Method for predicting lncRNA-miRNA interaction based on sequence complementary site information

Abstract

The invention provides a method for predicting lncRNA-miRNA interaction based on sequence complementary site information, which solves the problems of predicting lncRNA-miRNA interaction and the like, and comprises the following steps: S1, constructing positive and negative samples; s2, extracting combined point location information of the positive set and the negative set by adopting Miranda; s3, selecting the maximum value of the binding fraction in the positive set and the negative set as a representative binding site representing the interaction of the lncRNA and the miRNA; s4, performing feature construction and feature selection; s5, predicting interaction by adopting a machine learning model; and S6, performing data comparison to evaluate the model prediction accuracy. The method has the advantages of good prediction effect, high universality and the like.

Description

Technical Field

[0001] The present invention belongs to the technical field of bioinformatics, and particularly relates to a method for predicting lncRNA-miRNA interactions based on sequence complementary site information. Background Art

[0002] Existing methods based on graph learning have difficulty predicting potential interactions with isolated lncRNAs or miRNAs that do not interact with any other miRNAs or lncRNAs. Because lncRNA sequences are poorly conserved across species, conclusions drawn from studies in one species may not apply to others. Interaction profiling-based methods rely on data obtained through wet-bed experiments, which can be incomplete and restrict the study of unverified lncRNA-miRNA pairs. Existing methods based on data mining and deep learning may have limitations in the accuracy and efficiency of predicting lncRNA-miRNA interactions. Existing methods are unable to investigate associations in unverified lncRNA-miRNA data, which may hold valuable research value.

[0003] To address the shortcomings of existing technologies, researchers have conducted extensive research and proposed a variety of solutions. For example, a Chinese patent document [202311482172.2] discloses a non-coding RNA interaction prediction model and method based on graph comparative learning. The model first acquires a graph feature extraction module and a prediction module. The graph feature extraction module uses graph comparative learning to obtain graph-augmented data of the lncRNA-miRNA interaction network. The prediction module predicts the probability of an interaction between lncRNA and miRNA based on a traversal set of node representations from lncRNAs and miRNAs in the graph-augmented data. Both the graph feature extraction module and the prediction module utilize neural networks.

[0004] The above scheme has solved the problem of RNA interaction prediction to a certain extent, but it still has many shortcomings, such as limitations in predicting lncRNA-miRNA interactions. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for predicting lncRNA-miRNA interactions based on sequence complementary site information, which has good applicability and high accuracy in predicting lncRNA-miRNA interactions, in order to solve the above problems.

[0006] To achieve the above objectives, the present invention adopts the following technical solutions: a method for predicting lncRNA-miRNA interactions based on sequence complementary site information, comprising the following steps:

[0007] S1: Construction of positive and negative samples;

[0008] S2: Use Miranda to extract the combined point information of the positive set and the negative set;

[0009] S3: The maximum binding score in the positive set and the negative set is selected as the representative binding site representing the interaction between lncRNA and miRNA;

[0010] S4: perform feature construction and feature selection;

[0011] S5: Predict interactions using machine learning models;

[0012] S6: Perform data comparison to evaluate the model prediction accuracy.

[0013] In the above-mentioned method for predicting lncRNA-miRNA interactions based on sequence complementary site information, step S1 obtains a verified positive sample dataset from the lncRNASNP2 database. The lncRNASNP2 database is selected because it contains laboratory-verified lncRNA-miRNA interaction data, ensuring the accuracy and reliability of the samples.

[0014] In the above method for predicting lncRNA-miRNA interactions based on sequence complementary site information, step S1 obtains human lncRNA sequences from GENCODE and extracts human miRNA sequences from the miRbase database.

[0015] In the above-mentioned method for predicting lncRNA-miRNA interactions based on sequence complementary site information, the construction of negative samples in step S1 includes the following steps:

[0016] S11: Construct a sequence set with a similarity of 80% or more between each lncRNA and miRNA, denoted as and ;

[0017] Among them, Lp represents the sequence set with a similarity greater than or equal to 80% with any lncRNA sequence p. represents any sequence in the sequence set, Mq represents the sequence set with a similarity greater than or equal to 80% with any miRNA sequence q, Represents any sequence in the sequence set;

[0018] S12: Randomly select one sequence from each of lncRNA and miRNA, marked as Li and Mj respectively, and the combination pair is (L i , M j );

[0019] S13: Extract the positive samples and L i All miRNAs that interact are recorded as LM i ={m q}, for each m q Extract similar sequences from Mq and construct a new similarity matrix , if LM1 i Including M j It is considered that there is a potential interaction and (L i , M j ) interaction pair and repeat step S12; if it does not contain, retain (L i , M j ) interaction pairs;

[0020] S14: Extract the positive samples with M j All lncRNAs that interact are recorded as ML j ={l p}, for each l p From l p Extract similar sequences and construct a new similarity matrix , if ML1 j Contains L i It is considered that there is a potential interaction and (L i , M j ) interaction pair and repeat steps S12 and S13; if it does not contain, retain (L i , M j ) interaction pairs;

[0021] S15: Repeat steps S12 to S14 until the amount of positive and negative samples reaches a balance.

