Method for enhancing chromatin interaction information based on RNA-DNA and DNA-DNA interaction information
By integrating RNA-DNA and DNA-DNA interaction information, the Hi-C matrix is enhanced using the CNN network model, and the problem of insufficient integration of RNA interaction information and Hi-C data in the existing technology is solved, and efficient multimodal data analysis and refined chromatin structure research are achieved.
Patent Information
- Application Number
- CN202510656424.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-21
- Publication Date
- 2025-08-05
- Estimated Expiration
- 2045-05-21
AI Technical Summary
Existing genomic research technologies cannot effectively integrate RNA interaction information and Hi-C data, lack a systematic computational analysis framework, and cannot fully reveal the role of RNA in chromatin regulation.
By constructing a chromatin interaction information enhancement method based on RNA-DNA and DNA-DNA interaction information, the CNN network model is used to integrate iMARGI data and Hi-C data to generate the enhanced Hi-C matrix, and realize the effective integration of RNA information to chromatin interaction and the fusion analysis of multimodal data.
It improves the spatial resolution and prediction accuracy of Hi-C data, has good generalization ability across RNA types, supports quantitative functional analysis and visual verification, and promotes the development of RNA-chromatin interaction research.
Smart Images

Figure CN120183510B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of bioinformatics, and in particular to a method for enhancing chromatin interaction information based on RNA-DNA and DNA-DNA interaction information. Background Art
[0002] In recent years, research on the functions of RNA in the nucleus of eukaryotic cells and its interactions with chromatin has deepened. RNA is no longer viewed solely as a mediator of gene expression; it plays diverse and important roles in the dynamic regulation of chromatin conformation. Multiple RNA species exist within cells, including coding RNA and noncoding RNA (such as lncRNA, snoRNA, and snRNA). These RNAs participate in gene transcription, epigenetic regulation, and cell fate determination by regulating chromatin structure and function. The study of RNA-chromatin interactions provides new insights into the regulation of gene expression.
[0003] Currently, the interaction between RNA and chromatin is believed to play an important role in the organization of the cell nucleus. By binding to chromatin, RNA can regulate the open state of chromatin, affect the transcriptional program of genes, and ensure that cells maintain normal gene expression patterns during physiological processes. Existing genomic research technologies, such as Hi-C and eCLIP-seq, can already provide preliminary data on the interaction between chromatin and RNA. However, most existing technologies focus on analyzing the three-dimensional structure of DNA and lack tools to integrate RNA interaction information, making it impossible to fully reveal the role of RNA in chromatin regulation.
[0004] Among existing technologies, MARGI (Mapping of RNA-Genome Interaction) technology, an early approach for studying RNA-DNA interactions, combines cross-linking immunoprecipitation with high-throughput sequencing to initially resolve RNA binding sites on the genome. However, this method still has limitations in terms of spatial accuracy and resolution. The subsequent development of iMARGI (in situ MARGI) technology has achieved significant improvements in in situ capture and spatial resolution, allowing for more accurate localization of RNA-DNA interaction sites and enabling experiments to be conducted under conditions that preserve chromatin structure, enhancing their biological significance.
[0005] GRID-seq (Global RNA Interaction with DNA), a deep sequencing technology, can capture the global interaction between RNA and DNA, revealing chromatin-associated RNA sites and the specific locations of their binding to chromatin. This technology provides an important tool for studying the role of RNA in chromatin regulation. However, GRID-seq currently lacks the ability to comprehensively process multimodal interaction data between different RNA types and chromatin, and lacks the ability to integrate with other genomic data (such as Hi-C and epigenetic data).
[0006] RIC-seq (RNA in situ conformation sequencing) provides a deeper understanding of the spatial organization of RNA and chromatin by measuring RNA secondary and tertiary structure and RNA-RNA interactions at high resolution. Although RIC-seq can provide rich information on RNA structure, the complexity of interpreting its data and how to integrate this data with other data on chromatin interactions remain technical challenges.
