CN110322926A

CN110322926A - The recognition methods of miRNA sponge module and device

Info

Publication number: CN110322926A
Application number: CN201910684738.7A
Authority: CN
Inventors: 张俊鹏; 饶妮妮
Original assignee: University of Electronic Science and Technology of China
Current assignee: Sun Shaoping
Priority date: 2019-07-26
Filing date: 2019-07-26
Publication date: 2019-10-11
Anticipated expiration: 2039-07-26
Also published as: CN110322926B

Abstract

The present invention provides recognition methods and the device of a kind of miRNA sponge module, and method includes: the expression matrix of the expression matrix and target gene that obtain the sponge gene of matched sample.According to the expression matrix of the expression matrix of sponge gene and target gene, multiple sponge genes-target gene coexpression module is obtained.It obtains in each sponge gene-target gene coexpression module, shares the sensibility canonical correlation coefficient when canonical correlation coefficient and shared miRNA between significance value, sponge gene and the target gene of miRNA.And determine whether each sponge gene-target gene coexpression module is miRNA sponge module according to the above parameter.Since this programme can measure the competition intensity between sponge gene and target gene under module level, and whether Dinghai silk floss gene-target gene coexpression module is miRNA sponge module really according to competition intensity between sponge gene and target gene, so that the identification of miRNA sponge module is more accurate.

Description

The recognition methods of miRNA sponge module and device

Technical field

The present invention relates to gene identification technical fields, recognition methods and dress more particularly, to a kind of miRNA sponge module It sets.

Background technique

MiRNA (MicroRNA, miRNA) is the endogenous a kind of non-coding tiny RNA for being about 21-23 nucleotide Molecule, it and messenger RNA (Messenger RNA, mRNA) are completed shearing by base pair complementarity principle, inhibited Translation and protein degradation process.Share identical Microrna response element (miRNA response elements, MREs) again The different transcripts for constituting competitive relation are referred to as miRNA sponge.MiRNA sponge is not independent but clustering or module Mode completes the important mission in its physiology and pathologic process.Therefore, miRNA sponge module is excavated for studying human physiological It is of great significance with pathologic process.

In the prior art, miRNA sponge module can be identified based on miRNA sponge interaction network, such as by Markov Clustering algorithm (Markov Cluster Algorithm, MCL) is identified.

Since miRNA sponge interaction network is mutually opposed and integrated by single sponge gene-target gene, do not consider altogether Influence of the miRNAs to competition intensity between sponge gene and target gene is enjoyed, the miRNA sponge module identified is not accurate enough.

Summary of the invention

The purpose of the present invention is to provide a kind of recognition methods of miRNA sponge module and devices, to solve the prior art Present in the not accurate enough problem of the miRNA sponge module that identifies.

In a first aspect, the embodiment of the present invention provides a kind of recognition methods of miRNA sponge module, comprising:

Obtain the expression matrix of the sponge gene of matched sample and the expression matrix of target gene.According to the expression of sponge gene The expression matrix of matrix and target gene obtains multiple sponge genes-target gene coexpression module, wherein each sponge gene-target Gene co-expressing module indicates a kind of gene that can be co-expressed with sponge gene and target gene.Obtain each sponge gene-target In gene co-expressing module, shares the canonical correlation coefficient between significance value, sponge gene and the target gene of miRNA and be total to Enjoy sensibility canonical correlation coefficient when miRNA.According to sponge gene and target in each sponge gene-target gene coexpression module Canonical correlation coefficient between the quantity of gene, the significance value of shared miRNA, sponge gene and target gene and shared Sensibility canonical correlation coefficient when miRNA determines whether each sponge gene-target gene coexpression module is miRNA sponge mould Block.

In alternative embodiments, it according to the expression matrix of the expression matrix of sponge gene and target gene, obtains multiple Sponge gene-target gene co-expresses module, comprising: according to clustering algorithm, the table of expression matrix and target gene to sponge gene It is clustered up to matrix, obtains multiple cluster results of sponge gene and target gene.By the sponge gene in each cluster result And target gene, module is co-expressed as a sponge gene-target gene.

In alternative embodiments, clustering algorithm includes: unidirectional clustering algorithm or bidirectional clustering algorithm.If clustering algorithm For unidirectional clustering algorithm, then according to the expression matrix of sponge gene, the expression matrix and matched sample of target gene, to sponge base Cause and target gene are clustered.If clustering algorithm is bidirectional clustering algorithm, according to the expression matrix of sponge gene, target gene The matched sample of expression matrix and predetermined fraction clusters the matched sample of sponge gene, target gene and predetermined fraction.

In alternative embodiments, the significance value of shared miRNA is obtained, comprising: according to preset miRNA- target base Because of regulation relationship, sponge gene-target gene is obtained by hypergeometric distribution method of inspection and is co-expressed in module, sponge gene and target base The significance value of miRNA is shared because between.

In alternative embodiments, the canonical correlation coefficient between sponge gene and target gene is obtained, comprising: according to sea The expression matrix of continuous gene and the expression matrix of target gene obtain the column vector of sponge gene and the column vector of target gene.It obtains Take the variance matrix and covariance matrix between the expression matrix of sponge gene and the expression matrix of target gene.According to sponge Column vector, the column vector of target gene, variance matrix, covariance matrix and the preset representative vectors of gene calculate canonical correlation Coefficient.

In alternative embodiments, sensibility canonical correlation coefficient when shared miRNA is obtained, comprising: obtain shared The canonical correlation coefficient between canonical correlation coefficient and shared miRNA and target gene between miRNA and sponge gene.According to sea Canonical correlation coefficient, shared miRNA between continuous gene and target gene and the canonical correlation coefficient between sponge gene are shared Canonical correlation coefficient between miRNA and target gene calculates the inclined canonical correlation coefficient between sponge gene and target gene.It will be extra large Canonical correlation coefficient between continuous gene and target gene subtracts the inclined canonical correlation coefficient between sponge gene and target gene, obtains Sensibility canonical correlation coefficient when shared miRNA.

In alternative embodiments, it obtains the canonical correlation coefficient between shared miRNA and sponge gene and shares Canonical correlation coefficient between miRNA and target gene, comprising: obtain the expression matrix of shared miRNA.According to shared miRNA's The expression matrix of expression matrix, the expression matrix of sponge gene and target gene obtains column vector, the sponge gene of shared miRNA Column vector and target gene column vector.Obtain the side between the expression matrix of shared miRNA and the expression matrix of sponge gene Poor matrix and covariance matrix.According to the expression of the column vector of shared miRNA, the column vector of sponge gene, shared miRNA Variance matrix and covariance matrix and preset representative vectors between matrix and the expression matrix of sponge gene calculate altogether Enjoy the canonical correlation coefficient between miRNA and sponge gene.Obtain the expression matrix of shared miRNA and the expression matrix of target gene Between variance matrix and covariance matrix.According to the column vector of shared miRNA, the column vector of target gene, shared miRNA Expression matrix and target gene expression matrix between variance matrix and covariance matrix and preset representative vectors, meter Calculate the canonical correlation coefficient between shared miRNA and target gene.

In alternative embodiments, according to sponge gene and target base in each sponge gene-target gene coexpression module Canonical correlation coefficient and shared miRNA between the quantity of cause, the significance value of shared miRNA, sponge gene and target gene When sensibility canonical correlation coefficient determine whether each sponge gene-target gene coexpression module is miRNA sponge module, packet It includes: if the quantity of sponge gene and target gene is all larger than equal to 2 in sponge gene-target gene coexpression module, sharing miRNA's For significance value less than 0.05, the canonical correlation coefficient between sponge gene and target gene is greater than 0.8, shares sensitivity when miRNA Property canonical correlation coefficient be greater than 0.1, it is determined that sponge gene-target gene coexpression module is miRNA sponge module.

In alternative embodiments, determining whether each sponge gene-target gene coexpression module is miRNA sponge After module, further includes: if sponge gene-target gene coexpression module is miRNA sponge module, obtain miRNA sponge mould The correlation data of block and target gene, wherein correlation data includes miRNA sponge module and default miRNA sponge module Saliency data, miRNA sponge module and miRNA sponge module whether be target gene marker, miRNA sponge module Including long-chain non-encoding ribonucleic acid LncRNA and protein coding ribonucleic acid mRNA.

In alternative embodiments, the saliency data of miRNA sponge module and default miRNA sponge module is obtained, It include: the first Pearson came phase for obtaining the average absolute value of all LncRNA-mRNA coexpressions pair in each miRNA sponge module Coefficient values (mean absolute Pearson correlation, Pearson).By presetting random algorithm, according to each All LncRNA and mRNA in miRNA sponge module generate multiple default miRNA sponge modules.Obtain each default sea miRNA Second Pearson correlation coefficient value of the average absolute value of all LncRNA-mRNA coexpressions pair in continuous module.Pass through pre- establishing MiRNA sponge is obtained according to the first Pearson correlation coefficient value and the second Pearson correlation coefficient value to difference test algorithm The saliency data of module and default miRNA sponge module.