[0022] In the above-mentioned method for predicting lncRNA-miRNA interactions based on sequence complementary site information, step S2 includes the following steps:

[0023] S21: lncRNA and miRNA involved in the interaction are used as inputs of Miranda;

[0024] S22: The Miranda tool contains two key parameters: score and energy. Setting the score to 80 and the energy to -1 generates binding site information that covers all positive samples. We name the resulting binding site dataset J under these parameter settings.

[0025] S23: Select the binding site information involved in the positive and negative samples from J, name the binding site information data set extracted from the positive samples Z, and name the binding site information data set extracted from the negative samples F;

[0026] S24: In Z and F, each pair of lncRNA-miRNA interactions contains N binding sites, where N is greater than 1. We will screen each site in these interaction pairs, and finally retain only the binding sites that can truly reflect the interaction between lncRNA and miRNA;

[0027] S25: The binding site information set obtained after positive set screening is used as J p , the binding site information set obtained after negative set screening is used as J n The binding site information of the negative set is also processed as above, and then J p and J n The binding site sequences of lncRNA and miRNA were extracted.

[0028] In the above method for predicting lncRNA-miRNA interactions based on sequence complementary site information, step S3 obtains the binding site information set selected by the maximum binding score, and the positive set is named J ps , name the set of negative sets J ns , then J ps and J ns The binding site sequences of lncRNA and miRNA were extracted.

[0029] In the above method for predicting lncRNA-miRNA interactions based on sequence complementary site information, the feature selection process of step S4 takes the sequence in step S3 as input, and combines the k-mer features of the collection site sequence with the J ps and J ns Binding fraction and binding site free energy as signatures of lncRNA-miRNA interactions.

[0030] In the above-mentioned method for predicting lncRNA-miRNA interactions based on sequence complementary site information, step S5 inputs the features extracted in step S4 into a machine learning model, and the machine learning model includes RF, XGBoost, SVM, Decisiontree, and LightGBM.

[0031] In the above method for predicting lncRNA-miRNA interactions based on sequence complementary site information, step S5 selects the RF model as the prediction model.

[0032] In the above-mentioned method for predicting lncRNA-miRNA interactions based on sequence complementary site information, step S6 is compared with the LncMirNet method on the same dataset.

[0033] Compared with existing technologies, the advantages of the present invention are: taking the binding sites between lncRNA and miRNA as the key factor for predicting their interaction, a balanced set of positive and negative samples is constructed through a series of filtering and similarity analysis, which helps to improve the generalization ability and accuracy of the prediction model; after using the Miranda tool to predict the binding sites, the most representative binding site information is retained through a sophisticated screening process, which helps to improve the accuracy of the prediction; key features including kmer features, binding scores and binding site free energy are extracted, and these features are further optimized through a feature selection process, thereby enhancing the feature expression ability of the model; compared with the existing method LncMirNet, the prediction success rate (ACC) and AUC value are significantly improved, verifying the effectiveness and superiority of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 is a flow chart of the method of the present invention;

[0035] Figure 2 This is a comparison table of various evaluation indicators between the present invention and the LncMirNet method. DETAILED DESCRIPTION

[0036] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0037] like Figure 1-2 As shown in the figure, a method for predicting lncRNA-miRNA interactions based on sequence complementary site information is used to predict potential interactions. This method is the first lncRNA-miRNA interaction prediction method to use binding sites to predict interactions between them. The theoretical basis is that the interaction between lncRNA and miRNA may depend on specific binding sites. These sites may have specific patterns or motifs in the lncRNA sequence that interact with the seed region of the miRNA (usually the first 2-8 nucleotides of the miRNA).

[0038] This method uses Miranda to find relevant lncRNA and miRNA interaction sites. There are two parameters in Miranda (score, energy). When the parameters score = 80 and energy = -1 are set, the binding sites generated can cover all positive samples. The binding site data set generated under this parameter is named J. Therefore, in subsequent work, the binding site information involved in the positive and negative samples will be selected from J. The binding site information data set extracted from the positive samples is then named Z. At this time, each pair of interactions in Z contains N (N>1) binding sites. The N interaction pairs contained in each interaction pair are then screened. Finally, each pair of interactions only retains the binding sites that can truly reflect the interaction between lncRNA and miRNA. Finally, the binding site information set obtained after the positive set screening is named J p .