[0007] To fill these gaps, RD-SPRITE technology can help reveal the role of RNA in the spatial regulation of chromatin by mapping the spatial interactions between RNA and DNA into high-resolution maps. However, although the above technologies provide a wealth of tools and data sources for the study of RNA-chromatin interactions, due to the diversity of technical processes, there is still a lack of a systematic computational analysis framework that can effectively integrate RNA interaction data produced by different technologies with chromatin conformation data such as Hi-C. Therefore, the development of an efficient analysis tool that can integrate RNA interaction and Hi-C data has become an urgent problem to be solved in the current field of bioinformatics. Summary of the Invention
[0008] In response to the above problems, the purpose of the present invention is to provide a chromatin interaction information enhancement method based on RNA-DNA and DNA-DNA interaction information. This method integrates RNA interaction data with Hi-C data to construct a multimodal data integration and analysis platform, which can make up for the shortcomings of existing technologies in multimodal data integration and promote the further development of RNA-chromatin interaction research.
[0009] In order to solve the above technical problems, the present invention provides the following technical solutions:
[0010] A method for enhancing chromatin interaction information based on RNA-DNA and DNA-DNA interaction information, the method comprising the following steps:
[0011] S1. Collect iMARGI data of RNA-DNA interactions and raw Hi-C data of DNA-DNA interactions from the same cell line.
[0012] S2. Perform feature extraction on the iMARGI data and calculate the Pearson correlation coefficient to obtain a DNA-DNA feature correlation coefficient matrix of the fused RNA information of the whole genome; perform FAN completion processing on the original Hi-C data after downsampling to obtain a complete Hi-C matrix;
[0013] S3. Construct a CNN network model, use the DNA-DNA feature correlation coefficient matrix of the fused RNA information and the complete Hi-C matrix as the model input, train the model, and output the enhanced Hi-C matrix;
[0014] S4. For the trained model, input data of different RNA types and predict the corresponding Hi-C matrix for downstream analysis.
[0015] Optionally, step S1 specifically includes:
[0016] The iMARGI data and the original Hi-C data are obtained from public databases or experimental data; quality control is performed on the obtained original data to remove noise data.
[0017] Optionally, step S2 specifically includes:
[0018] Feature extraction was performed on the RNA-DNA interaction data, and the Pearson correlation coefficient of the extracted features was calculated to obtain the DNA-DNA feature correlation coefficient matrix that fused the RNA information.
[0019] The original Hi-C data were KR normalized and smoothed using downsampling rates of 1 / 25 and 1 / 100. The downsampled Hi-C data were then completed using the FAN method to obtain a complete Hi-C matrix.
[0020] Optionally, in step S3, the model input includes two parts: a DNA-DNA feature correlation coefficient matrix fused with RNA information and a complete Hi-C matrix;
[0021] The sample division is performed in the following way: the DNA-DNA feature correlation coefficient matrix of the fused RNA information and the complete Hi-C matrix are divided into n K×K dimensional sub-matrices as input samples;
[0022] The input samples are divided into training set, validation set and test set for model training, verification and evaluation respectively.
[0023] Optionally, in step S3, the completed Hi-C matrix is down-sampled using two DoubleConv modules plus a DownConv layer, up-sampled using two DoubleConv modules plus an UpConv layer, and finally passed through a DoubleConv module to obtain the first result;
[0024] For the DNA-DNA feature correlation coefficient matrix fused with RNA information, the second result was obtained using the DoubleConv module three times;
[0025] After merging the first and second results, the enhanced Hi-C matrix is obtained by performing two DoubleConv operations.
[0026] Optionally, in step S3, the error between the predicted Hi-C matrix and the actual Hi-C matrix is measured by a loss function;
[0027] Combine L1 loss and perceptual loss to optimize model weights to handle data imbalance and improve model performance on sparse data;
[0028] The Adam optimizer is used to train the model, and the generalization ability of the model is improved through cross-validation and parameter tuning.
[0029] Optionally, step S3 further includes:
[0030] The model was applied to data of different RNA types, with preprocessed RNA data as input and the predicted chromatin interaction enhancement matrix, i.e., the enhanced Hi-C matrix, as output. The prediction accuracy and stability of the model were verified by comparison with the actual Hi-C matrix.
[0031] Optionally, in step S4, the downstream analysis includes: Hi-C chromatin loop analysis mediated by different RNA types, CLIP-seq data and ChIP-seq data verification, enrichment of chromatin loops in subcellular nuclear structures and EP sites, the effects of different types of RNA on the hierarchical structure of chromatin, exploring the effects of RNA on chromatin conformation through model attribution, exploring the effects of RNA on TAD boundaries and chromatin loop anchors through model attribution, and visual analysis of RNA and chromatin high-order structure.
