A hypergraph-based chromatin domain identification method and system

Abstract

The application provides a chromatin domain recognition method and system based on a hypergraph, and the method comprises the following steps: acquiring epigenome data; and pre-processing the data, fragment cutting the chromatin according to a certain base pair quantity, regarding each fragment as a node of the genome, and acquiring a Hi-C interaction matrix; calculating a feature vector of each genome node represented by spatial interaction data and epigenome data; generating an initial TADs division for the Hi-C data matrix, clustering the nodes according to the node features by using a clustering algorithm, adjusting the nodes in the cluster, and obtaining an optimized TADs structure division. The optimized TADs division is expressed as a hypergraph, a TADs node set is used as a hyperedge, and the hypergraph division is recognized. The divided chromatin region is mapped to a cell nucleus to explore the biological significance. The application is suitable for the recognition and function-related research of chromatin domains in the field of biological information calculation.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of biological computing, and in particular to a chromatin domain identification method and device based on hypergraph. BACKGROUND

[0002] Chromosome conformation capture (3C) technology and its derivative technology Hi-C combined with high-throughput sequencing technology provide a technical basis for the generation of chromatin interaction data on the whole genome. The number of interactions is represented as an n*n matrix, where n is the number of fragments of the chromosome window (bin) cut by a restriction enzyme, and the size of the bin becomes the resolution of Hi-C. With the continuous development of Hi-C technology, we can obtain the contact matrix of the whole genome, i.e. Hi-C data, for three-dimensional genome modeling and hierarchical structure research.

[0003] Eukaryotic chromosomes are usually composed of multiple levels, including A / B compartments, topologically associated domains (TADs), and chromatin loops (loops). Among them, based on the exploration of local spatial relationships, it can be found out whether there is a potential relationship between the regulatory elements and the genes and what kind of potential relationship exists, so as to carry out deeper research.

[0004] In mathematics, a hypergraph is a generalization of a graph, where an edge can connect any number of vertices. Hypergraph structure has strong data sample non-linear high-order correlation description and mining ability compared with general graph structure. Applying hypergraph to the field of biological information can better represent the high-order interaction relationship between chromosome nodes.

[0005] At present, chromatin domain division algorithms generally divide A / B compartments with large scales or identify topologically associated domains (TADs) at the MB level, but for chromatin structures between TADs and A / B compartments in the nuclear space, such as A1, B1, A2, B2 compartments. The current algorithm does not perform a structural and functional clustering analysis on this type of structure. Therefore, the present application provides a chromatin domain division algorithm based on hypergraph, which first constructs a hypergraph using the identified TADs, and then performs hypergraph division based on the hypergraph to realize functional clustering of TADs and realize corresponding research of chromatin domains and nuclear functions. SUMMARY

[0006] The present application provides a chromatin domain division method and device based on hypergraph to solve the technical problems of chromatin domain identification and research.

[0007] To solve the above technical problems, the present application provides the following technical solutions:

[0008] In one aspect, the present application provides a chromatin domain division method, comprising:

[0009] Obtain the raw Hi-C data and the epigenomic data corresponding to the cell lines containing the Hi-C data;

[0010] Chromatin is fragmented according to a certain number of base pairs. Each fragment (bin) is regarded as a node in the genome. The number of base pairs in the node is the resolution. The Hi-C interaction matrix at a specific resolution is obtained.

[0011] Calculate the feature vectors of each genomic node, represented by spatial interaction data and epigenome data;

[0012] The initial TADS partitioning is generated using the Hi-C data matrix. The existing TADS partitioning is represented as inter-domain, intra-domain, and domain boundary node clusters. The intra-cluster nodes are adjusted using the K-means clustering algorithm based on the feature vectors to obtain the optimized TADS partitioning.

[0013] The optimized TADs partition is represented as a hypergraph, and the set of TAD nodes is used as a hyperedge. By partitioning the hypergraph, the partitioned chromatin regions are mapped to the cell nucleus space to explore their positional relationship with the cell nucleus, thereby further studying the biological significance of the chromatin domain.

