evaluating synthesization of organic small-molecule compounds based on reverse synthesis |
-
2019
- 2019-03-14 CN CN201910194016.3A patent/CN109872780A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008106080A2 (en) * | 2007-02-27 | 2008-09-04 | University Of Pittsburgh - Of The Commonwealth System Of Higher Education | Synthesis of desacetoxytubulysin h and analogs thereof |
| CN101789047A (en) * | 2010-02-05 | 2010-07-28 | 四川大学 | Method for evaluating synthesization of organic small-molecule compounds based on reverse synthesis |
Non-Patent Citations (3)
| Title |
|---|
| JAMES LAW ET.AL: "Route Designer: A Retrosynthetic Analysis Tool Utilizing Automated Retrosynthetic Rule Generation", 《J. CHEM. INF. MODEL》 * |
| JAVIER L. BAYLON ET.AL: "Enhancing Retrosynthetic Reaction Prediction with Deep Learning Using Multiscale Reaction Classification", 《JOURNAL OF CHEMICAL INFORMATION AND MODELING》 * |
| MARWIN H. S. SEGLER ET.AL: "Neural-Symbolic Machine Learning for Retrosynthesis and Reaction Prediction", 《CHEMISTRY》 * |
Cited By (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2599837A (en) * | 2019-06-18 | 2022-04-13 | Ibm | Generating control settings for a chemical reactor |
| US11675334B2 (en) | 2019-06-18 | 2023-06-13 | International Business Machines Corporation | Controlling a chemical reactor for the production of polymer compounds |
| WO2020254884A1 (en) * | 2019-06-18 | 2020-12-24 | International Business Machines Corporation | Generating control settings for a chemical reactor |
| GB2599837B (en) * | 2019-06-18 | 2023-05-03 | Ibm | Generating control settings for a chemical reactor |
| US11520310B2 (en) | 2019-06-18 | 2022-12-06 | International Business Machines Corporation | Generating control settings for a chemical reactor |
| CN113140260A (en) * | 2020-01-20 | 2021-07-20 | 腾讯科技（深圳）有限公司 | Method and device for predicting reactant molecular composition data of composition |
| CN113140260B (en) * | 2020-01-20 | 2023-09-08 | 腾讯科技（深圳）有限公司 | Method and device for predicting reactant molecular composition data of composition |
| WO2021174895A1 (en) * | 2020-03-05 | 2021-09-10 | 腾讯科技（深圳）有限公司 | Retrosynthesis processing method and apparatus, and electronic device and computer-readable storage medium |
| CN112652365A (en) * | 2020-03-05 | 2021-04-13 | 腾讯科技（深圳）有限公司 | Inverse synthesis processing method, device, electronic equipment and computer readable storage medium |
| EP4092682A4 (en) * | 2020-03-05 | 2023-08-23 | Tencent Technology (Shenzhen) Company Limited | Retrosynthesis processing method and apparatus, and electronic device and computer-readable storage medium |
| EP3896699A1 (en) * | 2020-03-06 | 2021-10-20 | Accenture Global Solutions Limited | Using machine learning for generating chemical product formulations |
| CN111524557B (en) * | 2020-04-24 | 2024-04-05 | 腾讯科技（深圳）有限公司 | Inverse synthesis prediction method, device, equipment and storage medium based on artificial intelligence |
| CN111524557A (en) * | 2020-04-24 | 2020-08-11 | 腾讯科技（深圳）有限公司 | Inverse synthesis prediction method, device, equipment and storage medium based on artificial intelligence |
| CN112037868A (en) * | 2020-11-04 | 2020-12-04 | 腾讯科技（深圳）有限公司 | Training method and device for neural network for determining molecular reverse synthetic route |
| WO2022095659A1 (en) * | 2020-11-04 | 2022-05-12 | 腾讯科技（深圳）有限公司 | Method and apparatus for training neural network for determining molecule retrosynthesis route |
| EP4213153A4 (en) * | 2020-11-04 | 2024-03-20 | Tencent Tech Shenzhen Co Ltd | Method and apparatus for training neural network for determining molecule retrosynthesis route |
| CN112037868B (en) * | 2020-11-04 | 2021-02-12 | 腾讯科技（深圳）有限公司 | Training method and device for neural network for determining molecular reverse synthetic route |