[0032] In the downstream analysis, the Hi-C chromatin loop analysis mediated by different RNA types includes: identifying chromatin loops using the mustache tool on the original Hi-C data and the predicted Hi-C data, and performing veen analysis on the chromatin loops of the original data and the predicted data;
[0033] Validation of CLIP-seq and ChIP-seq data involves: using CLIP-seq data to assess whether RBPs bound to nascent RNAs specifically colocalize near chromatin loop anchors mediated by different types of RNA; and combining public ChIP-seq data to analyze whether more chromatin regulatory factors are enriched at the chromatin loop anchors in the predicted results.
[0034] The enrichment of chromatin loops in subcellular nuclear structures and EP sites includes: calculating the enrichment of Hi-C data predicted by different types of RNA in subcellular nuclear structures and EP sites, revealing that lncRNAs frequently participate in the construction of promoter-enhancer interactions, and the roles of snRNAs and snoRNAs in nucleoli and nuclear speckles;
[0035] The effects of different types of RNA on the hierarchical structure of chromatin include: calculating the effects of different types of RNA on chromatin loops, topological domains and AB compartments of Hi-C data, revealing that lncRNA is involved in local fine regulation, while snRNA / snoRNA has advantages in long-range interactions and larger-scale structural remodeling.
[0036] The downstream analysis explored the impact of RNA on chromatin conformation through model attribution, including: calculating the contribution characteristics of lncRNA, mRNA, snRNA, and snoRNA at TAD boundary regions and chromatin loop anchors, revealing the promoting role of RNA in chromatin loop structure, as well as negative or neutral regulation at TAD boundary regions, indicating that the effect of RNA on chromatin conformation is scale-dependent, and is more likely to mediate small-scale, specific chromatin loop structures;
[0037] Exploring the impact of RNA on TAD boundaries and chromatin loop anchors through model attribution includes: calculating the contribution characteristics of lncRNA, mRNA, snRNA and snoRNA in TAD boundary regions and chromatin loop anchors, and displaying the interaction distribution map containing position information of chromatin interactions at a certain site that are affected by RNA and not affected by RNA.
[0038] The beneficial effects brought about by the technical solution provided by the present invention include at least:
[0039] (1) Realize the effective integration of RNA information into chromatin interaction modeling: This invention constructs a DNA-DNA feature correlation coefficient matrix that integrates RNA-DNA interaction features, regards RNA as an "endogenous factor" for chromatin conformation regulation, and for the first time transforms the asymmetric RNA-DNA interaction matrix system into a symmetric matrix that can be embedded in chromatin analysis, overcoming the problem that RNA positioning information cannot directly reflect the DNA-DNA interaction relationship.
[0040] (2) Integrating multimodal omics data to improve the spatial resolution of Hi-C maps: By introducing a deep learning model based on a dual-channel convolutional neural network and combining RNA-DNA interaction information with the Hi-C contact matrix, the interaction signal of Hi-C data is effectively enhanced, especially in the context of low-resolution or sparse data, showing stronger generalization ability and prediction accuracy, thereby improving the modeling accuracy of the spatial structure of chromatin.
[0041] (3) It has good generalization ability across RNA types: The model shows stable prediction performance on multiple RNA types (such as lncRNA, mRNA, snRNA, snoRNA), can analyze the regulatory characteristics of different RNAs in chromatin structure, and capture the differences in their functions in different hierarchical structures (such as TAD, Loop, Compartment), providing a theoretical basis for revealing the diversity of RNA regulatory mechanisms.
[0042] (4) Realize model attribution and support quantitative functional analysis: The attribution analysis module in the present invention can track the source contribution of the model to the prediction of specific chromatin structure, thereby realizing the quantitative evaluation of the regulatory effects of different RNAs in key areas such as TAD boundaries and chromatin loop anchors, revealing the dominant role of RNA in promoting the formation and stability of chromatin loops, as well as the potential mechanism of certain negative regulatory effects at TAD boundaries, reflecting the unique advantages of this method in functional attribution and scale-dependent analysis.
[0043] (5) Support visualization and cross-omics verification to improve biological interpretability: Combine CLIP-seq, ChIP-seq data and subcellular nuclear structure annotation information to verify whether RNA-mediated chromatin loop anchors are enriched with RNA-binding proteins and epigenetic marks; further visualize the spatial enrichment characteristics of different RNA types in chromatin regulation, so that the prediction results have higher biological interpretability and verifiability.