In alternative embodiments, the saliency data of miRNA sponge module and target gene is obtained, comprising: obtain Preset data concentrates the quantity of LncRNA, mRNA and preset data to concentrate and LncRNA, mRNA of target gene coexpression Quantity.Obtain what the quantity and miRNA sponge module of LncRNA, mRNA and target gene in miRNA sponge module co-expressed The quantity of LncRNA, mRNA.The quantity of LncRNA, mRNA are concentrated according to preset data and preset data is concentrated and target gene In the quantity of LncRNA, mRNA of coexpression, miRNA sponge module the quantity and miRNA sponge module of LncRNA, mRNA with The quantity of LncRNA, mRNA of target gene coexpression calculate miRNA sponge module and mesh by hypergeometric distribution method of inspection Mark the saliency data of gene.

In alternative embodiments, determine miRNA sponge module whether be target gene marker, comprising: according to MiRNA calculates the value-at-risk of each matched sample by the first preset algorithm.The value-at-risk of each matched sample of binaryzation, is obtained The first risk value set and the second risk value set are obtained, wherein the value-at-risk in the first risk value set is greater than the second value-at-risk collection Value-at-risk in conjunction.The first risk is calculated by the second preset algorithm according to the first risk value set and the second risk value set The Hazard ratio of value set and the second risk value set.According to the first risk value set and the second risk value set, calculated according to examining Method obtains the significance of difference of the first risk value set and the second risk value set.It is determined according to Hazard ratio and the significance of difference MiRNA whether be target gene marker.

Second aspect, the embodiment of the present invention provide a kind of identification device of miRNA sponge module, comprising: obtain module, use In the expression matrix for the sponge gene for obtaining matched sample and the expression matrix of target gene.Module is obtained, is also used to according to sponge The expression matrix of gene and the expression matrix of target gene obtain multiple sponge genes-target gene coexpression module, wherein each Sponge gene-target gene coexpression module indicates a kind of gene that can be co-expressed with sponge gene and target gene.Module is obtained, It is also used to obtain in each sponge gene-target gene coexpression module, shares significance value, sponge gene and the target base of miRNA Because between canonical correlation coefficient and shared miRNA when sensibility canonical correlation coefficient.Determining module, for according to each The quantity of sponge gene and target gene, the significance value of shared miRNA, sponge base in sponge gene-target gene coexpression module Because determining each sponge base with sensibility canonical correlation coefficient when canonical correlation coefficient and shared miRNA between target gene Cause-target gene co-expresses whether module is miRNA sponge module.

In alternative embodiments, module is obtained, is specifically used for according to clustering algorithm, to the expression matrix of sponge gene It is clustered with the expression matrix of target gene, obtains multiple cluster results of sponge gene and target gene.By each cluster result In sponge gene and target gene, co-express module as a sponge gene-target gene.

In alternative embodiments, module is obtained, is specifically used for being led to according to preset miRNA- target gene regulation relationship It crosses hypergeometric distribution check algorithm to obtain in sponge gene-target gene coexpression module, be shared between sponge gene and target gene The significance value of miRNA.

In alternative embodiments, module is obtained, specifically for according to the expression matrix of sponge gene and target gene Expression matrix obtains the column vector of sponge gene and the column vector of target gene.Obtain the expression matrix and target base of sponge gene Variance matrix and covariance matrix between the expression matrix of cause.According to the column vector of sponge gene, target gene column to Amount, variance matrix, covariance matrix and preset representative vectors calculate canonical correlation coefficient.

In alternative embodiments, module is obtained, specifically for obtaining the typical case between shared miRNA and sponge gene Canonical correlation coefficient between related coefficient and shared miRNA and target gene.According to the typical case between sponge gene and target gene Canonical correlation coefficient, shared miRNA between related coefficient, shared miRNA and sponge gene and the typical phase between target gene Relationship number calculates the inclined canonical correlation coefficient between sponge gene and target gene.By the typical case between sponge gene and target gene Related coefficient subtracts the inclined canonical correlation coefficient between sponge gene and target gene, and sensibility when obtaining shared miRNA is typical Related coefficient.

In alternative embodiments, module is obtained, specifically for obtaining the expression matrix of shared miRNA.According to shared Expression matrix, the expression matrix of sponge gene and the expression matrix of target gene of miRNA obtains column vector, the sea of shared miRNA The column vector of continuous gene and the column vector of target gene.Obtain shared miRNA expression matrix and sponge gene expression matrix it Between variance matrix and covariance matrix.According to the column vector of shared miRNA, the column vector of sponge gene, shared miRNA Expression matrix and sponge gene expression matrix between variance matrix and covariance matrix and preset representative vectors, Calculate the canonical correlation coefficient between shared miRNA and sponge gene.Obtain the expression matrix of shared miRNA and the table of target gene Up to the variance matrix and covariance matrix between matrix.According to the column vector of shared miRNA, the column vector of target gene, share Variance matrix and covariance matrix between the expression matrix of miRNA and the expression matrix of target gene and it is preset it is typical to Amount calculates the canonical correlation coefficient between shared miRNA and target gene.

In alternative embodiments, determining module, if being specifically used for sponge in sponge gene-target gene coexpression module The quantity of gene and target gene is all larger than equal to 2, shares the significance value of miRNA less than 0.05, sponge gene and target gene it Between canonical correlation coefficient be greater than 0.8, share miRNA when sensibility canonical correlation coefficient be greater than 0.1, it is determined that sponge base It is miRNA sponge module that cause-target gene, which co-expresses module,.

In alternative embodiments, module is obtained, if being also used to sponge gene-target gene coexpression module is miRNA Sponge module then obtains the correlation data of miRNA sponge module and target gene, wherein correlation data includes the sea miRNA Whether the saliency data of continuous module and default miRNA sponge module, miRNA sponge module and miRNA sponge module are target The marker of gene, miRNA sponge module include long-chain non-encoding ribonucleic acid (Long non-coding RNA, lncRNA) And mRNA.

In alternative embodiments, module is obtained, is specifically used for obtaining in each miRNA sponge module and own The first Pearson came (Pearson) correlation coefficient value of the average absolute value of LncRNA-mRNA coexpression pair.Pass through default random calculation Method generates multiple default miRNA sponge modules according to all LncRNA and mRNA in each miRNA sponge module.It obtains each Second Pearson correlation coefficient of the average absolute value of all LncRNA-mRNA coexpressions pair in default miRNA sponge module Value.By default pairing difference test algorithm, according to the first Pearson correlation coefficient value and the second Pearson correlation coefficient value, Obtain the saliency data of miRNA sponge module and default miRNA sponge module.

In alternative embodiments, module is obtained, the quantity of LncRNA, mRNA are concentrated specifically for obtaining preset data And preset data concentrates the quantity with LncRNA, mRNA of target gene coexpression.It obtains in miRNA sponge module The quantity of LncRNA, mRNA of quantity and miRNA sponge module and the target gene coexpression of LncRNA, mRNA.According to pre- If the quantity and preset data of LncRNA, mRNA concentrate the number with LncRNA, mRNA of target gene coexpression in data set Amount, the LncRNA of the quantity and miRNA sponge module of LncRNA, mRNA and target gene coexpression in miRNA sponge module, The quantity of mRNA calculates the saliency data of miRNA sponge module and target gene by hypergeometric distribution method of inspection.

In alternative embodiments, module is obtained, is specifically used for, by the first preset algorithm, being calculated every according to miRNA The value-at-risk of a matched sample.The value-at-risk of each matched sample of binaryzation obtains the first risk value set and the second value-at-risk Set, wherein the value-at-risk in the first risk value set is greater than the value-at-risk in the second risk value set.According to the first value-at-risk Set and the second risk value set calculate the wind of the first risk value set and the second risk value set by the second preset algorithm Dangerous ratio.The first risk value set and second is obtained according to check algorithm according to the first risk value set and the second risk value set The significance of difference of risk value set.According to Hazard ratio and the significance of difference determine miRNA whether be target gene marker.

The third aspect, the embodiment of the present invention provide a kind of identification equipment of miRNA sponge module, comprising: processor, storage Medium and bus, storage medium is stored with the executable machine readable instructions of processor, when the identification equipment of miRNA sponge module When operation, by bus communication between processor and storage medium, processor executes machine readable instructions, to execute above-mentioned first The step of aspect either method.

Fourth aspect, the embodiment of the present invention also provide a kind of computer readable storage medium, computer readable storage medium On the step of being stored with computer program, executing when computer program is run by processor such as above-mentioned first aspect either method.

