CN114649097A - Medicine efficacy prediction method based on graph neural network and omics information - Google Patents

Medicine efficacy prediction method based on graph neural network and omics information Download PDF

Info

Publication number
CN114649097A
CN114649097A CN202210206033.6A CN202210206033A CN114649097A CN 114649097 A CN114649097 A CN 114649097A CN 202210206033 A CN202210206033 A CN 202210206033A CN 114649097 A CN114649097 A CN 114649097A
Authority
CN
China
Prior art keywords
drug
node
graph
medicine
predicting
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210206033.6A
Other languages
Chinese (zh)
Inventor
鲁路
邓健平
倪世豪
孙术宁
李悦
刘欣
欧阳小露
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangzhou University of Traditional Chinese Medicine
Original Assignee
Guangzhou University of Traditional Chinese Medicine
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Guangzhou University of Traditional Chinese Medicine filed Critical Guangzhou University of Traditional Chinese Medicine
Priority to CN202210206033.6A priority Critical patent/CN114649097A/en
Publication of CN114649097A publication Critical patent/CN114649097A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
    • G16H20/10—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to drugs or medications, e.g. for ensuring correct administration to patients

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Medical Informatics (AREA)
  • General Health & Medical Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • Primary Health Care (AREA)
  • Epidemiology (AREA)
  • Biomedical Technology (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Pathology (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • Computing Systems (AREA)
  • Biophysics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Medicinal Chemistry (AREA)
  • Investigating Or Analysing Biological Materials (AREA)

Abstract

The invention relates to the technical field of drug screening, and particularly discloses a drug efficacy prediction method based on a graph neural network and omics information. The method comprises the following steps: collecting unified transcriptome data; converting the unified transcriptome data into a biological process as a node characteristic, expressing the relationship as an edge, and constructing a structural relationship diagram of the existing medicine and the biological process; vectorizing and expressing the node characteristics by using a method to be selected; and performing aggregation updating on the node characteristics according to the structural relationship diagram, comparing the similarity between the compound to be detected and the existing medicine through the node characteristic vector, and predicting and screening the functions of the compound to be detected. The invention starts with the medicine transcriptome data as the basis, integrates mass medicine transcriptome data through a big data processing method, extracts medicine characteristics through a graph neural network and performs functional clustering on the existing medicines. And then comparing the compound to be predicted with the clustered existing drugs, and predicting the function of the compound to be predicted by the method.