[0044] (6) Promote the refined development of chromatin three-dimensional structure and function research: This invention not only provides the ability to predict and enhance Hi-C maps, but also clarifies the regulatory patterns of RNA in promoter-enhancer interactions, nucleolus / nuclear speckle enrichment, and long-distance chromatin association through downstream analysis, providing key tools and methodological support for exploring the coupling relationship between RNA regulation of chromatin structure and function.
[0045] (7) Wide applicability and scalability: The method of the present invention is applicable to RNA-DNA interaction and Hi-C data in different cell types and multiple public databases. It has good versatility and can be expanded to various bioinformatics application scenarios such as epigenetic state prediction, regulatory element function identification, and RNA molecule function annotation. BRIEF DESCRIPTION OF THE DRAWINGS
[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0047] Figure 1 Schematic diagram of a method for enhancing chromatin interaction information based on RNA-DNA and DNA-DNA interaction information provided by an embodiment of the present invention;
[0048] Figure 2 Schematic diagram of correlation coefficient matrix calculation provided by an embodiment of the present invention;
[0049] Figure 3 This is a sample division diagram provided by an embodiment of the present invention;
[0050] Figure 4 This is a schematic diagram of the network model structure provided by an embodiment of the present invention;
[0051] Figure 5 This is a graph showing the variation of the model loss function with the number of training rounds provided by an embodiment of the present invention;
[0052] Figure 6 Schematic diagram of the overlap between the Hi-C prediction results of four RNA types provided in the embodiment of the present invention and the original data of chromatin loops;
[0053] Figure 7 This is a heat map of co-localization clustering of ChIP-seq data of different epigenetic signals at chromatin loop anchor points in the original data and four types of RNA predicted Hi-C data provided by the embodiment of the present invention;
[0054] Figure 8 This is a heat map of co-localization clustering of different CLIP-seq data at chromatin loop anchor points in original Hi-C data and four types of RNA predicted Hi-C data provided by the embodiments of the present invention;
[0055] Figure 9 Schematic diagram of subcellular nuclear structure enrichment and EP enrichment provided by an embodiment of the present invention;
[0056] Figure 10 Schematic diagram of Hi-C data distribution characteristics of different RNA types at different scales provided by an embodiment of the present invention;
[0057] Figure 11 This is a schematic diagram of the effect of RNA on TAD boundaries provided by an embodiment of the present invention;
[0058] Figure 12 Schematic diagram of the effect of RNA on chromatin loop anchor points provided by an embodiment of the present invention;
[0059] Figure 13 Schematic diagram of the positional distribution of the effects of RNA on TAD boundary interactions provided by an embodiment of the present invention;
[0060] Figure 14 Schematic diagram of the position distribution of the influence of RNA on chromatin loop anchor points provided by an embodiment of the present invention;
[0061] Figure 15 This is a multi-omics visualization of the effect of lncRNA on the SMYD3 locus provided by an embodiment of the present invention: a schematic diagram of ChIP-seq co-localization;
[0062] Figure 16 This is a multi-omics visualization of the effect of lncRNA on the SMYD3 locus provided by an embodiment of the present invention: a schematic diagram of CLIP-seq co-localization. DETAILED DESCRIPTION
[0063] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0064] An embodiment of the present invention provides a method for enhancing chromatin interaction information based on RNA-DNA and DNA-DNA interaction information. This method implements a deep learning-based prediction model (HiClip model) that can infer RNA-affected Hi-C maps from RNA-DNA interaction data and Hi-C data.
[0065] refer to Figure 1 As shown, the method includes the following steps:
[0066] S1. Collect iMARGI data of RNA-DNA interactions and raw Hi-C data of DNA-DNA interactions from the same cell line.
[0067] In this step, iMARGI data and raw Hi-C data can be obtained from public databases or experimental data, and the acquired raw data can be quality controlled to remove noisy data and low-quality data.
[0068] S2. After feature extraction of iMARGI data, the Pearson correlation coefficient is calculated to obtain a DNA-DNA feature correlation coefficient matrix that integrates RNA-DNA interaction information for the entire genome. The original Hi-C data is downsampled and then FAN-completed to obtain a complete Hi-C matrix.