In the present invention, the conspicuousness of shared miRNA in module is co-expressed by obtaining multiple sponge genes-target gene Sensibility canonical correlation coefficient when canonical correlation coefficient and shared miRNA between value, sponge gene and target gene, and root According to the quantity of sponge gene and target gene in each sponge gene-target gene coexpression module, the significance value of shared miRNA, Sensibility canonical correlation coefficient when canonical correlation coefficient and shared miRNA between sponge gene and target gene determines each Whether sponge gene-target gene coexpression module is miRNA sponge module.Can under module level measure sponge gene with Competition intensity between target gene, and according to Dinghai silk floss gene-target gene really of competition intensity between sponge gene and target gene Co-express whether module is miRNA sponge module, so that the identification of miRNA sponge module is more accurate.

Detailed description of the invention

It, below will be to specific in order to illustrate more clearly of the specific embodiment of the invention or technical solution in the prior art Embodiment or attached drawing needed to be used in the description of the prior art be briefly described, it should be apparent that, it is described below Attached drawing is some embodiments of the present invention, for those of ordinary skill in the art, before not making the creative labor It puts, is also possible to obtain other drawings based on these drawings.

Fig. 1 is the recognition methods flow diagram for the miRNA sponge module that one embodiment of the invention provides；

Fig. 2 be another embodiment of the present invention provides miRNA sponge module recognition methods flow diagram；

Fig. 3 be another embodiment of the present invention provides miRNA sponge module recognition methods flow diagram；

Fig. 4 be another embodiment of the present invention provides miRNA sponge module recognition methods flow diagram；

Fig. 5 be another embodiment of the present invention provides miRNA sponge module recognition methods flow diagram；

Fig. 6 be another embodiment of the present invention provides miRNA sponge module recognition methods flow diagram；

Fig. 7 be another embodiment of the present invention provides miRNA sponge module recognition methods flow diagram；

Fig. 8 be another embodiment of the present invention provides miRNA sponge module recognition methods flow diagram；

Fig. 9 be another embodiment of the present invention provides miRNA sponge module recognition methods flow diagram；

Figure 10 is in the embodiment of the present invention, and the miRNA sponge module of scene one is horizontal with the coexpression of corresponding randomized blocks Comparison schematic diagram；

Figure 11 is in the embodiment of the present invention, and the miRNA sponge module of scene two is horizontal with the coexpression of corresponding randomized blocks Comparison schematic diagram；

Figure 12 is the identification device structural schematic diagram for the miRNA sponge module that one embodiment of the invention provides；

Figure 13 is the identification device structure schematic diagram for the miRNA sponge module that one embodiment of the invention provides.

Specific embodiment

In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is A part of the embodiment of the present invention, instead of all the embodiments.The present invention being usually described and illustrated herein in the accompanying drawings is implemented The component of example can be arranged and be designed with a variety of different configurations.

Therefore, the detailed description of the embodiment of the present invention provided in the accompanying drawings is not intended to limit below claimed The scope of the present invention, but be merely representative of selected embodiment of the invention.Based on the embodiments of the present invention, this field is common Technical staff's every other embodiment obtained without creative efforts belongs to the model that the present invention protects It encloses.

It should also be noted that similar label and letter indicate similar terms in following attached drawing and formula, therefore, once it is a certain It is defined in Xiang Yi attached drawing or formula, does not then need that it is carried out further to define reconciliation in subsequent attached drawing or formula It releases.

With reference to the accompanying drawing, it elaborates to some embodiments of the present invention.In the absence of conflict, following Feature in embodiment and embodiment can be combined with each other.

Fig. 1 is the recognition methods flow diagram for the miRNA sponge module that one embodiment of the invention provides.The sea the miRNA The executing subject of the recognition methods of continuous module can be the terminal device with computing capability, such as: desktop computer, notebook electricity Brain, server, cloud, custom terminal or intelligent terminal etc., herein with no restrictions.

As shown in Figure 1, the recognition methods of miRNA sponge module, comprising:

S110, obtain matched sample sponge gene expression matrix and target gene expression matrix.

It, can be according to the sponge gene expression profile data of matched sample and the gene expression of target gene in some embodiments Modal data obtains the gene of sponge gene expression profile data and target gene since gene expression data generallys use matrix form Expression modal data can be obtained the expression matrix of sponge gene and the expression matrix of target gene.

In some embodiments, gene expression profile data is obtained, genetic chip can be first prepared, e.g., in sheet glass, polypropylene Or genetic chip etc. is formed by the fixed multiple gene probes of preset arrangement mode on the carriers such as nylon membrane.Then gene is visited Needle purified, reverse transcription or fluorescent marker, obtains label probe, wherein fluorescent marker is usually using Cye3-dUTP (Cy3) It is marked with Cye5-dUTP (Cy5).Subsequently, for carrying out fluorescent marker, can according to the condition that sets by label probe with Chip hybridization after a certain period of time, washes away unbonded probe, carries out the scanning and analysis of fluorescence signal, obtains hybridization image.Most Afterwards, gene expression profile data is extracted from hybridization image, can be obtained gene expression profile data.

Wherein, the target that sponge gene and target gene are miRNA, the type of RNA can be other coding RNAs, false base Because of (Pseudogene), cyclic annular ribonucleic acid (CircRNA), one of LncRNA group and mRNA group.

S120, according to the expression matrix of sponge gene and the expression matrix of target gene, obtain multiple sponge gene-target genes Co-express module.

Wherein, each sponge gene-target gene coexpression module indicates that one kind can use sponge gene and the common table of target gene The gene reached.

In some embodiments, the expression matrix D of sponge gene₁It may is that

The expression matrix D of target gene₂It may is that

Wherein, G indicates that the gene in expression matrix, R indicate that matched sample, S are the number of matched sample, n₁Indicate each The number of sponge gene, n in matched sample₂Indicate the number of target gene in each matched sample.

S130, it obtains in each sponge gene-target gene coexpression module, shares significance value, the sponge gene of miRNA Sensibility canonical correlation coefficient when canonical correlation coefficient and shared miRNA between target gene.

It should be noted that canonical correlation coefficient between the significance value of shared miRNA, sponge gene and target gene with And sensibility canonical correlation coefficient when shared miRNA is the important parameter in miRNA sponge module competition mechanism.

Wherein, miRNA express spectra data sample has to and sponge gene and expression of target gene modal data sample matches, table It is shown asWherein S is the number of matched sample, n₃Indicate each matched sample The number of middle miRNA.

In some embodiments, in miRNA sponge module competition mechanism, there are the RNA groups of five seed types: other codings RNA group, Pseudogene group, CircRNA group, LncRNA group and mRNA group, typical type of competition include: other coding RNAs One of group, Pseudogene group, CircRNA group or LncRNA group, vie each other with mRNA group.

In some embodiments, if other coding RNA groups, Pseudogene group, CircRNA group or LncRNA group competition victory Benefit, then mRNA group will all translate into protein, if the competition triumph of mRNA group, mRNA group is by whole degradations.

Optionally, there is also other competitive modes, for example, other coding RNA groups and Pseudogene group, other codings RNA group and CircRNA group, other coding RNA groups and LncRNA group, Pseudogene group and CircRNA group, Pseudogene group With LncRNA group and CircRNA group and LncRNA group etc., herein with no restrictions.

S140, the quantity of sponge gene and target gene in module is co-expressed according to each sponge gene-target gene, is shared Sensibility when canonical correlation coefficient and shared miRNA between the significance value of miRNA, sponge gene and target gene is typical Related coefficient determines whether each sponge gene-target gene coexpression module is miRNA sponge module.

It should be noted that can determine sponge gene-target gene coexpression mould according to miRNA sponge module competition mechanism When target gene in block is competed successfully, the canonical correlation system between significance value, sponge gene and the target gene of miRNA is shared The design parameter of sensibility canonical correlation coefficient when several and shared miRNA, to determine the sponge gene-according to the parameter Target gene co-expresses whether module is miRNA sponge module.

In the present embodiment, the conspicuousness of shared miRNA in module is co-expressed by obtaining multiple sponge genes-target gene Sensibility canonical correlation coefficient when canonical correlation coefficient and shared miRNA between value, sponge gene and target gene, and root According to the quantity of sponge gene and target gene in each sponge gene-target gene coexpression module, the significance value of shared miRNA, Sensibility canonical correlation coefficient when canonical correlation coefficient and shared miRNA between sponge gene and target gene determines each Whether sponge gene-target gene coexpression module is miRNA sponge module.Can under module level measure sponge gene with Competition intensity between target gene, and according to Dinghai silk floss gene-target gene really of competition intensity between sponge gene and target gene Co-express whether module is miRNA sponge module, so that the identification of miRNA sponge module is more accurate.

Fig. 2 be another embodiment of the present invention provides miRNA sponge module recognition methods flow diagram.

In alternative embodiments, as shown in Fig. 2, according to the expression square of the expression matrix of sponge gene and target gene Battle array obtains multiple sponge genes-target gene coexpression module, comprising:

S121, according to clustering algorithm, the expression matrix of expression matrix and target gene to sponge gene clusters, obtain Multiple cluster results of sponge gene and target gene.

In some embodiments, by cluster, sponge gene and the target gene integration that can will be co-expressed, to obtain Obtain multiple cluster results.