Description

Medicine efficacy prediction method based on graph neural network and omics information
Technical Field
The invention relates to the technical field of drug screening, in particular to a drug efficacy prediction method based on a graph neural network and omics information.
Background
Drug development is a field with great risk, long cycle and high cost. According to statistics, the development of a new drug usually costs 5-10 billion dollars, and needs 10-15 years or even longer development time, and has great contingency and blindness. The drug screening is the basis of new drug development, and is an engineering like a sea fishing needle to find out a suitable compound as a candidate drug for preclinical experiments from tens of millions of molecules.
Current drug screening methods can be divided into two categories: firstly, phenotype screening, wherein the phenotype is the sum of the character characteristics of individuals with specific genotypes under certain environmental conditions. Drug phenotype screening is a screening method based on phenotypic changes in specific environments. Phenotypic drug screening has been over a long period of time, and more so, traditional phenotypic drug screening is based on animal disease models by observing changes in the drug's phenotype in the body. With the advance of technology, minute elements such as cells, proteins, genes, etc., which constitute a living body, can be studied at a deeper level. In particular, the method can be divided into animal disease model screening and in-vitro drug screening. Secondly, computer aided drug design, which is a method for designing and optimizing a lead compound by taking computer chemistry as the basis and simulating, calculating and budgeting the relationship between the drug and the biological macromolecule of a receptor through a computer. Computer-aided drug design is actually the optimization and design of lead compounds by modeling and calculating the interaction between receptor and ligand. In particular, the method can be divided into direct drug design in which a specific ligand is designed directly according to the three-dimensional structure of a receptor protein molecule, and indirect drug design based on the ligand under the condition that the protein structure of a drug target is not clear.
However, the above drug screening methods also have certain drawbacks: for example, phenotype screening does not depend on the structure of a drug, but faces huge challenges, the first is that the drug screening is finally aimed at being applied to human, human diseases are very complicated processes, and the first difficulty is how to determine a phenotype available for screening. Furthermore, our current phenotypic screening systems are based primarily on cells, drosophila, nematodes, zebrafish, etc. These animals can be screened for high throughput, but the animals are also more divergent from human disease. We can generalize the challenges of phenotypic screening as a contradiction between complex diseases and simple approaches.
For example, computer-aided drug design suffers from excessive dependence on structure and targeting. Screening drugs, especially the first-created drugs, always likes the drugs with definite target, clear structure and targeting property. The disadvantages of the above methods, whether molecular docking in direct drug design or pharmacophore modeling, quantitative structure-activity relationships in indirect drug design, can be attributed to the dependence on compound structure as well as targeting. The first is the challenge of the method itself, one of which is the accuracy of protein structure information in public databases, which is the primary condition for ensuring the accuracy of the results. Although researchers have high-end crystallography hardware, data processing, and structure optimization software in structural studies, false positives and inconsistent data may still occur. The second method relies on protein structure, and the method for analyzing protein structure mainly uses X-ray crystallography, NMR (nuclear magnetic resonance, NMR), nuclear magnetic resonance and cryoelectron microscopy, which are limited by cost and time, and have certain experimental difficulty, and only when the protein can be crystallized, the X-ray crystallography can analyze the protein structure. However, most proteins are difficult to crystallize, accounting for 60% of currently approved drugs. Nuclear magnetic resonance is also only applicable to small molecule proteins. Furthermore, the method is based on the ideal situation that a disease can be treated by a target and a target can be precisely targeted by a drug. It is known that most diseases, especially various medical diseases, are a systemic, comprehensive pathophysiological progression not driven by which single gene, single target, on the other hand, a compound is also difficult to target only one target.
In this regard, the scientific community has also tried to overcome these problems, for example, using human-derived model animals and 3D culture techniques, which aim to make the screened phenotypes more and more towards human complex diseases, and there has been a series of progress, especially in the fields of regenerative medicine, tumor and immunity. However, the complexity of the phenotype means that the more investment, the more expensive the leading-edge phenotypic model, the less significant the community can use. Therefore, even if complex phenotypic screening systems are established, the previous high investment has limited the means of drug screening by many practitioners.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a medicine efficacy prediction method based on a graph neural network and omics information. The invention starts with the medicine transcriptome data as the basis, integrates mass medicine transcriptome data through a big data processing method, extracts medicine characteristics through a graph neural network and performs functional clustering on the existing medicines. And then comparing the compound to be predicted with the clustered existing drugs, and predicting the function of the compound to be predicted by the method.