[0014] Furthermore, the epigenomic data includes signals that are significantly enriched at the boundaries of TADs, including four types of signals: CTCF, RAD21, H3K36me3, and H3K4me3, and signals that are significantly enriched inside TADs, including two types of signals: H3K9me3 and H3K27me3.

[0015] Furthermore, the preprocessing of the raw Hi-C data includes: using the Juicer tool to obtain a Hi-C data matrix at a specified resolution from the raw Hi-C data.

[0016] Furthermore, the process of partitioning the initial TADs structure includes: taking the Hi-C matrix data as input, implementing the TADtree algorithm, and obtaining the initial TADs boundaries.

[0017] Furthermore, the adjustment process for TADS partitioning includes: using the K-means algorithm, based on the feature vector representation of genomic nodes, iteratively updating the clustering results to obtain the final TADs partitioning results.

[0018] Furthermore, the hypergraph representation and partitioning process includes: representing each TAD as a hyperedge, and achieving hypergraph partitioning through stages such as hypergraph coarsening, initial partitioning, and optimization.

[0019] On the other hand, the present invention also provides a chromatin domain partitioning device based on a hypergraph, comprising:

[0020] The data acquisition module is used to acquire the raw Hi-C data and the epigenomic data of the cell line containing the raw Hi-C data;

[0021] The data preprocessing module is used to cut the chromatin into fragments according to a certain number of base pairs in the raw Hi-C data acquired by the data acquisition module, treat each fragment (bin) as a node in the genome, and the number of base pairs in the node is the resolution, and obtain the Hi-C interaction matrix at a specific resolution.

[0022] The feature representation module is used to calculate the feature vectors of each genomic node, which are represented by spatial interaction data and epigenome data.

[0023] The TADS identification module is used to generate an initial TADs structure partition using existing algorithms based on the Hi-C data generated by the data preprocessing module, and to update the TADs partitioning results using the K-means algorithm based on the node features generated by the feature representation module.

[0024] The hypergraph partitioning module is used to represent the TADs partitioning results obtained by the TADs identification module as a hypergraph, and to partition the hypergraph to obtain the corresponding chromatin domains.

[0025] The beneficial effects of the above-described technical solution of the present invention are as follows:

[0026] This invention acquires raw Hi-C data and epigenomic data of the cell line in which the data is derived. The raw data is preprocessed by segmenting chromatin into fragments according to a certain number of base pairs. Each fragment (bin) is considered a node in the genome, and the number of base pairs in the node represents the resolution, resulting in a Hi-C interaction matrix at a specific resolution. Feature vectors representing each genomic node are obtained from the spatial interaction data and epigenomic data. Using a pre-researched algorithm, initial TADs are generated from the Hi-C data matrix, and the K-means algorithm is used to optimize the TADs partitioning results. The TADs are represented as hyperedges, and based on a hypergraph partitioning model, the partitioned chromatin domains are obtained. This fully utilizes Hi-C data and epigenomic data to achieve the identification and study of chromatin domains based on hypergraph technology. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1This is a flowchart illustrating the chromatin domain identification method based on a hypergraph provided in the first embodiment of the present invention.

[0029] Figure 2 This is a schematic diagram of the spatial interaction feature representation of chromosome nodes provided in the first embodiment of the present invention.

[0030] Figure 3 This is a schematic diagram representing the TAD partitioning result provided in the first embodiment of the present invention.

[0031] Figure 4 This is a schematic diagram of the hypergraph partitioning process provided in the first embodiment of the present invention.

[0032] Figure 5 This is a logical block diagram of a chromatin domain identification system based on a hypergraph according to the present invention. Detailed Implementation

[0033] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0034] like Figure 1 As shown, this invention provides a chromatin domain identification method based on a hypergraph, which specifically includes the following steps:

[0035] S1) Obtain the raw Hi-C data and the epigenetic data corresponding to the cell line containing the Hi-C data;

[0036] S2) Chromatin is fragmented according to a certain number of base pairs. Each fragment (bin) is regarded as a node in the genome. The number of base pairs in the node is used as the resolution to obtain the Hi-C interaction matrix at a specific resolution.