[0039] The binding site information of the negative set is also processed as above. In the next step, J p The lncRNA and miRNA binding site sequences are extracted and feature extraction is performed. The features used in this method are the 2mer features of the sequence, the binding score, the binding site free energy, and the binding site accessibility score. This step is also performed on the negative set. Finally, these features are used as input to the machine learning model to obtain the final prediction results. The specific steps include the following:

[0040] S1: Construction of positive and negative samples;

[0041] S2: Use Miranda to extract the combined point information of the positive set and the negative set;

[0042] S3: The maximum binding score in the positive set and the negative set is selected as the representative binding site representing the interaction between lncRNA and miRNA;

[0043] S4: perform feature construction and feature selection;

[0044] S5: Predict interactions using machine learning models;

[0045] S6: Perform data comparison to evaluate the model prediction accuracy.

[0046] Specifically, step S1 obtains a positive data set from the lncRNASNP2 database, in which real interactions have been confirmed by laboratory tests and research literature.

[0047] Specifically, step S1 retrieved human lncRNA sequences from GENCODE and human miRNA sequences from the miRbase database. To filter out true interactions, positive lncRNA-miRNA pairs were selected when both hsa-miR and ENST were present in the lncRNASNP2 record. Finally, 266 miRNAs, 1,663 lncRNAs, and 15,640 validated lncRNA-miRNA interactions and their corresponding sequences were obtained.

[0048] Furthermore, the construction of negative samples in step S1 includes the following steps:

[0049] S11: Construct a sequence set with a similarity of 80% or more between each lncRNA and miRNA, denoted as and ;

[0050] Among them, Lp represents the sequence set with a similarity greater than or equal to 80% with any lncRNA sequence p. Represents any sequence in the sequence set, and the value of m may range from 1 to 1663; Mq represents the sequence set with a similarity greater than or equal to 80% with any miRNA sequence q. Represents any sequence in the sequence set, and the value of n may range from 1 to 266;

[0051] S12: Randomly select one sequence each from 1663 lncRNAs and 266 miRNAs, labeled as Li and Mj respectively, and the combination pair is (L i , M j );

[0052] S13: Extract the positive samples and L i All miRNAs that interact are recorded as LM i ={m q}, for each m q Extract similar sequences from Mq and construct a new similarity matrix , if LM1 i Including M j It is considered that there is a potential interaction and (L i ,M j ) interaction pair and repeat step S12; if it does not contain, retain (L i , M j ) interaction pairs;

[0053] S14: Extract the positive samples with M j All lncRNAs that interact are recorded as ML j ={l p}, for each lp From l p Extract similar sequences and construct a new similarity matrix , if ML1 j Contains L i It is considered that there is a potential interaction and (L i , M j ) interaction pair and repeat steps S12 and S13; if it does not contain, retain (L i , M j ) interaction pairs;

[0054] S15: Repeat steps S12 to S14 until the amount of positive and negative samples reaches a balance.

[0055] In addition, step S2 includes the following steps:

[0056] S21: 1663 lncRNAs and 266 miRNAs involved in the interactions were used as inputs for Miranda;

[0057] S22: The Miranda tool contains two key parameters: score and energy. Setting the score to 80 and the energy to -1 generates binding site information that covers all positive samples. We name the resulting binding site dataset J under these parameter settings.

[0058] S23: Select the binding site information involved in the positive and negative samples from J, combine the binding site information data extracted from the positive samples into Z, and combine the binding site information data extracted from the negative samples into F;

[0059] S24: In Z and F, each pair of lncRNA-miRNA interactions contains N binding sites, where N is greater than 1. We will screen each site in these interaction pairs, and finally retain only the binding sites that can truly reflect the interaction between lncRNA and miRNA;

[0060] S25: The binding site information set obtained after positive set screening is used as J p , the binding site information set obtained after negative set screening is used as J n The binding site information of the negative set is also processed as above, and then J p and J n The binding site sequences of lncRNA and miRNA were extracted.