In some embodiments, unidirectional clustering algorithm includes: weight gene co-expressing Network Analysis Method (weighted Gene co-expression network analysis, WGCNA), K mean cluster method (K-means), hierarchical clustering method, base In application space clustering procedure (the Density-Based Spatial Clustering of Applications of density noise With Noise, DBSCAN), one of fuzzy C-means clustering method (Fuzzy C-Means, FCM), but not limited to this.With For WGCNA, unidirectional cluster is illustrated, for sponge gene and target gene that above-mentioned expression matrix determines, each pair of gene i With the gene co-expressing similarity s of gene j_ijAre as follows:

s_ij=| cor (i, j) |

Wherein, | cor (i, j) | it is the Pearson correlation coefficient absolute value of gene i and gene j.

Gene co-expressing similar matrix can be defined as S=[sij], select soft-threshold using scale-free topology standard, and Adjacency matrix A is converted by the similar matrix according to soft-threshold, in general, minimum scale-free topology fit indices R²It is usually not small In 0.8.Topology overlapping matrix (topological overlap matrix, TOM) W=is generated based on adjacency matrix A, WGCNA [wi_j].The then TOM similar value w of gene i and gene j_ijAre as follows:

Wherein, wherein u represents all genes in matched sample in sponge gene expression matrix and expression of target gene matrix, The TOM non-similarity value of gene i and j are di_j=1-wi_j.In order to identify sponge gene-target gene coexpression module, WGCNA is adopted With hierarchy clustering method (Hierarchical Clustering, HC) to non-similarity matrix D=[di of TOM_j] gathered Class.The sponge gene identified-target gene coexpression module has high topological plyability.

Bidirectional clustering algorithm include: sparse group factor analytic approach (Sparse Group Factor Analysis, SGFA), Double focusing class factor analysis (Factor Analysis for Bicluster Acquisition) FABIA, double focusing class mass spectrum point Double clustering procedures of analysis method (Bicluster Clustering of Spectral, BCSpectral), grid pattern etc. One of (Bicluster Clustering of Plaid, BCSpectral), but not limited to this.It is right by taking SGFA as an example Bidirectional clustering is illustrated, for sponge gene and target gene that above-mentioned expression matrix determines, if to excavate the number of bidirectional clustering Amount is B, need to calculate each gene (sponge gene and target gene) and each matched sample belong to each bidirectional clustering be subordinate to or The degree of association.Wherein, n-th of gene and k-th of bidirectional clustering degree of association g_{N, k}It is closed with d-th of matched sample and k-th of bidirectional clustering Connection degreeCalculation it is as follows:

Wherein:

M represents input data set label, since SGFA input data is two kinds of numbers of sponge gene and expression of target gene matrix According to collection, therefore m maximum value is 2.Two-valued variableRepresent k-th of bidirectional clustering whether include n-th of gene (be included as 1, Not comprising for 0),Represent the scale factor of k-th of bidirectional clustering.Two-valued variableWhether represent k-th of bidirectional clustering Comprising expressed from m-th d-th of sample in modal data (be included as 1, do not include for 0),Represent m-th of express spectra The scale factor of k-th of bidirectional clustering in data.ForWhen probability, hyper parameter a^π, b^π, a^α, b^αWith it is initial Change rate parameter δ₀Initial value be defaulted as 1.Degree of association g_{N, k}WithValue range be [- 1,1], pass through degree of association absolute value (absolute value of association, AVA) determines being associated with for each gene and matched sample and each bidirectional clustering Intensity, AVA threshold value are usually not less than 0.8.

S122, by the sponge gene and target gene in each cluster result, as a sponge gene-target gene coexpression Module.

Such as the clustering method in S121, can get multiple cluster results, each cluster result include a sponge gene and One target gene co-expresses module using the sponge gene and the target gene as a sponge gene-target gene.

In the present embodiment, by unidirectionally clustering or bidirectional clustering method clusters sponge gene and target gene, with It obtains sponge gene-target gene and co-expresses module, so that the sponge gene obtained-target gene coexpression module is more accurate, from And when determining miRNA sponge module, obtained miRNA sponge module is also more accurate.

In alternative embodiments, the significance value of shared miRNA is obtained, comprising: according to preset miRNA- target base Because of regulation relationship, sponge gene-target gene is obtained by hypergeometric distribution check algorithm and is co-expressed in module, sponge gene and target The significance value of miRNA is shared between gene.

In some embodiments, preset miRNA- target gene regulation relationship is by merging a variety of confirmatory numbers of different experiments It is obtained according to library mode.Preset miRNA- target gene regulation relationship includes: miRNA-lncRNA regulation relationship data and miRNA- MRNA regulation relationship data.Wherein, miRNA-lncRNA regulation relationship data can be by integrating NPInter v3.0 and LncBase Two kinds of databases of v2.0 experiment module obtain, and miRNA-mRNA regulation relationship data can by integrate miRTarBase v7.0, Tri- kinds of databases of TarBase v7.0 and miRWalk v2.0 obtain, and but not limited to this.

In some embodiments, the significance value p of miRNA is shared between sponge gene and target gene, can pass through following public affairs Formula is calculated:

Wherein, N₁Represent the quantity of miRNA all in data set, M₁And K₁Respectively represent regulation sponge gene and target base The quantity of the miRNA of cause, L₁Indicate the miRNA number that sponge gene and target gene are shared, L₁Value, which is typically larger than, is equal to 3.

In the present embodiment, preset miRNA- target base is obtained by merging the confirmatory database mode of a variety of different experiments Because of regulation relationship, and is obtained according to the regulation relationship by hypergeometric distribution checking computation and shared between sponge gene and target gene The significance value of miRNA so that the accuracy of obtained significance value is higher, and will acquire between sponge gene and target gene altogether The significance value of miRNA is enjoyed for determining miRNA sponge module, so that the miRNA sponge module determined is more accurate.

Fig. 3 be another embodiment of the present invention provides miRNA sponge module recognition methods flow diagram.

In alternative embodiments, as shown in figure 3, obtaining the canonical correlation coefficient between sponge gene and target gene, Include:

S131, according to the expression matrix of sponge gene and the expression matrix of target gene, obtain the column vector of sponge gene with And the column vector of target gene.

In some embodiments, in sponge gene-target gene coexpression module, the column vector and target gene of sponge gene Column vector be sponge gene expression matrix and target gene expression matrix in, the element set of each column, for example, sponge base Column vector can be X=(x in the expression matrix of cause₁, x₂..., x_p)^T, column vector can be Y=in expression of target gene matrix (y₁, y₂..., y_q)^T, but not limited to this.

Variance matrix and association side between S132, the expression matrix for obtaining sponge gene and the expression matrix of target gene Poor matrix.

In some embodiments, ∑ can be used_XXAnd ∑_YYRespectively represent the variance square being calculated from matrix X and Y Battle array, ∑_XYRepresent the covariance matrix between X and Y matrix.

S133, according to the column vector of sponge gene, the column vector of target gene, variance matrix, covariance matrix and preset Representative vectors calculate canonical correlation coefficient.

In a kind of possible implementation, the canonical correlation (Canonical between sponge gene and target gene Correlation, CC) coefficient, it can be calculated by the following formula:

Wherein, RNA₁Indicate sponge gene, RNA₂Indicate target gene, a a ∈ R^pWith b b ∈ R^qTo maximize canonical correlation Coefficient (corr (a^TX, b^TY)) the preset representative vectors of numerical value.

In the present embodiment, by the expression matrix of sponge gene and the expression matrix of target gene, sponge gene is obtained Column vector and the column vector of target gene, variance matrix and covariance matrix, and according to according to the column vector of sponge gene, Column vector, variance matrix, covariance matrix and the preset representative vectors of target gene calculate canonical correlation coefficient, by typical phase Relationship number is for determining miRNA sponge module, so that the miRNA sponge module determined is more accurate.

Fig. 4 be another embodiment of the present invention provides miRNA sponge module recognition methods flow diagram.

In alternative embodiments, as shown in figure 4, obtaining sensibility canonical correlation coefficient when sharing miRNA, packet It includes:

S134, it obtains between the canonical correlation coefficient and shared miRNA and target gene shared between miRNA and sponge gene Canonical correlation coefficient.

In some embodiments, obtain canonical correlation coefficient between shared miRNA and sponge gene and shared miRNA with The mode of canonical correlation coefficient between target gene is identical as the mode in S131-S133, and details are not described herein.

Between S135, the canonical correlation coefficient according between sponge gene and target gene, shared miRNA and sponge gene Canonical correlation coefficient between canonical correlation coefficient, shared miRNA and target gene, calculates inclined between sponge gene and target gene Canonical correlation coefficient.

In some embodiments,Indicate the inclined canonical correlation system between sponge gene and target gene Number, i.e., the canonical correlation coefficient under the precondition for considering shared miRNAs, between sponge gene and target gene.Can calculate according to the following formula:

Wherein,The canonical correlation coefficient between shared miRNA group and sponge gene is represented,Represent the canonical correlation coefficient between shared miRNAs group and target gene.