[0069] Specifically, referring to Figure 2, feature extraction is performed on the RNA-DNA interaction data (iMARGI data), and the Pearson correlation coefficient of the extracted features is calculated to obtain a DNA-DNA feature correlation coefficient matrix that integrates RNA information.
[0070] The original Hi-C data were KR normalized to eliminate technical noise and batch effects. The KR-normalized Hi-C data were smoothed using downsampling rates of 1 / 25 and 1 / 100 to ensure that sufficient features were captured from the low-resolution matrix. The downsampled Hi-C data were completed using the FAN method to obtain a complete Hi-C matrix.
[0071] It should be noted that the Hi-C data, Hi-C matrix, and Hi-C map mentioned in the embodiments of the present invention can sometimes be used interchangeably. When the distinction between them is not emphasized, the meanings they intend to express are the same.
[0072] S3. Construct a CNN network model, use the DNA-DNA feature correlation coefficient matrix fused with RNA information and the complete Hi-C matrix as the model input, train the model, and output the enhanced Hi-C matrix.
[0073] The input of the CNN network model includes two parts: a DNA-DNA feature correlation coefficient matrix fused with RNA information and a complete Hi-C matrix.
[0074] The sample division is performed in the following way: the DNA-DNA feature correlation coefficient matrix of the fused RNA information and the complete Hi-C matrix are divided into n K×K dimensional sub-matrices as input samples, and the reference Figure 3 As shown in Figure 2, the input samples are divided into training set, validation set and test set for model training, validation and evaluation respectively.
[0075] The structure of the model is as follows Figure 4As shown in the figure, for the complete Hi-C matrix, two DoubleConv modules plus DownConv layers are used for downsampling, two DoubleConv modules plus UpConv layers are used for upsampling, and finally a DoubleConv module is used to obtain the first result c5; for the DNA-DNA feature correlation coefficient matrix that fused RNA information, three DoubleConv modules are used to obtain the second result e3; after merging the first result c5 and the second result e3, two more DoubleConvs are performed to obtain the final enhanced Hi-C matrix.
[0076] The CNN network model constructed above is trained, and the error between the predicted Hi-C matrix and the actual Hi-C matrix is measured by the loss function.
[0077] Specifically, the L1 loss and perceptual loss are combined to optimize the model weights to deal with the data imbalance problem and improve the performance of the model on sparse data; the Adam optimizer is used to train the model, and the generalization ability of the model is improved through cross-validation and parameter tuning.
[0078] The prediction and validation of the model involves applying the model to data from different RNA types, with preprocessed RNA data as input and the predicted chromatin interaction enhancement matrix, i.e., the enhanced Hi-C matrix, as output; and verifying the model's prediction accuracy and stability by comparing it with the actual Hi-C matrix.
[0079] Among them, the experimental results on different RNA types are as follows Figure 5 As shown in Figure 3, cross-validation and multiple experiments can ensure that the model has good stability and accuracy.
[0080] S4. For the trained model, input data of different RNA types and predict the corresponding (enhanced) Hi-C matrix for downstream analysis.
[0081] In the embodiments of the present invention, downstream analysis includes: Hi-C chromatin loop analysis mediated by different RNA types, CLIP-seq data and ChIP-seq data verification, enrichment of chromatin loops in subcellular nuclear structures and EP sites, the influence of different types of RNA on the hierarchical structure of chromatin, exploration of the influence of RNA on chromatin conformation through model attribution, exploration of the influence of RNA on TAD boundaries and chromatin loop anchors through model attribution, and visualization analysis of RNA and chromatin high-order structure.
[0082] Among them, the Hi-C chromatin loop analysis mediated by different RNA types includes: using the mustache tool to identify chromatin loops on the original Hi-C data and the predicted Hi-C data, and performing veen analysis on the chromatin loops of the original data and the chromatin loops of the predicted data. The results are as follows Figure 6 shown.
[0083] CLIP-seq data and ChIP-seq data validation include: using CLIP-seq data to evaluate whether RBPs that bind to nascent RNAs are specifically co-localized near chromatin loop anchors mediated by different types of RNA; combining public ChIP-seq data to analyze whether more chromatin regulatory factors (such as transcription factors and histone modification marks) are enriched at the chromatin loop anchors in the predicted results to further verify the accuracy and reliability of the model, such as Figure 7 and Figure 8 shown.