[0061] At the same time, step S3 obtains the binding site information set that is screened by the maximum binding score. In the interaction between lncRNA and miRNA, the completely complementary pairing between the seed region of miRNA (usually the first 8 nucleotides) and the target lncRNA will obtain a higher score. When miRNA binds to the target lncRNA, the lower the free energy of the entire double-stranded structure, the more stable the binding, and thus the higher the score. Some studies have shown that miRNA tends to bind to the 3' untranslated region (3'UTR) of lncRNA, and Miranda binding sites in these regions may obtain higher scores. Therefore, in J p and J n The binding site information with the maximum binding score is selected to better represent the interaction characteristics of the entire interaction pair. The set of positive sets is named j ps , name the set of negative sets J ns , then J ps and J ns The binding site sequences of lncRNA and miRNA were extracted.

[0062] It can be seen that the feature selection process of step S4 in ifeature takes the sequence in step S3 as input, and tries a total of 24 features about RNA sequences (2mer, 3mer, ASDC, CKSNAP, DAC, DACC, DCC, DPCP, Geary, mismatch, MMI, Moran, NAC, NMBroto, PCPseDNC, PseDNC(2_0.1), PseKNC(2_0.1_2), PseKNC(2_0.1_3), SCPseDNC, Subsequence, z_curve_9bit, z_curve_12bit, z_curve_36bit, z_curve_48bit). In addition to these features, J ps and J ns The binding fraction and free energy of the binding site are used as the characteristics of the interaction between lncRNA and miRNA. ps and J ns Binding fraction and binding site free energy as signatures of lncRNA-miRNA interactions.

[0063] Finally, after screening, the k-mer features and J ps and J nsThe binding score and binding site free energy are calculated in [1]. The k-mer is the most fundamental characteristic of RNA sequences. A k-mer is a sequence segment of length k. For RNA sequences, a k-mer can be a nucleotide sequence of length k, including four types: adenine (A), cytosine (C), guanine (G), and uracil (U). The k-mer feature is the statistical frequency of each k-mer in the RNA sequence, studying the distribution pattern of k-mers throughout the sequence. The binding score in step S3 is the maximum value. A higher score indicates a higher likelihood of binding between the miRNA and its target mRNA, and potentially biologically significant results. Free energy features are important parameters for assessing the stability of intermolecular interactions, particularly in the binding of biomolecules such as proteins and nucleic acids. In the case of miRNA binding with its target lncRNA, the binding free energy is used to quantify the stability of the interaction. Binding free energy refers to the energy released or absorbed when one molecule binds to another molecule from its free state under constant temperature and pressure. For miRNAs and their target lncRNAs, the binding free energy reflects the stability of the duplex structure formed between them. Binding free energy is typically estimated by calculating the free energy change (ΔG) of the double-stranded RNA (the duplex formed by the miRNA and its target mRNA). A lower value (a more negative value) indicates a more stable duplex structure and a tighter binding of the miRNA to its target mRNA.

[0064] Clearly, step S5 feeds the features extracted in step S4 into a machine learning model, including RF, XGBoost, SVM, decision tree, and LightGBM. RF performs well on the data, achieving an ACC of 86.334% on the training set and 87.02% on the test set. Therefore, step S5 selects the RF model as the prediction model.

[0065] Obviously, step S6 is compared with the LncMirNet method on the same dataset. The results show that the prediction success rate is improved by 34.109% and the AUC value is improved by 30.96% compared with the LncMirNet method. Therefore, these results show that the method of predicting lncRNA-miRNA interactions based on sequence complementary site information has good performance.

[0066] In summary, the principle behind this example is to innovatively leverage the interaction sites of long noncoding RNAs (lncRNAs) and microRNAs (miRNAs) to predict potential relationships between them, identifying binding sites as a key factor in predicting their interactions for the first time. Experimentally validated positive sample data were collected from the lncRNASNP2 database and combined with sequence information from the GENCODE and miRbase databases. A balanced set of positive and negative samples was constructed through a series of filtering and similarity analyses. Using the Miranda tool, lncRNA and miRNA binding sites were predicted, and the most representative binding sites were screened. Furthermore, key features, including kmer features, binding scores, and binding site free energies, were extracted and optimized through a feature selection process. After comparing multiple machine learning models, the random forest (RF) model was selected as the final prediction model due to its outstanding performance, achieving an accuracy of 86.334% on the training set and 87.02% on the test set.

[0067] The specific examples described in this document are intended solely to illustrate the principles and applications of the present invention. Persons skilled in the art may modify, supplement, or substitute the described embodiments with similar approaches without departing from the spirit of the present invention or exceeding the scope of the appended claims.

[0068] Although this article frequently uses terms such as positive and negative samples, it does not exclude the possibility of using other terms. These terms are used only to more conveniently describe and explain the essence of the present invention; interpreting them as any additional limitations is contrary to the spirit of the present invention.