S136, the canonical correlation coefficient between sponge gene and target gene is subtracted it is inclined between sponge gene and target gene Canonical correlation coefficient obtains sensibility canonical correlation coefficient when shared miRNA.

In some embodiments, sensibility between sponge gene and target gene in sponge gene-target gene coexpression module Canonical correlation (sensitivity canonical correlation, SCC) coefficientIt is defined as follows:

In the present embodiment, modal data is expressed by incorporating miRNA, under the precondition for considering shared miRNAs, meter Sensibility canonical correlation coefficient when obtaining shared miRNA is calculated, and for determining miRNA sponge module, so that the miRNA determined Sponge module is more accurate.

Fig. 5 be another embodiment of the present invention provides miRNA sponge module recognition methods flow diagram.

In alternative embodiments, as shown in figure 5, obtaining the canonical correlation system between shared miRNA and sponge gene Canonical correlation coefficient between several and shared miRNA and target gene, comprising:

S1341, the expression matrix for obtaining shared miRNA.

S1342, basis share expression matrix, the expression matrix of sponge gene and the expression matrix of target gene of miRNA, obtain Take the column vector, the column vector of sponge gene and the column vector of target gene of shared miRNA.

Variance matrix between S1343, the expression matrix for obtaining shared miRNA and the expression matrix of sponge gene and Covariance matrix.

S1344, basis share column vector, the column vector of sponge gene, the expression matrix of shared miRNA and the sea of miRNA Variance matrix and covariance matrix and preset representative vectors between the expression matrix of continuous gene, calculate shared miRNA with Canonical correlation coefficient between sponge gene.

Variance matrix and association between S1345, the expression matrix for obtaining shared miRNA and the expression matrix of target gene Variance matrix.

S1346, column vector, the column vector of target gene, the expression matrix and target base for sharing miRNA according to shared miRNA Variance matrix and covariance matrix and preset representative vectors between the expression matrix of cause calculate shared miRNA and target base Canonical correlation coefficient because between.

Wherein, it obtains between the canonical correlation coefficient and shared miRNA and target gene between shared miRNA and sponge gene Canonical correlation coefficient step S1341-S1345 in S131-S133 calculate the method for canonical correlation coefficient it is identical, herein not It repeats again.

Fig. 6 be another embodiment of the present invention provides miRNA sponge module recognition methods flow diagram.

In alternative embodiments, as shown in fig. 6, whether determining each sponge gene-target gene coexpression module After miRNA sponge module, further includes:

If S150, sponge gene-target gene coexpression module are miRNA sponge module, obtain miRNA sponge module with The correlation data of target gene.

Wherein, correlation data includes the saliency data of miRNA sponge module and default miRNA sponge module, miRNA Sponge module and miRNA sponge module whether be target gene marker, miRNA sponge module includes long-chain non-coding ribose Nucleic acid LncRNA and protein coding ribonucleic acid mRNA.

In some embodiments, after obtaining miRNA sponge module, it can also be verified, be obtained with determining MiRNA sponge module it is whether related to target gene, for example, can verify miRNA sponge module whether with breast cancer cause a disease Gene-correlation, that is, obtain the correlation data of miRNA sponge module and breast cancer Disease-causing gene.

Fig. 7 be another embodiment of the present invention provides miRNA sponge module recognition methods flow diagram.

In alternative embodiments, as shown in fig. 7, obtaining the aobvious of miRNA sponge module and default miRNA sponge module Work property data, comprising:

S151, obtain the average absolute value of all LncRNA-mRNA coexpressions pair in each miRNA sponge module first Pearson correlation coefficient value.

In some embodiments, the value range of the first Pearson correlation coefficient value is [0,1], and the value is bigger, then it represents that Coexpression level between lncRNA and mRNA is higher.

S152, it is generated more by default random algorithm according to all LncRNA and mRNA in each miRNA sponge module A default miRNA sponge module.

In some embodiments, it is default random alignment can be carried out to lncRNA in each miRNA sponge module and mRNA Number, such as 1000 times, then can be randomly generated it is identical as each miRNA sponge module gene number (i.e. lncRNA and mRNA's Number is identical) random miRNA sponge module, totally 1000, as default miRNA sponge module.

S153, the average absolute value of all LncRNA-mRNA coexpressions pair in each default miRNA sponge module is obtained Second Pearson correlation coefficient value.

Wherein, with reference to S152, being averaged between lncRNA and mRNA in 1000 default miRNA sponge modules can be calculated The coexpression that horizontal (i.e. the second Pearson correlation coefficient value) is co-expressed as each miRNA sponge module defaultization is horizontal.

S154, algorithm is matched by default, according to the first Pearson correlation coefficient value and the second Pearson correlation coefficient Value obtains the saliency data of miRNA sponge module and default miRNA sponge module.

In some embodiments, it can be compared by Welch paired t-test (Welch's two sample t-test) The horizontal significance of difference is co-expressed between miRNA sponge module and default miRNA sponge module.Welch paired t-test formula It is specific as follows:

Wherein, whereinWithRespectively represent miRNA sponge module and default miRNA sponge module is averagely total to table Up to level,WithRespectively represent the variance of miRNA sponge module and default miRNA sponge module coexpression level, N₁And N₂ It is identical and represent miRNA sponge number of modules.

It should be noted that the t value calculated is bigger, significance of difference p value is smaller, indicates the total table of miRNA sponge module The coexpression for being significantly higher than default miRNA sponge module up to level is horizontal, and conspicuousness p value is less than 0.05 in the present embodiment.

Fig. 8 be another embodiment of the present invention provides miRNA sponge module recognition methods flow diagram.

In alternative embodiments, as shown in figure 8, obtaining the saliency data of miRNA sponge module and target gene, Include:

S155, the quantity for obtaining preset data concentration LncRNA, mRNA and preset data are concentrated and are total to table with target gene The quantity of LncRNA, mRNA for reaching.

Wherein, it includes in multiple LncRNA and mRNA and multiple LncRNA and mRNA, with target base that preset data, which is concentrated, Because of relevant LncRNA and mRNA, preset data integrates the data set that can be obtained as priori.

S156, the quantity and miRNA sponge module and target gene for obtaining LncRNA, mRNA in miRNA sponge module The quantity of LncRNA, mRNA of coexpression.

Wherein, i.e. sponge gene-target of the quantity of LncRNA, mRNA of miRNA sponge module and target gene coexpression Gene co-expressing module is the quantity of miRNA sponge module.

S157, the quantity that LncRNA, mRNA are concentrated according to preset data and preset data concentration are total to table with target gene The quantity and miRNA sponge module and target of LncRNA, mRNA in the quantity of LncRNA, mRNA for reaching, miRNA sponge module The quantity of LncRNA, mRNA of gene co-expressing calculate miRNA sponge module and target base by hypergeometric distribution method of inspection The saliency data of cause.

In some embodiments, by taking target gene relevant to breast cancer as an example, it is illustrated, calculation are as follows:

Wherein, N₂Represent the number of gene in data set (lncRNA and mRNA), M₂Represent mastocarcinoma gene in data set The number of (lncRNA and mRNA), K₂Indicate the number of gene (lncRNA and mRNA) in miRNA sponge module, L₂Indicate miRNA The number of mastocarcinoma gene (lncRNA and mRNA) in sponge module.

Fig. 9 be another embodiment of the present invention provides miRNA sponge module recognition methods flow diagram.

In alternative embodiments, as shown in figure 9, determine miRNA sponge module whether be target gene marker, Include:

S158, the value-at-risk of each matched sample is calculated by the first preset algorithm according to miRNA.

In some embodiments, by taking target gene relevant to breast cancer as an example, it is illustrated.

Wherein, the value-at-risk of each matched sample can be calculate by the following formula:

H (t, Z)=h₀(t) exp (β Z)=h₀(t)exp(β₁Z₁+β₂Z₂+...+β_kZ_k)

Wherein, h (t, Z) is the risk function value for having the breast cancer sample of covariant Z in moment t, t for survival when Between, Z=(Z₁, Z₂..., Z_k) ' be possible influence the gene (LncRNA and mRNA) of life span, h₀It (t) is all covariants Risk function value β=(β when taking 0₁, β₂..., β_kThe regression coefficient of) ' be Cox model.

The value-at-risk of each matched sample of S159, binaryzation obtains the first risk value set and the second risk value set, In value-at-risk in the first risk value set be greater than the value-at-risk in the second risk value set.

In some embodiments, according to the risk function value h (t, Z) of each sample, 500 breast cancer samples are divided into Two sample sets of high risk and low-risk, i.e. the first risk value set (high risk) and the second risk value set (low-risk), Wherein, the mode of binaryzation can be with are as follows: the risk function value of each sample is subjected to descending sort, according to preset percentage into Row divides, and determines the first risk value set and the second risk value set, for example, can be used as the first value-at-risk collection for preceding 50% It closes, rear 50% is used as the second risk value set；Or the first risk value set of preceding 20% conduct, rear 80% is used as the second value-at-risk collection It closes, but not limited to this.It should be noted that the set of the first value-at-risk should at least account for preceding the 50% of all risk function values.

