Biomolecule pair matching analysis method, device, equipment and medium
By constructing the similarity matrix of influence factor between biomolecules and using machine learning classifiers, the problem of inefficient correlation analysis of biomolecules is solved, efficient and accurate matching analysis is achieved, and errors and inefficiency in manual operations are avoided.
Patent Information
- Application Number
- CN202411135644.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-19
- Publication Date
- 2024-12-13
AI Technical Summary
In the prior art, there is a lack of efficient and accurate matching methods for correlation analysis between biomolecules, resulting in low artificial comparison efficiency and prone to human errors.
By obtaining the influencing factors of at least two biomolecule samples, constructing the influencing factor similarity matrix, using artificial intelligence language model coding to obtain embedded vectors, calculating cosine similarity, and generating matching results through pre-trained machine learning classifiers to realize biological molecules automated matching analysis between the two.
Automatic matching analysis of massive influencing factors is realized, more similarity information is extracted, and more accurate prediction results are formed, avoiding a large number of manual operation and accuracy problems.
Smart Images
Figure CN119132418A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of bioinformatics, and in particular relates to a biomolecule matching method, device, equipment and medium. Background Art
[0002] Biomolecules are the basic units that make up living organisms, including proteins, nucleic acids (such as DNA and RNA), enzymes, etc. Biomolecules play a vital role in living organisms, and abnormalities or changes in their functions often lead to the occurrence of diseases. For example, misfolding of proteins may lead to Alzheimer's disease; gene mutations may cause cancer; and enzyme dysfunction may lead to metabolic disorders. The normal functioning of biomolecules is essential to maintaining the health and function of living organisms.
[0003] At present, there are many biomolecular disease and function annotation databases, which provide rich resources for researchers. For example, in China, the well-known one is the National Genome Science Data Center (NGDC). Internationally famous ones include Gene Ontology (GO), KEGG (Kyoto Encyclopedia of Genes and Genomes) and UniProt databases. These databases contain a large number of disease and function annotations of biomolecules, providing valuable data support for scientific research.
[0004] Take genes as an example. Genes play a vital role in disease research and functional analysis. Through in-depth research on genes, we can understand their roles in different diseases and explore their potential biological significance. In order to promote research in this field, many annotation databases have been established. These databases bring together a large amount of gene-disease relationship data to help scientists and researchers conduct in-depth bioinformatics analysis. Taking DisGeNET as an example, the number of diseases related to TP53 and PTEN genes is huge. The number of diseases associated with TP53 genes is about 2,500, and the number of diseases associated with PTEN genes is about 1,300. These annotation information provides valuable data support for the study of disease mechanisms, but at the same time, it also brings huge data processing challenges. Traditionally, manually comparing the relevant disease information of two genes is a complex and time-consuming task. For example, a researcher needs to compare the similarity between diseases related to TP53 and PTEN genes. Assuming that the similarity evaluation between each pair of diseases takes 0.5 seconds, it takes a total of about 450 hours for all possible disease pairs. This is not only inefficient, but also prone to human errors.
[0005] In summary, the prior art lacks an efficient and accurate matching method for correlation analysis between biological molecules. Summary of the invention
[0006] In view of the above analysis, the embodiments of the present invention aim to provide a biomolecule pair matching analysis method, device, equipment and medium to solve the problem of lack of efficient and accurate matching method for correlation analysis between biomolecules in the prior art.
[0007] The embodiment of the present invention provides a biomolecule pair matching analysis method, comprising the following steps:
[0008] S1. Obtain at least two biomolecule samples, wherein the biomolecule samples include a number of influencing factors;
[0009] S2, constructing an influence factor similarity matrix between the two biomolecule samples, wherein the influence factor similarity matrix includes the semantic similarity of each influence factor pair, and the two influence factors in the influence factor pair come from different biomolecule samples;
[0010] S3. Inputting the influencing factor similarity matrix into a classifier to obtain a matching result between the two biomolecule samples, wherein the classifier is pre-trained by machine learning based on labeled training data.
[0011] In some embodiments, the influencing factors include regulatory functions and diseases corresponding to the biomolecules.
[0012] In some embodiments, the modulated function and disease form comprises a character string.