[0037] S3) Calculate the feature vector of each genomic node represented by spatial interaction data and epigenomic data;

[0038] S4) Use the Hi-C interaction matrix obtained in S2) to generate the initial TADs partitioning method and perform partitioning. Use the K-means clustering algorithm to optimize the partitioning result and obtain the optimized TADs partitioning.

[0039] S5) The optimized TADs are represented as a hypergraph, and each set of TAD nodes is used as a hyperedge. The hypergraph is then identified through partitioning.

[0040] The epigenomic data in S1) are relevant factors that are significantly enriched at the TADS boundary.

[0041] The preprocessing of the Hi-C data used in S1) includes: using the Juicer tool to extract the original .hic file and generate a Hi-C data matrix with a specific resolution.

[0042] The process of preprocessing the epigenomic data used in S1) includes extracting epigenomic data corresponding to each genomic node with a specific resolution.

[0043] The specific steps in S2) are as follows: After dividing the genome into n nodes at a specific resolution, extract 6 types of epigenomic data for each node, and add the spatial interaction features of each node. The spatial interaction between the node and the genome within a 1Mb range to the left and right is the Hi-C interaction matrix.

[0044] Specifically, in S3), the original data is first extracted according to chromosome number and then converted into a matrix with the same dimensions as the Hi-C matrix, which serves as the final epigenomic data.

[0045] In S4), the Hi-C interaction matrix is ​​initially divided into TADs using a pre-researched algorithm, generating hierarchically nested TADs, and the nodes are represented as three clusters based on the TADs partitioning results, representing the set of nodes within the TADs domain, nodes between the domains, and nodes at the domain boundaries, respectively.

[0046] Then, cluster the initially partitioned node clusters, and adjust the node positions using the K-means clustering algorithm based on the node feature vectors to generate an optimized TADs partitioning method.

[0047] In S5), the optimized TADs structure is represented using a hypergraph, and TADs are represented as hyperedges.

[0048] In step S5), the hypergraph is divided, and the divided chromatin regions are mapped to the cell nucleus space for identification.

[0049] like Figure 5 As shown, the present invention discloses a chromatin domain identification system based on a hypergraph, the system comprising:

[0050] The data acquisition module is used to acquire the raw Hi-C data and the epigenomic data of the cell line containing the raw Hi-C data;

[0051] The data preprocessing module is used to cut the chromatin into fragments according to a certain number of base pairs in the raw Hi-C data acquired by the data acquisition module, treat each fragment as a node in the genome, and the number of base pairs in the node is the resolution, and obtain the Hi-C interaction matrix at a specific resolution.

[0052] The feature representation module is used to calculate the feature vectors of each genomic node, which are represented by spatial interaction data and epigenome data.

[0053] The TADS identification module is used to generate an initial TADs structure partition using existing algorithms based on the Hi-C data generated by the data preprocessing module, and to update the TADs partitioning results using the K-means algorithm based on the node features generated by the feature representation module.

[0054] The hypergraph partitioning module is used to partition the hypergraph by representing the TADs partitioning results obtained by the TADs identification module. Each hyperedge represents a continuous set of TADs nodes.

[0055] Example 1

[0056] This invention provides a chromatin domain identification method based on a hypergraph. This method can be implemented by an electronic device, which can be a terminal or a server. The execution flow of this method is as follows: Figure 1 As shown, it includes:

[0057] S1, Obtain the raw Hi-C data and the epigenomic data corresponding to the cell line containing the Hi-C data;

[0058] Specifically, the epigenomic data identified factors that were significantly enriched at the boundaries of TADs, including CTCF, RAD21, H3K36me3, and H3K4me3, as well as factors that were enriched within TADs, including H3K9me3 and H3K27me3.