S1510, the first wind is calculated by the second preset algorithm according to the first risk value set and the second risk value set The Hazard ratio of dangerous value set and the second risk value set.

In some embodiments, the calculation of Hazard ratio can be with are as follows:

HR=h (t, Z_h)/h (t, Z_l)=exp [β (Z_h-Z_l)]

Wherein, h (t, Z_h) be breast cancer high risk group risk function value, h (t, Z_l) be breast cancer low-risk group risk Functional value,It is the high risk gene (LncRNA and mRNA) that possible influence life span,It is the low-risk gene (LncRNA and mRNA) that possible influence life span.

In a kind of possible embodiment, the threshold value of HR can be set as 2.

S1511, the first value-at-risk is obtained according to check algorithm according to the first risk value set and the second risk value set The significance of difference of set and the second risk value set.

In some embodiments, can be compared according to Log-Rank Test (Log-rank test) breast cancer high risk and Whether two groups of sample life spans of low-risk are identical, and test statistics is card side χ², it calculates as follows:

Wherein, A is observation breast cancer deaths case number of cases, and T is theoretical breast cancer deaths case number of cases.The χ of calculating²Value is bigger, Significance of difference p value is smaller, and it is not identical to indicate that two groups of sample life spans of breast cancer high risk and low-risk are got over.

S1512, determined according to Hazard ratio and the significance of difference miRNA whether be target gene marker.

Wherein, when HR value is greater than 2 and Log-Rank Test conspicuousness p value is less than 0.05, miRNA sponge module and is just recognized It is set to breast cancer module biomarker, but not limited to this.

Here, passing through answering for recognition methods of the following two application scenarios to miRNA sponge module provided herein With explaining, those skilled in the art can be defined, and following example is only for example, and have to carry out in this way without representing.

Scene one

Firstly, heterogeneous data source is obtained, from cancer gene express spectra database TCGA (the cancer genome Atlas, https: //cancergenome.nih.gov/) in, collect miRNA, lncRNA and the mRNA of breast cancer matched sample Express the Survival data information of modal data and breast cancer sample.Pass through pretreatment (removal duplicate keys and not Gene Name MiRNA, lncRNA and mRNA), finally obtain 674 miRNA of 500 breast cancer matched samples, 12711 lncRNA and 18344 mRNA express modal data.In this scene, RNA₁For LncRNA, RNA₂For mRNA.Therefore:

D₁={ G_1,1；G_1,2；...；G_1,500}∈R^500×1271

D₂={ G_2,1；G_2,2；...；G_2,500)∈R^500×18344

D₃={ G_3,1；G_3,2；...；G_3,500}∈R^500×674

Then, identification lncRNA-mRNA co-expresses module, and the lncRNA and mRNA for giving matched sample express modal data, Module is co-expressed using WGCNA coexpression network analysis method identification lncRNA-mRNA.Wherein, the minimum nothing in WGCNA method Scale topology fit indices R²It is set as 0.8.

Module is co-expressed based on lncRNA-mRNA, by calculating shared miRNA conspicuousness p value, canonical correlation coefficient and quick Three Measure Indexes of perceptual canonical correlation coefficient identify miRNA sponge module.Each miRNA sponge module lncRNA and mRNA Number be no less than 2 respectively, and must satisfy condition: shared miRNAs conspicuousness p value < 0.05, canonical correlation coefficientSensibility canonical correlation coefficient

Finally, assessment miRNA sponge module, determines in miRNA sponge module for breast cancer module biomarker Number.

Figure 10 is in the embodiment of the present invention, and the miRNA sponge module of scene one is horizontal with the coexpression of corresponding randomized blocks Comparison schematic diagram.

In scene one, 17 miRNA sponge modules are identified in total, as shown in table 1.As shown in Figure 10,17 excavated What miRNA sponge module co-expressed horizontal conspicuousness is higher than horizontal (the conspicuousness p value=1.55E- of corresponding randomized blocks coexpression 05).As shown in table 2 and table 3, in 17 miRNA sponge modules, there is that 10 miRNA sponge modules are related to breast cancer enrichment, There are 15 miRNA sponge modules that can serve as breast cancer module biomarker.

The miRNA sponge module excavated in 1 scene one of table

MiRNA sponge module relevant to breast cancer enrichment in 2 scene one of table

The breast cancer miRNA sponge module of biomarker is served as in 3 scene one of table

Scene two

In scene two, the lncRNA and mRNA for giving matched sample express modal data, are analyzed using the sparse group factor of SGFA Method identifies that lncRNA-mRNA co-expresses module.Wherein, (the absolute value of of the degree of association absolute value in SGFA method Association, AVA) threshold value is set as 0.8.Other steps are identical as scene one.

Figure 11 is in the embodiment of the present invention, and the miRNA sponge module of scene two is horizontal with the coexpression of corresponding randomized blocks Comparison schematic diagram.

In scene two, 51 miRNA sponge modules are identified in total, as shown in table 4.As shown in figure 11,51 excavated What miRNA sponge module co-expressed horizontal conspicuousness is higher than horizontal (the conspicuousness p value=1.55E- of corresponding randomized blocks coexpression 14).As shown in table 5 and table 6, in 51 miRNA sponge modules, there is that 3 miRNA sponge modules are related to breast cancer enrichment, There are 49 miRNA sponge modules that can serve as breast cancer module biomarker.

The miRNA sponge module excavated in 4 scene two of table

MiRNA sponge module relevant to breast cancer enrichment in 5 scene two of table

The breast cancer miRNA sponge module of biomarker is served as in 6 scene two of table

Figure 12 is the identification device structural schematic diagram for the miRNA sponge module that one embodiment of the invention provides

As shown in figure 12, the identification device of miRNA sponge module, comprising:

Module 210 is obtained, for obtaining the expression matrix of the sponge gene of matched sample and the expression matrix of target gene.It obtains Modulus block 210 is also used to the expression matrix of the expression matrix and target gene according to sponge gene, obtains multiple sponge gene-targets Gene co-expressing module, wherein each sponge gene-target gene coexpression module indicates that one kind can use sponge gene and target gene The gene of co-expression.Module 210 is obtained, is also used to obtain in each sponge gene-target gene coexpression module, share Sensibility when canonical correlation coefficient and shared miRNA between the significance value of miRNA, sponge gene and target gene is typical Related coefficient.Determining module 220, for according to sponge gene and target gene in each sponge gene-target gene coexpression module Quantity, the significance value of shared miRNA, canonical correlation coefficient and shared miRNA between sponge gene and target gene when Sensibility canonical correlation coefficient determine whether each sponge gene-target gene coexpression module is miRNA sponge module.

In alternative embodiments, module 210 is obtained, the expression according to clustering algorithm, to sponge gene is specifically used for The expression matrix of matrix and target gene is clustered, and multiple cluster results of sponge gene and target gene are obtained.By each cluster As a result sponge gene and target gene in co-express module as a sponge gene-target gene.

In alternative embodiments, module 210 is obtained, is specifically used for regulating and controlling to close according to preset miRNA- target gene System obtains sponge gene-target gene by hypergeometric distribution method of inspection and co-expresses in module, between sponge gene and target gene altogether Enjoy the significance value of miRNA.

In alternative embodiments, module 210 is obtained, specifically for the expression matrix and target gene according to sponge gene Expression matrix, obtain sponge gene column vector and target gene column vector.Obtain the expression matrix and target of sponge gene Variance matrix and covariance matrix between the expression matrix of gene.According to the column vector of sponge gene, target gene column to Amount, variance matrix, covariance matrix and preset representative vectors calculate canonical correlation coefficient.

In alternative embodiments, module 210 is obtained, specifically for obtaining between shared miRNA and sponge gene Canonical correlation coefficient between canonical correlation coefficient and shared miRNA and target gene.According between sponge gene and target gene Canonical correlation coefficient, shared miRNA between canonical correlation coefficient, shared miRNA and sponge gene and the allusion quotation between target gene Type related coefficient calculates the inclined canonical correlation coefficient between sponge gene and target gene.It will be between sponge gene and target gene Canonical correlation coefficient subtracts the inclined canonical correlation coefficient between sponge gene and target gene, obtains sensibility when shared miRNA Canonical correlation coefficient.

In alternative embodiments, module 210 is obtained, specifically for obtaining the expression matrix of shared miRNA.According to altogether Enjoy the expression matrix, the expression matrix of sponge gene and the expression matrix of target gene of miRNA, obtain shared miRNA column vector, The column vector of sponge gene and the column vector of target gene.Obtain the expression matrix of shared miRNA and the expression matrix of sponge gene Between variance matrix and covariance matrix.According to the column vector of shared miRNA, the column vector of sponge gene, share Variance matrix and covariance matrix and preset typical case between the expression matrix of miRNA and the expression matrix of sponge gene Vector calculates the canonical correlation coefficient between shared miRNA and sponge gene.Obtain the expression matrix and target base of shared miRNA Variance matrix and covariance matrix between the expression matrix of cause.According to the column vector of shared miRNA, target gene column to Variance matrix and covariance matrix between amount, the expression matrix of shared miRNA and the expression matrix of target gene and preset Representative vectors calculate the canonical correlation coefficient between shared miRNA and target gene.