[0084] The enrichment of chromatin loops in subcellular nuclear structures and EP sites includes: calculating the enrichment of Hi-C data predicted by different types of RNA in subcellular nuclear structures and EP sites, revealing that lncRNAs frequently participate in the construction of promoter-enhancer interactions, and the important roles of snRNAs and snoRNAs in nucleolus and nuclear speckle regions, such as Figure 9 shown.
[0085] The effects of different types of RNA on the hierarchical structure of chromatin include: calculating the effects of different types of RNA on chromatin loops, topological domains, and AB compartments of Hi-C data, revealing that lncRNA is mainly involved in local fine regulation, while snRNA / snoRNA has obvious advantages in long-range interactions and larger-scale structural remodeling, such as Figure 10 shown.
[0086] The present invention also designs a model attribution module. Specifically, the model attribution is used to explore the influence of RNA on chromatin conformation, including: calculating the contribution characteristics of lncRNA, mRNA, snRNA and snoRNA in TAD boundary regions and chromatin loop anchors, revealing the promoting role of RNA in chromatin loop structure, and the existence of certain negative or neutral regulation in TAD boundary regions, indicating that the effect of RNA on chromatin conformation is scale-dependent and tends to mediate small-scale, specific chromatin loop structures, such as Figure 11 and Figure 12 shown.
[0087] The influence of RNA on TAD boundaries and chromatin loop anchors is explored through model attribution, including: calculating the contribution characteristics of lncRNA, mRNA, snRNA and snoRNA in TAD boundary regions and chromatin loop anchors, and displaying the interaction distribution map containing position information of chromatin interactions at a certain site affected by RNA and not affected by RNA. Figure 13 and Figure 14 As shown, there are more locations where TAD boundaries are not affected by RNA, and they are distributed at the lower left of the chromatin interaction site and the internal genomic position of the chromatin interaction, while the chromatin loop anchor points are less affected by RNA.
[0088] Furthermore, in the visualization analysis of RNA and chromatin high-order structure, taking the SMYD3 gene as an example, the promotion effect of lncRNA on specific chromatin sites and its possible molecular mechanism were demonstrated, further confirming that lncRNA actively regulates the three-dimensional structure of chromatin by mediating promoter-enhancer and CTCF-dependent chromatin interactions; in the transcription factory region, lncRNA does not participate in the formation of obvious chromatin interactions, but may participate in the post-transcriptional processing and modification of nascent RNA by aggregating RBPs, such as Figure 15 and Figure 16 shown.
[0089] In summary, the present invention provides a method for enhancing chromatin interaction information based on RNA-DNA and DNA-DNA interaction information. The core of the method is to introduce RNA as an endogenous influencing factor regulating the spatial structure of chromatin into the chromatin interaction modeling process, and accurately predict and enhance the Hi-C map through a deep learning model, thereby analyzing the specific regulatory effect of RNA on multi-scale chromatin hierarchical structure (such as chromatin loops, TADs, compartments, etc.).
[0090] Compared with the prior art, the method of the present invention has the following advantages:
[0091] The present invention constructs a DNA-DNA feature correlation coefficient matrix that integrates RNA-DNA interaction features, regards RNA as an "endogenous factor" regulating chromatin conformation, and for the first time converts the asymmetric RNA-DNA interaction matrix system into a symmetric matrix that can be embedded in chromatin analysis, overcoming the problem that RNA positioning information cannot directly reflect the DNA-DNA interaction relationship.
[0092] This paper introduces a deep learning model based on a dual-channel convolutional neural network, combines RNA-DNA interaction information with the Hi-C contact matrix, and effectively enhances the interaction signal of Hi-C data. In particular, it shows stronger generalization ability and prediction accuracy in the context of low-resolution or sparse data, thereby improving the modeling accuracy of chromatin spatial structure.
[0093] The prediction model of the present invention exhibits stable prediction performance on multiple RNA types (such as lncRNA, mRNA, snRNA, and snoRNA). It can analyze the regulatory characteristics of different RNAs in chromatin structure and capture the differences in their functions in different hierarchical structures (such as TAD, Loop, and Compartment), providing a theoretical basis for revealing the diversity of RNA regulatory mechanisms.
[0094] The present invention designs an attribution analysis module that can track the source contribution of the model to the prediction of specific chromatin structure, thereby realizing the quantitative evaluation of the regulatory effects of different RNAs at key regions such as TAD boundaries and chromatin loop anchors. It reveals the dominant role of RNA in promoting the formation and stability of chromatin loops, as well as the potential mechanism of certain negative regulatory effects at TAD boundaries, reflecting the unique advantages of this method in functional attribution and scale-dependent analysis.