In order to achieve the purpose, the invention adopts the technical scheme that:
the invention provides a medicine efficacy prediction method based on a graph neural network and omics information, which comprises the following steps:
1) collecting: collecting the existing medicine transcriptome data, and unifying the transcriptome data through gene set enrichment analysis to obtain unified transcriptome data;
2) constructing a structural relationship diagram: converting the unified transcriptome data into a biological process as a node characteristic, expressing the relationship as an edge, and constructing a structural relationship diagram of the existing medicine and the biological process;
3) node learning represents: expressing the learning idea by combining the structural relation graph, and performing vectorization expression on the node characteristics in the step 2) by using a candidate method to obtain a node characteristic vector;
4) graph neural network: and performing aggregation updating on the node characteristics according to the constructed structural relationship diagram, and comparing the similarity between the compound to be detected and the existing medicine through the extracted node characteristic vector of the medicine so as to predict and screen the functions of the compound to be detected.
As a preferred embodiment of the drug efficacy prediction method, the specific steps of collecting the existing drug transcriptome data are as follows:
collecting information of the existing medicines:
firstly, comprehensively collecting the information of the existing drugs, ensuring the normal work of a neural network of a subsequent graph, and integrating a drug catalog on a drug bank;
mining transcriptome information:
data retrieval is intended to be performed in the GEO, Array Express database according to the integrated drug catalog.
High-throughput data often accompanies massive information, for example, at present, the one-time sequencing depth can reach more than 10G, and although the transcriptome data is rich, the problems of large experimental time span and non-uniform measurement standards (different selected samples, different sequencing methods and the like) exist. The technical key point of the invention is that complicated data must be unified, and how to apply a big data processing method to process the data, how to select and set the algorithm of the graph neural network, and the like are important problems.
As a preferred embodiment of the method for predicting the efficacy of the medicine, the candidate method in the step 3) comprises DEEP WALK, LINE, Metapath2Vec + +, GATNE.
As a preferred embodiment of the method for predicting drug efficacy of the present invention, the method for aggregate updating of node characteristics in step 4) includes GAT and HAN.
In a preferred embodiment of the method for predicting the efficacy of a drug according to the present invention, the drug to be tested includes acacetin.
As a preferred embodiment of the method for predicting drug efficacy of the present invention, the node characteristics are classified into two expression modes, Up and Down, according to the biological process.
As a preferred embodiment of the method for predicting the efficacy of the medicament, Metapath2Vec + + is adopted as a candidate method according to Drug-Biological Process-Drug (V)bp-ap) And Drug- -Biological Process- -Drug (V)bp-down) The two element paths carry out random walk on the node characteristics, the sequence obtained by the random walk is used as a training corpus, the training corpus is collected according to a positive and negative sampling optimization method of metapath2Vec + +, and the nodes are expressed in a vectorization mode.
As a preferred embodiment of the Drug efficacy prediction method, when HAN is selected to perform aggregation update on node characteristics, a structural relationship graph which is a heterogeneous graph G with three types of node attributes is constructed according to Drug-Biological Process-Drug (V)bp-ap) And Drug- -Biological Process- -Drug (V)bp-down) The two element paths split the abnormal composition graph G into two abnormal composition graphs Gup、Gdown(ii) a And importing the learned vector as a node feature, and calculating the attention of the node and the semantic attention to obtain a final node feature vector.
Compared with the prior art, the invention has the following beneficial effects:
in phenotype screening, although an effective screening model can relatively well restore the human disease environment, the cost is high, and although virtual screening solves the cost problem, a method which excessively depends on a structure and an excessively idealized targeting idea also become inevitable defects. The invention provides a medicine efficacy prediction method based on a graph neural network and omics information, which not only avoids the high cost problem of phenotype screening, but also solves the problem that virtual screening excessively depends on structural information by informationizing phenotype data.
Drawings
FIG. 1 is a visualization of an implementation of example 1;
FIG. 2 is an enlarged view of a portion of FIG. 1;
FIG. 3 is a graph showing the change in fasting plasma glucose of rats in each group;
FIG. 4 is a graph showing the postprandial blood glucose changes of rats in each group;
FIG. 5 is a graph of the results of the insulin resistance test.
Detailed Description
To better illustrate the objects, aspects and advantages of the present invention, the present invention will be further described with reference to the accompanying drawings and specific embodiments.
In the following examples, the experimental methods used were all conventional methods unless otherwise specified, and the materials, reagents and the like used were commercially available without otherwise specified.
The invention realizes the efficacy prediction of the compound based on the graph neural network and omics information. Among them, the graph neural network belongs to the category of Artificial Intelligence (AI).
Examples
This example was carried out using a portion of the marketed drug transcriptome and the test compound farnesin (Acacetin) (fig. 1, fig. 2 are schematic representations).