| WO2022130648A1 (en) * | 2020-12-18 | 2022-06-23 | 富士通株式会社 | Information processing program, information processing method, and information processing device |
| CN113053474A (en) * | 2021-04-13 | 2021-06-29 | 南京信息工程大学 | Light chargeable material screening system and operation method thereof |
| CN113782109A (en) * | 2021-09-13 | 2021-12-10 | 烟台国工智能科技有限公司 | Reactant derivation method and reverse synthesis derivation method based on Monte Carlo tree |
| CN115312135A (en) * | 2022-07-21 | 2022-11-08 | 苏州沃时数字科技有限公司 | Method, system, device and storage medium for predicting chemical reaction conditions |
| CN115312135B (en) * | 2022-07-21 | 2023-10-20 | 苏州沃时数字科技有限公司 | Chemical reaction condition prediction method, system, device and storage medium |
| CN115206450B (en) * | 2022-09-15 | 2022-12-06 | 药融云数字科技(成都)有限公司 | Synthetic route recommendation method and terminal |
| CN115206450A (en) * | 2022-09-15 | 2022-10-18 | 药融云数字科技(成都)有限公司 | Synthetic route recommendation method and terminal |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109872780A (en) | A kind of determination method and device of chemical synthesis route | |
| US10622098B2 (en) | Systems and methods for predicting chemical reactions | |
| Aziz et al. | Dimension reduction methods for microarray data: a review | |
| CN107862173B (en) | Virtual screening method and device for lead compound | |
| Nyathi et al. | Comparison of a genetic algorithm to grammatical evolution for automated design of genetic programming classification algorithms | |
| Nemati et al. | A novel ACO–GA hybrid algorithm for feature selection in protein function prediction | |
| Bebek et al. | PathFinder: mining signal transduction pathway segments from protein-protein interaction networks | |
| Liu et al. | Evolving feature selection | |
| He et al. | A generalization of vit/mlp-mixer to graphs | |
| CN108108743A (en) | Abnormal user recognition methods and the device for identifying abnormal user | |
| CN109727637B (en) | Method for identifying key proteins based on mixed frog-leaping algorithm | |
| JP2022505540A (en) | Systems and methods for active transfer learning using deep feature extraction | |
| CA2942106A1 (en) | Aligning and clustering sequence patterns to reveal classificatory functionality of sequences | |
| CN111724866A (en) | Automated prediction of biological response of compounds based on chemical information | |
| Al-Maqaleh et al. | A genetic algorithm for discovering classification rules in data mining | |
| Wang et al. | Molecular property prediction based on a multichannel substructure graph | |
| Ye et al. | Adaptive unsupervised feature learning for gene signature identification in non-small-cell lung cancer | |
| Flores-Garrido et al. | Mining maximal frequent patterns in a single graph using inexact matching | |
| Azevedo et al. | Hybrid approaches to optimization and machine learning methods: a systematic literature review | |
| Garimella et al. | Churn prediction using optimized deep learning classifier on huge telecom data | |
| CN112115996B (en) | Image data processing method, device, equipment and storage medium | |
| Hansen et al. | A VNS heuristic for escaping local extrema entrapment in normalized cut clustering | |
| Nelson et al. | Higher order interactions: detection of epistasis using machine learning and evolutionary computation | |
| Ye et al. | High‐Dimensional Feature Selection Based on Improved Binary Ant Colony Optimization Combined with Hybrid Rice Optimization Algorithm | |
| Choong et al. | Jack and Masters of all Trades: One-Pass Learning Sets of Model Sets From Large Pre-Trained Models |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190611 |
|
| RJ01 | Rejection of invention patent application after publication |