In alternative embodiments, determining module 220, if being specifically used in sponge gene-target gene coexpression module The quantity of sponge gene and target gene is all larger than equal to 2, and the significance value of shared miRNA is less than 0.05, sponge gene and target base Canonical correlation coefficient because between is greater than 0.8, and sensibility canonical correlation coefficient when sharing miRNA is greater than 0.1, it is determined that sponge Gene-target gene coexpression module is miRNA sponge module.

In alternative embodiments, module 210 is obtained, is if being also used to sponge gene-target gene coexpression module MiRNA sponge module, then obtain the correlation data of miRNA sponge module and target gene, wherein correlation data includes Whether the saliency data of miRNA sponge module and default miRNA sponge module, miRNA sponge module and miRNA sponge module For the marker of target gene, miRNA sponge module include long-chain non-encoding ribonucleic acid (Long non-coding RNA, ) and mRNA lncRNA.

In alternative embodiments, module 210 is obtained, is specifically used for obtaining in each miRNA sponge module and own The first Pearson came (Pearson) correlation coefficient value of the average absolute value of LncRNA-mRNA coexpression pair.Pass through default random calculation Method generates multiple default miRNA sponge modules according to all LncRNA and mRNA in each miRNA sponge module.It obtains each Second Pearson correlation coefficient of the average absolute value of all LncRNA-mRNA coexpressions pair in default miRNA sponge module Value.By default pairing difference test algorithm, according to the first Pearson correlation coefficient value and the second Pearson correlation coefficient value, Obtain the saliency data of miRNA sponge module and default miRNA sponge module.

In alternative embodiments, module 210 is obtained, concentrates LncRNA, mRNA specifically for obtaining preset data Quantity and preset data concentrate the quantity with LncRNA, mRNA of target gene coexpression.It obtains in miRNA sponge module The quantity of LncRNA, mRNA of quantity and miRNA sponge module and the target gene coexpression of LncRNA, mRNA.According to pre- If the quantity and preset data of LncRNA, mRNA concentrate the number with LncRNA, mRNA of target gene coexpression in data set Amount, the LncRNA of the quantity and miRNA sponge module of LncRNA, mRNA and target gene coexpression in miRNA sponge module, The quantity of mRNA calculates the saliency data of miRNA sponge module and target gene by hypergeometric distribution check algorithm.

In alternative embodiments, module 210 is obtained, is specifically used for passing through the first preset algorithm, meter according to miRNA Calculate the value-at-risk of each matched sample.The value-at-risk of each matched sample of binaryzation obtains the first risk value set and the second wind Dangerous value set, wherein the value-at-risk in the first risk value set is greater than the value-at-risk in the second risk value set.According to the first wind Dangerous value set and the second risk value set calculate the first risk value set and the second risk value set by the second preset algorithm Hazard ratio.According to the first risk value set and the second risk value set, according to check algorithm, obtain the first risk value set and The significance of difference of second risk value set.According to Hazard ratio and the significance of difference determine miRNA whether be target gene mark Remember object.

Due to miRNA sponge module identification device for realizing above-mentioned miRNA sponge module recognition methods, Beneficial effect is identical, and details are not described herein.

Figure 11 is the identification device structure schematic diagram for the miRNA sponge module that one embodiment of the invention provides.

The above module can be arranged to implement one or more integrated circuits of above method, such as: one Or multiple specific integrated circuits (Application Specific Integrated Circuit, abbreviation ASIC), or, one Or multi-microprocessor (Digital SignalProcessor, abbreviation DSP), or, one or more field-programmable gate array Arrange (Field Programmable Gate Array, abbreviation FPGA) etc..For another example, when some above module passes through processing element When the form of scheduler program code is realized, which can be general processor, such as central processing unit (Central Processing Unit, abbreviation CPU) or it is other can be with the processor of caller code.For another example, these modules can integrate Together, it is realized in the form of system on chip (System-On-a-Chip, abbreviation SOC).

Figure 13 is the identification device structure schematic diagram for the miRNA sponge module that one embodiment of the application provides.

As shown in figure 13, the identification equipment of miRNA sponge module, comprising: processor 301, storage medium 302 and bus 303, storage medium 302 is stored with the executable machine readable instructions of processor, when the identification equipment of miRNA sponge module is run When, it is communicated between processor 301, storage medium 302 by bus 303, processor 301 executes machine readable instructions, to execute The step of recognition methods of above-mentioned miRNA sponge module.

The identification equipment of miRNA sponge module can be general purpose computer, server or mobile terminal etc., not limit herein System.The identification equipment of miRNA sponge module for realizing the application above method embodiment.

It should be noted that processor 301 may include one or more processing cores (for example, single core processor or multicore Processor).Only as an example, processor may include central processing unit (Central Processing Unit, CPU), specially With integrated circuit (Application Specific Integrated Circuit, ASIC), dedicated instruction set processor (Application Specific Instruction-set Processor, ASIP), graphics processing unit (Graphics Processing Unit, GPU), physical processing unit (Physics Processing Unit, PPU), digital signal processor (Digital Signal Processor, DSP), field programmable gate array (Field Programmable Gate Array, FPGA), programmable logic device (Programmable Logic Device, PLD), controller, microcontroller list Member, risc (Reduced Instruction Set Computing, RISC) or microprocessor etc. or its Any combination.

Storage medium 302 may include: including mass storage, removable memory, volatile read-write memory or Read-only memory (Read-Only Memory, ROM) etc., or any combination thereof.As an example, mass storage may include Disk, CD, solid state drive etc.；Removable memory may include flash drive, floppy disk, CD, storage card, zip disk, Tape etc.；Volatile read-write memory may include random access memory (Random Access Memory, RAM)；RAM can To include dynamic ram (Dynamic Random Access Memory, DRAM), Double Data Rate synchronous dynamic ram (Double Date-Rate Synchronous RAM, DDR SDRAM)；Static RAM (Static Random-Access Memory, SRAM), thyristor RAM (Thyristor-Based Random Access Memory, T-RAM) and zero capacitor RAM (Zero-RAM) etc..As an example, ROM may include mask rom (Mask Read-Only Memory, MROM), can compile Journey ROM (Programmable Read-Only Memory, PROM), erasable programmable ROM (Programmable Erasable Read-only Memory, PEROM), electrically erasable ROM (Electrically Erasable Programmable read only memory, EEPROM), CD ROM (CD-ROM) and digital versatile disk [Sony] ROM etc..

For ease of description, a processor 301 is only described in the identification equipment of miRNA sponge module.However, answering When note that the identification equipment of the miRNA sponge module in the application can also include multiple processors 301, therefore in the application The step of one processor of description executes can also be combined by multiple processors to be executed or is individually performed.For example, if miRNA is extra large The processor 301 of the identification equipment of continuous module executes step A and step B, then it should be understood that step A and step B can also be by two A different processor is executed jointly or is individually performed in a processor.For example, first processor executes step A, the Two processors execute step B or first processor and second processor executes step A and B jointly.

Optionally, it the present invention also provides a kind of computer readable storage medium, is stored on computer readable storage medium The step of having computer program, the recognition methods of miRNA sponge module is executed when computer program is run by processor.

In several embodiments provided by the present invention, 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, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list Member both can take the form of hardware realization, can also realize in the form of hardware adds SFU software functional unit.

The above-mentioned integrated unit being realized in the form of SFU software functional unit can store and computer-readable deposit at one In storage media.Above-mentioned SFU software functional unit is stored in a storage medium, including some instructions are used so that a computer Equipment (can be personal computer, server or the network equipment etc.) or processor (English: Processor) execute this hair The part steps of bright each embodiment the method.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (English: Read-Only Memory, abbreviation: ROM), random access memory (English: Random Access Memory, letter Claim: RAM), the various media that can store program code such as magnetic or disk.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention., rather than its limitations；To the greatest extent Pipe present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: its according to So be possible to modify the technical solutions described in the foregoing embodiments, or to some or all of the technical features into Row equivalent replacement；And these are modified or replaceed, various embodiments of the present invention technology that it does not separate the essence of the corresponding technical solution The range of scheme.