(ii) converting the gene set information of the transcriptome into a biological process by GSEA analysis.
② the graph structure G is (V, E) constructed by the medicine and all the involved biological processes and the relationship thereof. The node V of the biological process is divided into two representations, Up and Down, according to whether the biological process is Up or Down.
(iii) selecting Metapath2Vec + + graph learning representation method according to Drug-Biological Process-Drug (V)bp-ap) And Drug- -Biological Process- -Drug (V)bp-down) Two meta-paths (meta-path) perform random walk (random walk) on all nodes, and the sequence obtained by the random walk is used as the corpus. Training corpora are collected according to a positive and negative sampling optimization method of metapath2Vec + +, and nodes are expressed in a vectorization mode.
Metapath2Vec + + is a graph representation learning method developed based on the idea of Word2Vec in Natural Language Processing (NLP). The Word2Vec method is that all words in the corpus are traversed to be used as central words, words above and below the central words are selected to be used as positive samples of adjacent words according to the set window length, and other words are used as negative sample definition labels. And maximizing the learning parameters according to the positive and negative sample conditional probability, and performing word vectorization on all words.
Metapath2Vec + + uses a sequence obtained by randomly walking a graph structure according to a semantic relationship (in this embodiment, a relationship between a drug and a biological process) as a corpus, traverses nodes therein, and vectorizes and represents the nodes in the graph according to the method of Wrod2 Vec.
And 4, using HAN (heterologous Graph attachment network) to analyze the Graph neural network. According to the second step, the graph G constructed by the invention is an isomeric graph G with three types of node attributes, and the isomeric graph G is split into two isomeric graphs G according to the third step of two meta-pathsap、Gdown. And importing the vector learned in the third step as a node feature. Node-level and semantic-level are computed. And obtaining a final node feature vector. The characteristics are used for comparing, predicting and screening the efficacy of the medicine.
HAN is a Graph Attention network (GAT) that learns an anomaly Graph based on an Attention mechanism, and a Graph Attention network (GAT) is a space-based Graph Convolution Network (GCN).
GCN graph convolution networks generalize convolution operations from traditional data (e.g., images) to graph data. The core idea is to learn a function map f (), by which a node can aggregate its own features with those of its neighboring nodes to generate a new representation.
The GAT graph attention network differs from the graph convolution network by introducing an attention mechanism, which is a method of learning the weights of different neighbor nodes while aggregating the characteristics of the neighbor nodes around the node, where the weights are attention parameters, and the attention mechanism can assign different attention scores to each neighbor compared to the GCN treating all neighbors of the node equally, thereby identifying more important neighbor nodes.
The neural network algorithms of GCN, GAT and other graphs are also based on isomorphism, while the embodiment constructs an isomerous graph containing the semantic relation between the medicine and the biological process, and HAN is a graph neural network based on GAT learning heteromorphism. In a meta-path (meta-path) under a specific semantic relationship, each node performs GAT, aggregates the characteristics of neighboring nodes, and constructs node-level attention characteristics. In a heterogeneous graph, generally, a node contains multiple types of semantics, a node feature of a specific semantic only contains information of a certain aspect of the node, and to learn more complicated node features, multiple semantic information based on meta-paths needs to be fused. In the embodiment, node-level features are firstly aggregated according to meta-paths under two semantics of a drug ascending biological process and a drug descending biological process, and the features under the two semantics are aggregated again as final features of nodes through a semantic attention mechanics learning mechanism, so that the drug efficacy is predicted and screened according to the features. The dark spots of the left continuous part in fig. 1 are ascending biological processes and the light spots are descending biological processes; FIG. 1 shows an enlarged view of a portion of the agglomerated part of a marketed drug and farnesoid, and FIG. 2 shows a portion of the enlarged view of FIG. 1, wherein the vector similarity between Metformin (Metformin) and the compound to be tested is high, and the prediction result shows that the farnesoid may have the effect of controlling blood glucose.
And verifying and predicting results in-vitro and in-vivo experiments.
And (3) anti-blood sugar test: male SD rats (5-7 weeks old; body weight 140-. The same dose of distilled water was added as a control group and HFF + Acacetin was used as an experimental group.
Referring to FIGS. 3-4, there are shown fasting and postprandial blood glucose changes in rats, respectively, illustrating the significant blood glucose control effect of farnesoid.
Insulin resistance test: l6 rat skeletal muscle cells were treated with Free Fatty Acid (FFA) (0.75mM) for 24 hours and glucose uptake changes in rat skeletal muscle cells in the insulin resistant state after farnesoid treatment were assessed by the fluorescent indicator 2-NBDG. The results show that farnesin increased glucose uptake (see fig. 5). The results of the blood glucose control by farnesoid were verified in the above experiments.
Finally, it should be noted that the above embodiments are only used for illustrating the technical solutions of the present invention and not for limiting the protection scope of the present invention, and although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions can be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims (8)