[0059] S2, preprocess the raw data, cut the chromatin into fragments according to a certain number of base pairs, treat each fragment (bin) as a node in the genome, and the number of base pairs in the node is the resolution, and obtain the Hi-C interaction matrix at a specific resolution;

[0060] Specifically, in this embodiment, the preprocessing of the raw Hi-C data includes applying the Juicer tool to the raw Hi-C data to obtain a Hi-C data matrix with a specific resolution.

[0061] S3, calculate the feature vector of each genomic node represented by spatial interaction data and epigenomic data;

[0062] The process of preprocessing epigenomic data includes extracting data from the raw data by chromosome number and converting it into a matrix of the same dimensions as the Hi-C matrix, which serves as the final epigenomic data.

[0063] Specifically, the representation process of feature vectors for each genomic locus includes spatial interaction data features and epigenomic data features.

[0064] like Figure 2 As shown, the spatial interaction features of genomic locus i are represented by the interaction values ​​of the genomes in the 1Mb regions upstream and downstream of i. Taking a resolution of 25Kb as an example, 40 bins upstream and downstream of the locus are taken, for a total of 80 interaction values.

[0065] The epigenomic data features of genomic locus i consist of six signal data types: CTCF, RAD21, H3K36me3, H3K4me3, H3K9me3, and H3K27me3.

[0066] The features of site i are represented by a 86-dimensional feature vector, which is composed of spatial interaction features and epigenomic features.

[0067] S4, use the Hi-C data matrix to generate the initial TADs partitioning method, as shown below:

[0068] Algorithm: TAD structure optimization algorithm based on K-means clustering

[0069]

[0070] The existing TADs are represented as inter-domain, intra-domain, and domain boundary node clusters. The nodes within the clusters are adjusted based on the feature vector using the K-means clustering algorithm to obtain the optimized TADs partitioning.

[0071] Specifically, in this embodiment, the process of dividing the initial TADs includes:

[0072] Based on the Hi-C contact matrix A, the fold enrichment of each pair of contacts with a distance of d is calculated:

[0073]

[0074] Where J is the number of bins after the chromosome is divided equally, and A is a J×J matrix. ij This represents the contact frequency between bin i and bin j;

[0075] For each interval [i, j], estimate the parameters. These two parameters determine the expected contact frequency of paired bins within each TAD;

[0076]

[0077] Next, we calculate the Boundary index (BI) for each genomic locus i, which is a one-dimensional statistical statistic used to find local variations in the interaction frequency at the boundaries of TADs;

[0078]

[0079] BI measures the change in contact near i. Specifically, within a length p containing i, BI calculates the total difference in contact frequency between q bins upstream and downstream of i.

[0080] Find the TADs tree with the maximum BI and the best contact matrix A through dynamic programming, and then select the optimal set of TADs trees to form a TADs forest;

[0081] To minimize the difference between A and the expected contact frequency, the following objective function is used:

[0082]

[0083] Specifically, in this embodiment, the pseudocode flow for optimizing the TADs structure partitioning based on the K-means algorithm is as follows: Figure 4 As shown:

[0084] First, the three types of node clusters obtained from the initial TADs partitioning—inter-domain, intra-domain, and domain boundary—are used as initial inputs, and the mean of each class is calculated as the center point.

[0085] Then, for each data point, the distance between it and each center point is calculated in turn, and it is assigned to the cluster with the closest distance.

[0086] For the updated cluster, repeat the above steps until the cluster's center point no longer changes or the number of iterations reaches its maximum value.

[0087] S5 represents the optimized TADs as a hypergraph, uses the set of TADs nodes as hyperedges, and maps the segmented chromatin regions to the cell nucleus space by segmenting the hypergraph, exploring their positional relationship with the cell nucleus, thereby further studying the biological significance of the chromatin domain.

[0088] The process of hypergraph segmentation is as follows: Figure 4 As shown, it includes:

[0089] During the coarsening stage, closely connected nodes are merged to reduce the size of the hypergraph until the predetermined size is reached.