[0095] The present invention combines CLIP-seq, ChIP-seq data and subcellular nuclear structure annotation information to verify whether RNA-mediated chromatin loop anchors are enriched with RNA-binding proteins and epigenetic marks; it further visualizes the spatial enrichment characteristics of different RNA types in chromatin regulation, making the prediction results more biologically interpretable and verifiable.
[0096] The present invention not only provides the ability to predict and enhance Hi-C maps, but also clarifies the regulatory patterns of RNA in promoter-enhancer interactions, nucleolar / nuclear speckle enrichment, and long-distance chromatin association through downstream analysis, providing key tools and methodological support for exploring the coupling relationship between RNA regulation of chromatin structure and function.
[0097] The present invention is applicable to RNA-DNA interaction and Hi-C data in different cell types and multiple public databases. It has good versatility and can be expanded to various bioinformatics application scenarios such as epigenetic state prediction, regulatory element function identification, and RNA molecule function annotation.
[0098] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or terminal device that includes a series of elements includes not only those elements but also other elements not explicitly listed, or also includes elements inherent to such process, method, article, or terminal device. In the absence of further limitations, an element defined by the phrase "comprises a..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes the element.
[0099] References in the specification to "one embodiment," "an embodiment," "exemplary embodiments," "some embodiments," etc. indicate that the described embodiments may include a particular feature, structure, or characteristic, but not every embodiment necessarily includes that particular feature, structure, or characteristic. In addition, when a particular feature, structure, or characteristic is described in conjunction with an embodiment, it is within the knowledge of persons skilled in the relevant art to implement that feature, structure, or characteristic in conjunction with other embodiments (whether or not explicitly described).
[0100] It should be understood that the term "and / or" as used herein simply describes an association between related objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. A and B can be singular or plural. Furthermore, the character " / " as used herein generally indicates an "or" relationship between the related objects, but it may also indicate an "and / or" relationship. For specific understanding, please refer to the context.
[0101] In this disclosure, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, "at least one of a, b, or c" can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or plural.
[0102] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.
[0103] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of the device or unit, which can be electrical, mechanical or other forms.
[0104] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.
[0105] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
[0106] If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or the portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical disks.
[0107] The present invention encompasses any alternatives, modifications, equivalents, and solutions that fall within the spirit and scope of the present invention. To provide a thorough understanding of the present invention, specific details are described in detail below in connection with the preferred embodiments of the present invention, but those skilled in the art will be able to fully understand the present invention without these detailed descriptions. Furthermore, to avoid unnecessary confusion regarding the essence of the present invention, well-known methods, processes, procedures, components, and circuits have not been described in detail.
[0108] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.
Claims
1. A method for enhancing chromatin interaction information based on RNA-DNA and DNA-DNA interaction information, characterized in that: The following steps are involved: S1. Collect iMARGI data of RNA-DNA interactions and raw Hi-C data of DNA-DNA interactions from the same cell line. S2. Perform feature extraction on the iMARGI data and calculate the Pearson correlation coefficient to obtain a DNA-DNA feature correlation coefficient matrix of the fused RNA information of the whole genome; perform FAN completion processing on the original Hi-C data after downsampling to obtain a complete Hi-C matrix; The step S2 specifically includes: Feature extraction was performed on the RNA-DNA interaction data, and the Pearson correlation coefficient of the extracted features was calculated to obtain the DNA-DNA feature correlation coefficient matrix that fused the RNA information. The original Hi-C data were KR normalized and smoothed using downsampling rates of 1 / 25 and 1 / 100. The downsampled Hi-C data were then completed using the FAN method to obtain a complete Hi-C matrix. S3. Construct a CNN network model, use the DNA-DNA feature correlation coefficient matrix of the fused RNA information and the complete Hi-C matrix as the model input, train the model, and output the enhanced Hi-C matrix; In step S3, the complete Hi-C matrix is downsampled using two DoubleConv modules plus a DownConv layer, upsampled using two DoubleConv modules plus an UpConv layer, and finally passed through a DoubleConv module to obtain a first result; the DNA-DNA feature correlation coefficient matrix fused with RNA information is passed through three DoubleConv modules to obtain a second result; the first result and the second result are merged and then passed through two more DoubleConv modules to obtain the final enhanced Hi-C matrix; S4. For the trained model, input data of different RNA types and predict the corresponding Hi-C matrix for downstream analysis.