Claims

1. A method for predicting lncRNA-miRNA interactions based on sequence complementary site information, characterized in that: The steps include: S1: Construction of positive and negative samples; S2: Use Miranda to extract the combined point information of the positive set and the negative set; S3: The maximum binding score in the positive set and the negative set is selected as the representative binding site representing the interaction between lncRNA and miRNA; S4: perform feature construction and feature selection; S5: Predict interactions using machine learning models; S6: Perform data comparison to evaluate the prediction accuracy of the model.

2. A method for predicting lncRNA-miRNA interactions based on sequence complementary site information according to claim 1, characterized in that: The step S1 obtains a verified positive sample data set from the lncRNASNP2 database.

3. The method for predicting lncRNA-miRNA interaction based on sequence complementary site information according to claim 1, characterized in that: The step S1 obtains human lncRNA sequences from GENCODE and extracts human miRNA sequences from the miRbase database.

4. The method for predicting lncRNA-miRNA interaction based on sequence complementary site information according to claim 1, characterized in that: The construction of negative samples in step S1 includes the following steps: S11: Construct a sequence set with a similarity of 80% or more for each lncRNA and miRNA, respectively, and record it as and Among them, Lp represents the sequence set with a similarity greater than or equal to 80% with any lncRNA sequence p. represents any sequence in the sequence set, Mq represents the sequence set with a similarity greater than or equal to 80% with any miRNA sequence q, Represents any sequence in the sequence set; S12: Randomly select a sequence from lncRNA and miRNA, marked as Li and Mj respectively, and the combination pair is (L i , M j ); S13: Extraction of positive samples and L i All miRNAs that interact are denoted as LM i ={m q }, for each m q′ Extract similar sequences from Mq and construct a new similarity matrix If LM1 i Contains M j It is considered that there is a potential interaction and (L i , M j ) interaction pair and repeat step S12; if it does not contain, retain (L i , M j ) interaction pairs; S14: Extract the positive samples with M j All lncRNAs that interact are recorded as ML j = {l p }, for each l p From l p Extract similar sequences and construct a new similarity matrix If ML1 j′ Contains L i It is considered that there is a potential interaction and (L i , M j ) interaction pair and repeat steps S12 and S13; if it does not contain, retain (L i , M j ) interaction pairs; S15: Repeat steps S12 to S14 until the data volume of positive and negative samples reaches a balance.

5. The method for predicting lncRNA-miRNA interaction based on sequence complementary site information according to claim 1, characterized in that: The step S2 comprises the following steps: S21: lncRNA and miRNA involved in the interaction are used as inputs of Miranda; S22: The Miranda tool contains two key parameters: score and energy. When the score is set to 80 and the energy is set to -1, the generated binding site information can cover all positive samples. We name the binding site data set obtained under this parameter setting J; S23: Select the binding site information involved in the positive and negative samples from J, combine the binding site information data extracted from the positive samples into Z, and combine the binding site information data extracted from the negative samples into F; S24: In Z and F, each pair of lncRNA-miRNA interactions contains N binding sites, where N is greater than 1. We will screen each site in these interaction pairs, and finally retain only the binding sites that can truly reflect the interaction between lncRNA and miRNA for each pair of interactions; S25: The binding site information set obtained after positive set screening is used as J p , the binding site information set obtained after negative set screening is taken as J n The binding site information of the negative set is also processed as above, and then J p and J n The binding site sequences of lncRNA and miRNA were extracted.

6. The method for predicting lncRNA-miRNA interaction based on sequence complementary site information according to claim 5, characterized in that: Step S3 obtains the binding site information set selected by the maximum binding score, and the positive set is named J ps , name the set of negative sets J ns , and then J ps and J ns The binding site sequences of lncRNA and miRNA were extracted.

7. The method for predicting lncRNA-miRNA interaction based on sequence complementary site information according to claim 6, characterized in that: The feature selection process of step S4 is described in ifeature, taking the sequence in step S3 as input, and combining the k-mer features of the collection site sequence and J ps and J ns Binding fraction and free energy of binding sites as signatures of lncRNA-miRNA interactions.

8. The method for predicting lncRNA-miRNA interaction based on sequence complementary site information according to claim 7, characterized in that: The step S5 inputs the features extracted in step S4 into the machine learning model, and the machine learning model includes RF, XGBoost, SVM, Decisiontree, and LightGBM.

9. The method for predicting lncRNA-miRNA interaction based on sequence complementary site information according to claim 8, characterized in that: The step S5 selects the RF model as the prediction model.

10. The method for predicting lncRNA-miRNA interaction based on sequence complementary site information according to claim 8, characterized in that: The step S6 is compared with the LncMirNet method on the same data set.

Images