[0090] Here, "close" means that for a given node, considering all nodes sharing the same hyperedge with it, if the number of times a node shares the same hyperedge with it is greater than the number of times the other nodes share the same hyperedge, then the two nodes are considered to be closely connected.

[0091] In the initial partitioning phase, the hypergraph is initially partitioned to obtain the initial k subgraphs;

[0092] The initial partitioning method involves random partitioning, traversing all partitions and nodes, and generating a random number between 0 and 1 for each node. If the random number is greater than 0.5, the current node is placed in a certain partition; otherwise, it is not placed.

[0093] In the optimization phase, the coarsened hypergraph is sequentially split and restored, where splitting refers to dividing a node into two, until the initial hypergraph size is reached. During this phase, the initial hypergraph partition is adjusted, gradually reducing the number of fragmented hypergraph nodes, ultimately yielding the optimal k-way hypergraph partition.

[0094] Example 2

[0095] This embodiment provides a chromatin domain partitioning device based on a hypergraph, which includes the following modules:

[0096] The data acquisition module is used to acquire the raw Hi-C data and the epigenomic data of the cell line containing the raw Hi-C data;

[0097] The data preprocessing module is used to cut the chromatin into fragments according to a certain number of base pairs in the raw Hi-C data acquired by the data acquisition module, treat each fragment (bin) as a node in the genome, and the number of base pairs in the node is the resolution, and obtain the Hi-C interaction matrix at a specific resolution.

[0098] The feature representation module is used to calculate the feature vectors of each genomic node, which are represented by spatial interaction data and epigenome data.

[0099] The TADs identification module is used to generate an initial TADs structure partition using existing algorithms based on the Hi-C data generated by the data preprocessing module, and to update the TADs partitioning results using the K-means algorithm based on the node features generated by the data preprocessing module.

[0100] The hypergraph partitioning module is used to represent the TADs partitioning results obtained by the TADs identification module as a hypergraph, and to segment the hypergraph to obtain the corresponding chromatin domains, thereby further studying the relevant biological significance of the chromatin domains.

[0101] The hypergraph-based chromatin domain partitioning device of this embodiment corresponds to the chromatin domain partitioning method of the first embodiment described above; wherein, the functions implemented by each functional module in the chromatin domain partitioning device of this embodiment correspond one-to-one with the process steps in the chromatin domain partitioning method of the first embodiment described above; therefore, it will not be described again here.

[0102] The foregoing has provided a detailed description of a hypergraph-based chromatin domain identification method and system provided in the embodiments of this application. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and its core ideas; furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

[0103] Certain terms are used in the specification and claims to refer to specific components. Those skilled in the art will understand that hardware manufacturers may use different names to refer to the same component. This specification and claims do not distinguish components based on differences in name, but rather on differences in function. The terms "comprising" and "including" used throughout the specification and claims are open-ended and should be interpreted as "comprising / including but not limited to". "Approximately" means that within an acceptable margin of error, those skilled in the art can solve the technical problem and substantially achieve the technical effect within a certain margin of error. The following descriptions in the specification are preferred embodiments for carrying out this application; however, these descriptions are for the purpose of illustrating the general principles of this application and are not intended to limit the scope of this application. The scope of protection of this application shall be determined by the appended claims.

[0104] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a product or system comprising a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a product or system. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the product or system that includes said element.

[0105] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0106] The foregoing description illustrates and describes several preferred embodiments of this application. However, as previously stated, it should be understood that this application is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the application concept described herein through the foregoing teachings or the technology or knowledge in related fields. Modifications and variations made by those skilled in the art do not depart from the spirit of this application.