Claims

1. a kind of recognition methods of miRNA miRNA sponge module characterized by comprising

Obtain the expression matrix of the sponge gene of matched sample and the expression matrix of target gene；

According to the expression matrix of the expression matrix of the sponge gene and the target gene, multiple sponge gene-target genes are obtained Co-express module, wherein each sponge gene-target gene coexpression module indicates that one kind can use sponge gene and target gene The gene of co-expression；

It obtains in each sponge gene-target gene coexpression module, shares significance value, sponge gene and the target of miRNA Sensibility canonical correlation coefficient when canonical correlation coefficient and shared miRNA between gene；

According to the quantity of sponge gene and target gene in each sponge gene-target gene coexpression module, described shared It is quick when canonical correlation coefficient and the shared miRNA between the significance value of miRNA, the sponge gene and target gene Perceptual canonical correlation coefficient determines whether each sponge gene-target gene coexpression module is miRNA sponge module.

2. the method according to claim 1, wherein according to the expression matrix of the sponge gene and the target base The expression matrix of cause obtains multiple sponge genes-target gene coexpression module, comprising:

According to clustering algorithm, the expression matrix of expression matrix and the target gene to the sponge gene is clustered, and is obtained Multiple cluster results of the sponge gene and the target gene；

By the sponge gene and the target gene in each cluster result, as a sponge gene-target base Because co-expressing module.

3. according to the method described in claim 2, it is characterized in that, the clustering algorithm includes: unidirectional clustering algorithm or two-way Clustering algorithm；

If the clustering algorithm is the unidirectional clustering algorithm, according to the expression matrix of the sponge gene, the target gene Expression matrix and the matched sample, the sponge gene and the target gene are clustered；

If the clustering algorithm is the bidirectional clustering algorithm, according to the expression matrix of the sponge gene, the target gene Expression matrix and predetermined fraction the matched sample, to the institute of the sponge gene, the target gene and predetermined fraction Matched sample is stated to be clustered.

4. method according to claim 1-3, which is characterized in that obtain the significance value of shared miRNA, packet It includes:

According to preset miRNA- target gene regulation relationship, the sponge gene-target base is obtained by hypergeometric distribution method of inspection Because sharing the significance value of miRNA between the sponge gene and the target gene in coexpression module.

5. method according to claim 1-3, which is characterized in that obtain the allusion quotation between sponge gene and target gene Type related coefficient, comprising:

According to the expression matrix of the expression matrix of the sponge gene and the target gene, the column vector of the sponge gene is obtained And the column vector of the target gene；

Obtain the variance matrix and covariance between the expression matrix of the sponge gene and the expression matrix of the target gene Matrix；

According to the column vector of the sponge gene, the column vector of the target gene, the variance matrix, the covariance matrix and Preset representative vectors calculate the canonical correlation coefficient.

6. method according to claim 1-3, which is characterized in that sensibility when obtaining shared miRNA is typical Related coefficient, comprising:

It obtains between the canonical correlation coefficient and shared miRNA and the target gene between shared miRNA and the sponge gene Canonical correlation coefficient；

According between the sponge gene and target gene canonical correlation coefficient, the shared miRNA and the sponge gene it Between canonical correlation coefficient, the canonical correlation coefficient between the shared miRNA and the target gene, calculate the sponge gene Inclined canonical correlation coefficient between the target gene；

Canonical correlation coefficient between the sponge gene and target gene is subtracted between the sponge gene and the target gene Inclined canonical correlation coefficient, obtain sensibility canonical correlation coefficient when the shared miRNA.

7. according to the method described in claim 6, it is characterized in that, described obtain between shared miRNA and the sponge gene Canonical correlation coefficient and shared miRNA and the target gene between canonical correlation coefficient, comprising:

Obtain the expression matrix of the shared miRNA；

According to the expression matrix of the shared miRNA, the expression matrix of the expression matrix of the sponge gene and the target gene, Obtain the column vector, the column vector of the sponge gene and the column vector of the target gene of the shared miRNA；

Obtain the variance matrix between the expression matrix of the shared miRNA and the expression matrix of the sponge gene and association Variance matrix；

According to the column vector of the shared miRNA, the column vector of the sponge gene, the shared miRNA expression matrix and Variance matrix and covariance matrix and preset representative vectors between the expression matrix of the sponge gene, described in calculating Canonical correlation coefficient between shared miRNA and the sponge gene；

Obtain the variance matrix between the expression matrix of the shared miRNA and the expression matrix of the target gene and association side Poor matrix；

According to the column vector of the shared miRNA, the column vector of the target gene, the expression matrix of the shared miRNA and institute Variance matrix between the expression matrix of target gene and covariance matrix and preset representative vectors are stated, are calculated described shared Canonical correlation coefficient between miRNA and the target gene.

8. the method according to claim 1, wherein described be total to table according to each sponge gene-target gene Up to the quantity of sponge gene and target gene, the significance value of the shared miRNA, the sponge gene and target gene in module it Between canonical correlation coefficient and the shared miRNA when sensibility canonical correlation coefficient determine each sponge gene- Target gene co-expresses whether module is miRNA sponge module, comprising:

If the quantity of sponge gene and target gene is all larger than equal to 2 in the sponge gene-target gene coexpression module, described total The significance value of miRNA is enjoyed less than 0.05, the canonical correlation coefficient between the sponge gene and target gene is described greater than 0.8 Sensibility canonical correlation coefficient when shared miRNA is greater than 0.1, it is determined that the sponge gene-target gene co-expresses module and is MiRNA sponge module.

9. the method according to claim 1, wherein determining each sponge gene-target gene coexpression After whether module is miRNA sponge module, further includes:

If the sponge gene-target gene coexpression module is miRNA sponge module, obtain the miRNA sponge module with The correlation data of target gene, wherein the correlation data includes the miRNA sponge module and default miRNA sponge The saliency data of module, the miRNA sponge module and the miRNA sponge module whether be target gene marker, The miRNA sponge module includes long-chain non-encoding ribonucleic acid LncRNA and protein coding ribonucleic acid mRNA.

10. according to the method described in claim 9, it is characterized in that, obtaining the miRNA sponge module and the sea default miRNA The saliency data of continuous module, comprising:

Obtain the first Pierre of the average absolute value of all LncRNA-mRNA coexpressions pair in each miRNA sponge module Inferior Pearson correlation coefficient value；

By presetting random algorithm, according to all LncRNA and mRNA in each miRNA sponge module, generate multiple described Default miRNA sponge module；

Obtain second of the average absolute value of all LncRNA-mRNA coexpressions pair in each default miRNA sponge module Pearson correlation coefficient value；

By default pairing difference test algorithm, according to the first Pearson correlation coefficient value and the 2nd Pearson phase Coefficient values obtain the saliency data of the miRNA sponge module and default miRNA sponge module.

11. according to the method described in claim 9, it is characterized in that, obtaining the miRNA sponge module and the target gene Saliency data, comprising:

Obtaining preset data concentrates the quantity of LncRNA, mRNA and the preset data to concentrate and target gene coexpression The quantity of LncRNA, mRNA；

It is total to obtain the quantity of LncRNA, mRNA and the miRNA sponge module and target gene in the miRNA sponge module The quantity of LncRNA, mRNA of expression；

It concentrates the quantity of LncRNA, mRNA and the preset data to concentrate according to the preset data to co-express with target gene The quantity of LncRNA, mRNA, in the miRNA sponge module LncRNA, mRNA quantity and the miRNA sponge module The miRNA sponge mould is calculated by hypergeometric distribution method of inspection with the quantity of LncRNA, mRNA of target gene coexpression The saliency data of block and the target gene.

12. according to the method described in claim 9, it is characterized in that, determining whether the miRNA sponge module is target gene Marker, comprising:

The value-at-risk of each matched sample is calculated by the first preset algorithm according to the miRNA；

The value-at-risk of each matched sample described in binaryzation obtains the first risk value set and the second risk value set, Described in value-at-risk in the first risk value set be greater than the value-at-risk in the second risk value set；

Described first is calculated by the second preset algorithm according to the first risk value set and the second risk value set The Hazard ratio of risk value set and the second risk value set；

First risk is obtained according to check algorithm according to the first risk value set and the second risk value set The significance of difference of value set and the second risk value set；

According to the Hazard ratio and the significance of difference determine the miRNA whether be the target gene marker.

13. a kind of identification device of miRNA sponge module characterized by comprising

Module is obtained, for obtaining the expression matrix of the sponge gene of matched sample and the expression matrix of target gene；

The acquisition module is also used to the expression matrix of the expression matrix and the target gene according to the sponge gene, obtains Multiple sponge genes-target gene co-expresses module, wherein each sponge gene-target gene coexpression module indicates a kind of The gene that can be co-expressed with sponge gene and target gene；

The acquisition module is also used to obtain in each sponge gene-target gene coexpression module, shares the aobvious of miRNA Write property value, sponge gene and target gene between canonical correlation coefficient and shared miRNA when sensibility canonical correlation coefficient；

Determining module, for the number according to sponge gene and target gene in each sponge gene-target gene coexpression module Canonical correlation coefficient between amount, the significance value of the shared miRNA, the sponge gene and target gene and described shared Sensibility canonical correlation coefficient when miRNA determines whether each sponge gene-target gene coexpression module is the sea miRNA Continuous module.