1. A drug efficacy prediction method based on a graph neural network and omics information is characterized by comprising the following steps:
1) collecting: collecting the existing medicine transcriptome data, and unifying the transcriptome data through gene set enrichment analysis to obtain unified transcriptome data;
2) constructing a structural relationship diagram: converting the unified transcriptome data into a biological process as a node characteristic, expressing the relationship as an edge, and constructing a structural relationship diagram of the existing medicine and the biological process;
3) node learning represents: expressing the learning idea by combining the structural relation graph, and performing vectorization expression on the node characteristics in the step 2) by using a candidate method to obtain a node characteristic vector;
4) graph neural network: and performing aggregation updating on the node characteristics according to the constructed structural relationship diagram, and comparing the similarity between the compound to be detected and the existing medicine through the extracted node characteristic vector of the medicine so as to predict and screen the functions of the compound to be detected.
2. The method of predicting drug efficacy of claim 1, wherein the step of collecting the existing drug transcriptome data comprises:
collecting information of the existing medicines:
firstly, comprehensively collecting the information of the existing drugs, ensuring the normal work of a neural network of a subsequent graph, and integrating a drug catalog on a drug bank;
mining transcriptome information:
data retrieval is intended to be performed in the GEO, Array Express database according to the integrated drug catalog.
3. The method for predicting drug efficacy according to claim 1, wherein the candidate methods in step 3) comprise DEEP wall, LINE, Metapath2Vec + +, GATNE.
4. The method for predicting drug efficacy according to claim 1, wherein the method for performing aggregate update on the node characteristics in step 4) comprises GAT and HAN.
5. The method of predicting the efficacy of a drug according to claim 1, wherein the drug to be tested comprises acacetin.
6. The method of predicting drug efficacy according to claim 1, wherein the node characteristics are classified into Up and Down according to biological processes.
7. The method for predicting Drug efficacy according to claim 3, wherein when Metapath2Vec + + is used as the candidate method, the node features are randomly walked according to two meta-paths of Drug-Biological Process-Drug and Drug-Biological Process-Drug, a sequence obtained by random walk is used as the training corpus, the training corpus is collected according to a positive and negative sampling optimization method of Metapath2Vec + +, and the nodes are expressed in a vectorization manner.
8. The method for predicting Drug efficacy according to claim 4, wherein when HAN is selected for aggregation and update of node features, the constructed structural relationship graph is an isomeric graph G with three types of node attributes, and the isomeric graph G is split into two isomeric graphs G according to two meta-paths of Drug-Biological Process-Drug and Drug-Biological Process-Drugup、Gdown(ii) a And importing the learned vector as a node feature, and calculating the attention of the node and the semantic attention to obtain a final node feature vector.
CN202210206033.6A 2022-03-04 2022-03-04 Medicine efficacy prediction method based on graph neural network and omics information Pending CN114649097A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210206033.6A CN114649097A (en) 2022-03-04 2022-03-04 Medicine efficacy prediction method based on graph neural network and omics information