[0013] In some embodiments, constructing the influencing factor similarity matrix between the two biomolecule samples in step S2 includes:
[0014] S21, inputting the impact factor into an artificial intelligence language model for encoding to obtain an embedding vector;
[0015] S22, calculating the cosine similarity between the embedding vectors corresponding to the two influencing factors in the influencing factor pair;
[0016] S23: forming the influencing factor similarity matrix based on the cosine similarities of at least some influencing factor pairs of the two biomolecule samples.
[0017] In some embodiments, the at least some influencing factor pairs include regulatory function pairs or disease pairs, wherein the regulatory function pairs are composed of two regulatory functions from different biomolecule samples, and the disease pairs are composed of two diseases from different biomolecule samples.
[0018] In some embodiments, the training process of the classifier includes:
[0019] constructing a training data set, wherein the training data set includes pairs of biomolecule samples with paired labels;
[0020] constructing an influencing factor similarity matrix between two biomolecule samples in the biomolecule sample pair;
[0021] Inputting the impact factor similarity matrix into the classifier to generate a pair analysis result;
[0022] A loss value is determined based on the difference between the pairing analysis result and the pairing label, and back-propagation training is performed on the classifier based on the loss value.
[0023] In some embodiments, the classifier includes a support vector machine, logistic regression, decision tree, random forest, gradient boosting, or a combination of one or more of a neural network.
[0024] The present invention also provides a biomolecule pair matching analysis device, comprising:
[0025] An acquisition module, acquiring at least two biomolecule samples, wherein the biomolecule samples include a number of influencing factors;
[0026] A similarity matrix construction module is used to construct an influencing factor similarity matrix between the two biomolecule samples, wherein the influencing factor similarity matrix includes the semantic similarity of each influencing factor pair, and the two influencing factors in the influencing factor pair come from different biomolecule samples;
[0027] The matching prediction module inputs the influencing factor similarity matrix into a classifier to obtain a matching result between the two biomolecule samples, wherein the classifier is pre-trained by machine learning based on labeled training data.
[0028] The present invention also provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the biological molecule pair matching analysis method as described in any of the above embodiments is implemented.
[0029] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the biomolecule pair matching analysis method as described in any of the above embodiments is implemented.
[0030] The above embodiments of the present invention have at least the following beneficial effects:
[0031] The present invention can convert numerous influencing factors contained in biological molecule samples into embedding vectors through an intelligent language model, and then calculate the similarity matrix of the biological molecule sample pair based on the embedding vectors of the influencing factor pairs across two biological molecule samples. Then, by performing feature extraction and binary classification on the similarity matrix for matching prediction, the present invention realizes automated matching analysis and prediction for biological molecule samples with massive influencing factors. By performing matching prediction based on the similarity matrix, the present invention can extract more information related to the similarity of two biological molecule samples, form a more accurate prediction result, and avoid a lot of manual work and the resulting accuracy problems. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the embodiments of this specification. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.
[0033] Figure 1 A schematic diagram of a process flow of a biomolecule pair matching analysis method provided by an embodiment of the present invention;
[0034] Figure 2 A schematic diagram of the structure of the biomolecule pair matching analysis device provided by the present invention;
[0035] Figure 3 This is a schematic diagram of the electronic device architecture provided by the present invention. DETAILED DESCRIPTION
[0036] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. It should be noted that, in the absence of conflict, the embodiments and features in the embodiments disclosed in this disclosure can be combined, separated, interchanged and / or rearranged with each other. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.
[0037] The terms used here are for the purpose of describing specific embodiments, and are not intended to be restrictive. As used here, unless the context clearly indicates otherwise, the singular forms "one (kind, person)" and "said (the)" are also intended to include plural forms. In addition, when the terms "comprise" and / or "include" and their variations are used in this specification, it is explained that there are stated features, integral bodies, steps, operations, parts, assemblies and / or their groups, but it is not excluded that there are or add one or more other features, integral bodies, steps, operations, parts, assemblies and / or their groups. It should also be noted that, as used here, the terms "substantially", "approximately" and other similar terms are used as approximate terms and not as degree terms, so that they are used to explain the inherent deviations of the measured values, calculated values and / or the values provided that will be recognized by those of ordinary skill in the art.
