Cell differentiation trajectory inference method based on deep hyperbolic manifold embedding

Abstract

The present application relates to a cell differentiation trajectory inference method based on deep hyperbolic manifold embedding, comprising: obtaining single-cell RNA sequencing data; introducing hyperbolic space embedding with negative curvature, extracting features of the single-cell RNA sequencing data through a deep feature extraction network, and mapping the extracted high-dimensional features to a low-dimensional hyperbolic space; constructing the transition path of cells from progenitor cells to various differentiated end cell types through trajectory inference in the low-dimensional hyperbolic space, and evaluating key genes and pathways using a sensitivity analysis method. The beneficial effects of the present application are: by introducing hyperbolic space embedding with negative curvature, the model can better capture the hierarchical structure of the data, and enhance the accuracy and consistency of the differentiation trajectory. Moreover, the present application can process high-dimensional single-cell sequencing data, overcome the limitations of traditional methods in high-dimensional cases, and improve the applicability of the model.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of bioinformatics, more specifically, it relates to a cell differentiation trajectory inference method based on deep hyperbolic manifold embedding. BACKGROUND

[0002] With the development of single-cell RNA sequencing technology, more and more cell types and their differentiation processes have been intensively studied. However, due to the high dimensionality and complexity of single-cell data, traditional cell differentiation trajectory inference methods have obvious shortcomings in capturing the hierarchical structure between cells. Existing methods, such as dimensionality reduction methods based on PCA or t-SNE, although can map data to low-dimensional space, but it is difficult to effectively preserve the hierarchical information of data and the global relationship of cell differentiation. In addition, traditional methods are mostly based on Euclidean space, which cannot accurately describe the inherent hierarchical structure characteristics in the process of cell differentiation. Therefore, it is of great significance to propose a deep learning method based on hyperbolic manifold embedding to better represent the hierarchical relationship between cells and infer the differentiation trajectory. SUMMARY

[0003] The purpose of the present application is to overcome the shortcomings of the prior art and propose a cell differentiation trajectory inference method based on deep hyperbolic manifold embedding.

[0004] In a first aspect, a cell differentiation trajectory inference method based on deep hyperbolic manifold embedding is provided, comprising:

[0005] Step 1, obtaining single-cell RNA sequencing data;

[0006] Step 2, introducing hyperbolic space embedding with negative curvature, extracting features of the single-cell RNA sequencing data through a deep feature extraction network, and mapping the extracted high-dimensional features to a low-dimensional hyperbolic space;

[0007] Step 3, constructing the transition path of cells from progenitor cells to each differentiated end cell type through trajectory inference in the low-dimensional hyperbolic space, and evaluating key genes and pathways using sensitivity analysis methods.

[0008] As a preferred, in step 2, the deep feature extraction network is composed of multiple fully connected layers and activation functions; the last layer of the deep feature extraction network is composed of a hyperbolic mapping layer and a hyperbolic activation function.

[0009] As a preferred, step 3 includes:

[0010] Step 3.1, calculating the hyperbolic distance between cells;

[0011] Step 3.2, constructing a cell differentiation tree using hierarchical clustering algorithm according to the hyperbolic distance between cells, and clearly defining the transition path of cells from progenitor cells to each differentiated end cell type.

[0012] Step 3.3, using a sensitivity analysis method, perturbing the trajectory in hyperbolic space, identifying genes and signaling pathways that play a key role in the differentiation process.

[0013] As preferred, step 2 further comprises:

[0014] calculating a reconstruction loss function for keeping the distance of neighboring cells in the low-dimensional hyperbolic space close.

[0015] As preferred, step 2 further comprises:

[0016] calculating a hyperbolic regularization loss function for regularizing the embedding points to avoid them deviating from the boundary of the hyperbolic space.

[0017] In a second aspect, a cell differentiation trajectory inference system based on deep hyperbolic manifold embedding is provided for performing the method of any of the first aspect, comprising:

[0018] an acquisition module for acquiring single-cell RNA sequencing data;

[0019] an extraction module for introducing a hyperbolic space embedding with negative curvature, performing feature extraction on the single-cell RNA sequencing data through a deep feature extraction network, and mapping the extracted high-dimensional features to a low-dimensional hyperbolic space;

[0020] a construction module for constructing a transition path of cells from progenitor cells to various differentiated end cell types through trajectory inference in the low-dimensional hyperbolic space, and evaluating key genes and pathways using a sensitivity analysis method.