Claims

1. A chromatin domain identification method based on hypergraphs, characterized in that, The identification method specifically includes the following steps: S1) Obtain the raw Hi-C data and the epigenetic data corresponding to the cell line containing the Hi-C data; S2) The chromatin is fragmented according to a certain number of base pairs. Each fragment (bin) is regarded as a node in the genome. The number of base pairs in the node is used as the resolution, and the resolution is used as the Hi-C interaction matrix. The specific steps are as follows: After dividing the genome into n nodes at the specified resolution, extract six types of epigenome data for each node, and add the spatial interaction features of each lineage node. The spatial interaction between the node and the genome within a 1Mb range to the left and right is the Hi-C interaction matrix. S3) Calculate the feature vector of each genomic node represented by spatial interaction data and epigenomic data; S4) Use the Hi-C interaction matrix obtained in S2) to generate the initial TADs partitioning method and perform partitioning. Use the K-means clustering algorithm to optimize the partitioning result and obtain the optimized TADs partitioning. The Hi-C interaction matrix is ​​initially partitioned into TADs using a pre-researched algorithm, generating hierarchically nested TADs. Based on the TADs partitioning results, the nodes are represented as three clusters, representing the sets of nodes within the TAD domain, nodes between the domains, and nodes at the domain boundaries, respectively. The node clusters after the initial partitioning are then clustered, and the K-means clustering algorithm is used to adjust the node positions based on the node's feature vectors, generating an optimized TADs partitioning method. S5) The optimized TADs are represented as a hypergraph, and each set of TAD nodes is used as a hyperedge. The hypergraph is then identified through partitioning.

2. The identification method according to claim 1, characterized in that, The epigenomic data in S1) are relevant factors that are significantly enriched at the TADS boundary.

3. The identification method according to claim 1, characterized in that, The preprocessing of the Hi-C data used in S2) includes: using the Juicer tool to extract the original .hic file and generate a resolution Hi-C interaction matrix.

4. The identification method according to claim 1, characterized in that, The process of preprocessing the epigenomic data used in S1) includes extracting the epigenomic data corresponding to each genomic node with a resolution.

5. The identification method according to claim 4, characterized in that, Specifically, in S3), the original data is first extracted according to chromosome number and then converted into a matrix with the same dimensions as the Hi-C interaction matrix, which serves as the final epigenomic data.

6. The identification method according to claim 1, characterized in that, In S5), the optimized TADs structure is represented using a hypergraph, and TADs are represented as hyperedges.

7. The identification method according to claim 1, characterized in that, In step S5), the hypergraph is divided, and the divided chromatin regions are mapped to the cell nucleus space for identification.

8. A chromatin domain identification device based on a hypergraph, characterized in that, The device includes: a data acquisition module for acquiring raw Hi-C data and epigenomic data of the cell line containing the raw Hi-C data; The data preprocessing module is used to cut the raw Hi-C data acquired by the data acquisition module into fragments according to a certain number of base pairs, treat each fragment as a node in the genome, and the number of base pairs in the node is the resolution, and obtain the Hi-C interaction matrix at the resolution. Specifically, at a certain resolution, after dividing the genome into n nodes, six types of epigenome data are extracted for each node. At the same time, the spatial interaction features of each lineage node are added. The spatial interaction between the node and the genome within a 1Mb range to the left and right is the Hi-C interaction matrix. The feature representation module is used to calculate the feature vectors of each genomic node, which are represented by spatial interaction data and epigenome data. The TADS identification module is used to generate an initial TADs structure partition using existing algorithms based on the Hi-C data generated by the data preprocessing module, and to update the TADs partitioning results using the K-means algorithm based on the node features generated by the feature representation module. The Hi-C interaction matrix is ​​initially partitioned into TADs using a pre-researched algorithm, generating hierarchically nested TADs. Based on the TADs partitioning results, the nodes are represented as three clusters, representing the sets of nodes within the TAD domain, nodes between the domains, and nodes at the domain boundaries, respectively. The node clusters after the initial partitioning are then clustered, and the K-means clustering algorithm is used to adjust the node positions based on the node's feature vectors, generating an optimized TADs partitioning method. The hypergraph partitioning module is used to represent the TADs partitioning results obtained by the TADs identification module as a hypergraph, where each hyperedge represents a continuous set of TAD nodes, and the hypergraph is partitioned and identified.

Images