[0038] The present disclosure is described below through several specific embodiments. In order to keep the following description of the embodiments of the present disclosure clear and concise, the present disclosure omits detailed descriptions of known functions and known components. Figure 1 The present invention provides a method for analyzing matching of biomolecule pairs, including the following steps:
[0039] S1. Obtain at least two biomolecule samples, wherein the biomolecule samples include a number of influencing factors;
[0040] S2, constructing an influence factor similarity matrix between the two biomolecule samples, wherein the influence factor similarity matrix includes the semantic similarity of each influence factor pair, and the two influence factors in the influence factor pair come from different biomolecule samples;
[0041] S3. Inputting the influencing factor similarity matrix into a classifier to obtain a matching result between the two biomolecule samples, wherein the classifier is pre-trained by machine learning based on labeled training data.
[0042] Among them, the biological molecules in the biological molecule sample refer to molecules that play a key function in the body. They mainly include the following categories: Proteins: composed of amino acids, responsible for catalytic reactions (enzymes), transport molecules, provide structural support, etc. Nucleic acids: including DNA and RNA, which are responsible for the storage and transmission of genetic information, as well as the translation of information during protein synthesis. Lipids: mainly fats and phospholipids, etc., are involved in energy storage, cell membrane construction and signal transduction. Carbohydrates: such as sugars and polysaccharides, are mainly used for energy provision, cell recognition and structural support. These molecules play a key role in the physiological activities and functions of cells. In some embodiments, biological molecules can also refer to specific units such as genes, which are the basic units of biological genetic information and are responsible for the transmission and expression of genetic characteristics. They are specific sequences in DNA (deoxyribonucleic acid) molecules that encode instructions for synthesizing proteins or RNA, thereby determining various physiological characteristics and functions of organisms.
[0043] Taking genes as an example, genes play a vital role in disease research and functional analysis. Through in-depth research on genes, we can understand their role in different diseases and explore their potential biological significance. In order to promote research in this field, many annotation databases have been established. These databases have brought together a large amount of gene-disease relationship data to help scientists and researchers conduct in-depth bioinformatics analysis. In the present invention, the items affected by these biomolecules are called impact factors, for example, the impact factors of genes include multiple regulatory functions and diseases.
[0044] In some embodiments, the influencing factors include regulatory functions and diseases corresponding to the biomolecules.
[0045] In some embodiments, the modulated function and disease form comprises a character string.
[0046] In some embodiments, constructing the influencing factor similarity matrix between the two biomolecule samples in step S2 includes:
[0047] S21, inputting the impact factor into an artificial intelligence language model for encoding to obtain an embedding vector;
[0048] S22, calculating the cosine similarity between the embedding vectors corresponding to the two influencing factors in the influencing factor pair;
[0049] S23: forming the influencing factor similarity matrix based on the cosine similarities of at least some influencing factor pairs of the two biomolecule samples.
[0050] In some embodiments, the at least some influencing factor pairs include regulatory function pairs or disease pairs, wherein the regulatory function pairs are composed of two regulatory functions from different biomolecule samples, and the disease pairs are composed of two diseases from different biomolecule samples.
[0051] It should be understood that language models are the core technology in the field of natural language processing (NLP). By calculating the probability of occurrence of language units (such as words and sentences), they can understand and generate natural language text. Modern language models, especially models based on the Transformer architecture such as BERT and GPT, use deep learning technology to pre-train on large-scale text data, significantly improving semantic understanding and generation capabilities. These models can efficiently capture contextual information and are widely used in tasks such as text generation, machine translation, and semantic analysis.
[0052] In some embodiments, the training process of the classifier includes:
[0053] constructing a training data set, wherein the training data set includes pairs of biomolecule samples with paired labels;
[0054] constructing an influencing factor similarity matrix between two biomolecule samples in the biomolecule sample pair;
[0055] Inputting the impact factor similarity matrix into the classifier to generate a pair analysis result;
[0056] A loss value is determined based on the difference between the pairing analysis result and the pairing label, and back-propagation training is performed on the classifier based on the loss value.
[0057] Preferably, in some embodiments, before inputting the impact factor similarity matrix into the classifier, the method further includes converting the impact factor similarity matrix into one dimension.
[0058] Preferably, in a preferred embodiment of the present invention, after obtaining the influencing factor similarity matrix of the biomolecule sample pair, it also includes:
[0059] The similarities in the influencing factor similarity matrix are screened to be greater than a preset threshold, and similar influencing factor pairs associated with the biomolecule sample pairs are determined according to the rows and columns corresponding to the screened similarities, and a database is formed based on the similar influencing factor pairs and the corresponding biomolecule sample pairs.