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210206033.6A CN114649097A (en) 2022-03-04 2022-03-04 Medicine efficacy prediction method based on graph neural network and omics information

Publications (1)

Publication Number Publication Date
CN114649097A true CN114649097A (en) 2022-06-21

Family

ID=81993289

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210206033.6A Pending CN114649097A (en) 2022-03-04 2022-03-04 Medicine efficacy prediction method based on graph neural network and omics information

Country Status (1)

Country Link
CN (1) CN114649097A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116451176A (en) * 2023-06-15 2023-07-18 武汉大学人民医院（湖北省人民医院） Deep learning-based medicine spectrum data analysis method and device
CN116597902A (en) * 2023-04-24 2023-08-15 浙江大学 Method and device for screening multiple groups of chemical biomarkers based on drug sensitivity data
WO2023217290A1 (en) * 2022-10-11 2023-11-16 之江实验室 Genophenotypic prediction based on graph neural network

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023217290A1 (en) * 2022-10-11 2023-11-16 之江实验室 Genophenotypic prediction based on graph neural network
CN116597902A (en) * 2023-04-24 2023-08-15 浙江大学 Method and device for screening multiple groups of chemical biomarkers based on drug sensitivity data
CN116597902B (en) * 2023-04-24 2023-12-01 浙江大学 Method and device for screening multiple groups of chemical biomarkers based on drug sensitivity data
CN116451176A (en) * 2023-06-15 2023-07-18 武汉大学人民医院（湖北省人民医院） Deep learning-based medicine spectrum data analysis method and device
CN116451176B (en) * 2023-06-15 2024-01-12 武汉大学人民医院（湖北省人民医院） Deep learning-based medicine spectrum data analysis method and device

Similar Documents

Publication Publication Date Title
CN114649097A (en) Medicine efficacy prediction method based on graph neural network and omics information
CN106529205B (en) It is a kind of based on drug minor structure, the drug targets Relationship Prediction method of molecule character description information
Erfanian et al. Deep learning applications in single-cell genomics and transcriptomics data analysis
US20090313189A1 (en) Method, system and apparatus for assembling and using biological knowledge
US20050165594A1 (en) System, method and apparatus for causal implication analysis in biological networks
WO2007084374A2 (en) Random forest modeling of cellular phenotypes
US20140052686A1 (en) Method, system and software arrangement for reconstructing formal descriptive models of processes from functional/modal data using suitable ontology
CN114334038B (en) Disease medicine prediction method based on heterogeneous network embedded model
Nagarajan et al. Predicting future scientific discoveries based on a networked analysis of the past literature
US20090187525A1 (en) System and method for network association inference, validation and pruning based on integrated constraints from diverse data
CN105740626A (en) Drug activity prediction method based on machine learning
CN114003734A (en) Breast cancer risk factor knowledge system model, knowledge map system and construction method
Alterovitz et al. Knowledge-based bioinformatics: from analysis to interpretation
Wolcott et al. Automated classification of estrous stage in rodents using deep learning
CN115472221A (en) Protein fitness prediction method based on deep learning
US20220208540A1 (en) System for Identifying Structures of Molecular Compounds from Mass Spectrometry Data
CN113223609B (en) Drug target interaction prediction method based on heterogeneous information network
Aziz et al. A novel hybrid approach for classifying osteosarcoma using deep feature extraction and multilayer perceptron
CN117476252A (en) Etiology and pathology prediction method based on knowledge graph
Sundar et al. An intelligent prediction model for target protein identification in hepatic carcinoma using novel graph theory and ann model
US20240303544A1 (en) Graph database techniques for machine learning
Wang et al. Prediction of the disease causal genes based on heterogeneous network and multi-feature combination method
Kumbier et al. Signed iterative random forests to identify enhancer-associated transcription factor binding
CN117976047B (en) Key protein prediction method based on deep learning
Darmawahyuni et al. Health-related Data Analysis using Metaheuristic Optimization and Machine Learning

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