2. The method according to claim 1, characterized in that The step S1 specifically includes: The iMARGI data and the original Hi-C data are obtained from public databases or experimental data; quality control is performed on the obtained original data to remove noise data.
3. The method according to claim 1, characterized in that In step S3, the model input includes two parts: the DNA-DNA feature correlation coefficient matrix of the fused RNA information and the complete Hi-C matrix; The sample division is performed in the following way: the DNA-DNA feature correlation coefficient matrix of the fused RNA information and the complete Hi-C matrix are divided into n K×K dimensional sub-matrices as input samples; The input samples are divided into training set, validation set and test set for model training, verification and evaluation respectively.
4. The method according to claim 1, wherein In step S3, the error between the predicted Hi-C matrix and the actual Hi-C matrix is measured by a loss function; Combine L1 loss and perceptual loss to optimize model weights to handle data imbalance and improve model performance on sparse data; The Adam optimizer is used to train the model, and the generalization ability of the model is improved through cross-validation and parameter tuning.
5. The method according to claim 1, wherein The step S3 further comprises: The model was applied to data of different RNA types, with preprocessed RNA data as input and the predicted chromatin interaction enhancement matrix, i.e., the enhanced Hi-C matrix, as output. The prediction accuracy and stability of the model were verified by comparison with the actual Hi-C matrix.
6. The method according to claim 1, characterized in that In step S4, downstream analysis includes: Hi-C chromatin loop analysis mediated by different RNA types, CLIP-seq data and ChIP-seq data verification, enrichment of chromatin loops in subcellular nuclear structures and EP sites, the effects of different types of RNA on chromatin hierarchical structure, exploration of the effects of RNA on chromatin conformation through model attribution, exploration of the effects of RNA on TAD boundaries and chromatin loop anchors through model attribution, and visualization analysis of RNA and chromatin high-order structure.
7. The method according to claim 6, characterized in that In the downstream analysis, the Hi-C chromatin loop analysis mediated by different RNA types includes: identifying chromatin loops using the mustache tool on the original Hi-C data and the predicted Hi-C data, and performing veen analysis on the chromatin loops of the original data and the predicted data; Validation of CLIP-seq and ChIP-seq data involves: using CLIP-seq data to assess whether RBPs bound to nascent RNAs specifically colocalize near chromatin loop anchors mediated by different types of RNA; and combining public ChIP-seq data to analyze whether more chromatin regulatory factors are enriched at the chromatin loop anchors in the predicted results. The enrichment of chromatin loops in subcellular nuclear structures and EP sites includes: calculating the enrichment of Hi-C data predicted by different types of RNA in subcellular nuclear structures and EP sites, revealing that lncRNAs frequently participate in the construction of promoter-enhancer interactions, and the roles of snRNAs and snoRNAs in nucleoli and nuclear speckles; The effects of different types of RNA on the hierarchical structure of chromatin include: calculating the effects of different types of RNA on chromatin loops, topological domains and AB compartments of Hi-C data, revealing that lncRNA is involved in local fine regulation, while snRNA / snoRNA has advantages in long-range interactions and larger-scale structural remodeling.
8. The method according to claim 6, characterized in that The downstream analysis explored the impact of RNA on chromatin conformation through model attribution, including: calculating the contribution characteristics of lncRNA, mRNA, snRNA, and snoRNA at TAD boundary regions and chromatin loop anchors, revealing the promoting role of RNA in chromatin loop structure, as well as negative or neutral regulation at TAD boundary regions, indicating that the effect of RNA on chromatin conformation is scale-dependent, and is more likely to mediate small-scale, specific chromatin loop structures; Exploring the impact of RNA on TAD boundaries and chromatin loop anchors through model attribution includes: calculating the contribution characteristics of lncRNA, mRNA, snRNA and snoRNA in TAD boundary regions and chromatin loop anchors, and displaying the interaction distribution map containing position information of chromatin interactions at a certain site that are affected by RNA and not affected by RNA.
Citation Information
Patent Citations
Single-cell Hi-C map prediction method based on single-cell RNA expression data
CN118645154A
Method for determination of 3D genome architecture with base pair resolution and further uses thereof
US20230032136A1