[0060] In some embodiments, the classifier includes a support vector machine, logistic regression, decision tree, random forest, gradient boosting, or a combination of one or more of a neural network.
[0061] The present invention also provides a biomolecule pair matching analysis device, such as Figure 2 As shown, including:
[0062] An acquisition module, acquiring at least two biomolecule samples, wherein the biomolecule samples include a number of influencing factors;
[0063] A similarity matrix construction module is used to construct an influencing factor similarity matrix between the two biomolecule samples, wherein the influencing factor similarity matrix includes the semantic similarity of each influencing factor pair, and the two influencing factors in the influencing factor pair come from different biomolecule samples;
[0064] The matching prediction module inputs the influencing factor similarity matrix into a classifier to obtain a matching result between the two biomolecule samples, wherein the classifier is pre-trained by machine learning based on labeled training data.
[0065] The present invention also provides an electronic device, such as Figure 3 As shown, it includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the biomolecule pair matching analysis method as described in any of the above embodiments is implemented.
[0066] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the biomolecule pair matching analysis method as described in any of the above embodiments is implemented.
[0067] Computer readable storage media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0068] The professionals should further realize that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to the function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.
[0069] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
[0070] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.
Claims
1. A biomolecule pair matching analysis method, characterized in that: The following steps are involved: S1. Obtain at least two biomolecule samples, wherein the biomolecule samples include a number of influencing factors; S2, constructing an influence factor similarity matrix between the two biomolecule samples, wherein the influence factor similarity matrix includes the semantic similarity of each influence factor pair, and the two influence factors in the influence factor pair come from different biomolecule samples; S3. Inputting the influencing factor similarity matrix into a classifier to obtain a matching result between the two biomolecule samples, wherein the classifier is pre-trained by machine learning based on labeled training data.
2. The biomolecule pair matching analysis method according to claim 1, characterized in that: The influencing factors include the regulatory functions and diseases corresponding to the biomolecules.
3. The biomolecule pair matching analysis method according to claim 2, characterized in that: The regulatory functions and disease forms include character strings.
4. The biomolecule pair matching analysis method according to claim 2, characterized in that: In step S2, the influencing factor similarity matrix between the two biomolecule samples is constructed, including: S21, inputting the impact factor into an artificial intelligence language model for encoding to obtain an embedding vector; S22, calculating the cosine similarity between the embedding vectors corresponding to the two influencing factors in the influencing factor pair; S23: forming the influencing factor similarity matrix based on the cosine similarities of at least some influencing factor pairs of the two biomolecule samples.
5. The biomolecule pair matching analysis method according to claim 4, characterized in that: The at least some influencing factor pairs include regulatory function pairs or disease pairs, wherein the regulatory function pairs are composed of two regulatory functions from different biomolecule samples, and the disease pairs are composed of two diseases from different biomolecule samples.
6. The biomolecule pair matching analysis method according to claim 1, characterized in that: The training process of the classifier includes: constructing a training data set, wherein the training data set includes pairs of biomolecule samples with paired labels; constructing an influencing factor similarity matrix between two biomolecule samples in the biomolecule sample pair; Inputting the impact factor similarity matrix into the classifier to generate a pair analysis result; A loss value is determined based on the difference between the pairing analysis result and the pairing label, and back-propagation training is performed on the classifier based on the loss value.
7. The biomolecule pair matching analysis method according to claim 6, characterized in that: The classifier includes one or more combinations of support vector machines, logistic regression, decision trees, random forests, gradient boosting or neural networks.
8. A biomolecule pair matching analysis device, characterized in that: include: An acquisition module, acquiring at least two biomolecule samples, wherein the biomolecule samples include a number of influencing factors; A similarity matrix construction module is used to construct an influencing factor similarity matrix between the two biomolecule samples, wherein the influencing factor similarity matrix includes the semantic similarity of each influencing factor pair, and the two influencing factors in the influencing factor pair come from different biomolecule samples; The matching prediction module inputs the influencing factor similarity matrix into a classifier to obtain a matching result between the two biomolecule samples, wherein the classifier is pre-trained by machine learning based on labeled training data.
9. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the biomolecule pair matching analysis method according to any one of claims 1 to 7 is implemented.
10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the biomolecule pair matching analysis method according to any one of claims 1 to 7 is implemented.