[0021] In a third aspect, a computer storage medium is provided, the computer storage medium having stored therein a computer program; the computer program, when executed on a computer, causes the computer to perform the method of any of the first aspect.

[0022] In a fourth aspect, an electronic device is provided, comprising:

[0023] a memory for saving a computer program;

[0024] a processor for executing the computer program to implement the method of any of the first aspect.

[0025] The beneficial effects of the present application are:

[0026] 1. The present application introduces a hyperbolic space embedding with negative curvature, and the model can better capture the hierarchical structure of the data, enhancing the accuracy and consistency of the differentiation trajectory.

[0027] 2. The application can process high-dimensional single-cell sequencing data, overcome the limitations of traditional methods in high-dimensional cases, and improve the applicability of the model.

[0028] 3. The application combines deep learning technology, can quickly extract nonlinear features, improve data processing efficiency, and shorten the research cycle.

[0029] 4. The application is suitable for various single-cell biology research and medical applications, such as disease mechanism research and new drug development, and has wide practical application value. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 is a general flowchart of a cell differentiation trajectory inference method based on deep hyperbolic manifold embedding;

[0031] Figure 2 is a general framework diagram of a cell differentiation trajectory inference method based on deep hyperbolic manifold embedding;

[0032] Figure 3 is a cell differentiation tree schematic diagram provided by the embodiment of the application;

[0033] Figure 4 is a structure schematic diagram of a cell differentiation trajectory inference system based on deep hyperbolic manifold embedding. DETAILED DESCRIPTION

[0034] The application will be further described below in conjunction with the embodiments. The following description of the embodiments is only to help understand the application. It should be noted that for ordinary people in the technical field, without departing from the principles of the application, the application can be modified, and these improvements and modifications also fall within the protection scope of the claims of the application.

[0035] Embodiment 1:

[0036] In order to solve the limitations of existing methods in cell differentiation trajectory inference, such as Figure 1 As shown, the application provides a cell differentiation trajectory inference method based on deep hyperbolic manifold embedding, comprising:

[0037] Step 1, obtaining single-cell RNA sequencing data.

[0038] Exemplarily, the single-cell RNA sequencing data is denoted as , wherein is the number of cells, is the number of genes.

[0039] Step 2, introducing hyperbolic space embedding with negative curvature, extracting features of the single-cell RNA sequencing data through a deep feature extraction network, and mapping the extracted high-dimensional features to a low-dimensional hyperbolic space.

[0040] Specifically, the deep feature extraction network is composed of multiple fully connected layers and activation functions. The nonlinear transformation of each layer can be represented as:

[0041]

[0042] where, and are the weight matrix and bias vector of the th layer, is the activation function, is the output feature of the th layer. The last layer is a hyperbolic mapping layer, which maps high-dimensional features to hyperbolic space using the hyperbolic activation function to ensure the effectiveness of the mapping:

[0043]

[0044] where, is the representation of the th cell in hyperbolic space.

[0045] In addition, the introduction of hyperbolic space embedding with negative curvature enables the model to better capture the hierarchical structure of the data, enhancing the accuracy and consistency of the differentiation trajectories.

[0046] Specifically, features are extracted through a deep neural network, and high-dimensional features are mapped to a low-dimensional hyperbolic space, denoted as where is the dimension of the embedding space:

[0047]

[0048] where, denotes a deep neural network controlled by parameters is the embedding representation in hyperbolic space. Step 3, through trajectory inference in low-dimensional hyperbolic space, construct the transition path of cells from progenitor cells to various differentiated end cell types, and use sensitivity analysis method to evaluate key genes and pathways.

[0049] Trajectory inference refers to inferring the continuous change path in dynamic biological processes (such as cell differentiation, development, or disease progression) through algorithms. It is based on single-cell data, and by constructing similarity relationships between cells, it arranges the sequence of changes in cells over time or state, thereby reconstructing the dynamic trajectory from the initial state to the final state.

[0050]

[0051] ​Specifically, the application constructs a cell differentiation trajectory inference module to perform step 4, which includes a trajectory inference unit, a differentiation tree construction unit, and a key gene and pathway identification unit.

[0052] Step 3 includes:

[0053] Step 3.1, calculating the hyperbolic distance between cells by the trajectory inference unit.

[0054] The calculation formula of step 3.1 is:

[0055]

[0056] wherein, represents the hyperbolic distance between cell and cell , represents the Euclidean norm of the vector.

[0057] Step 3.2, according to the hyperbolic distance between cells, the differentiation tree construction unit uses a hierarchical clustering algorithm to construct a cell differentiation tree, clearly showing the transition path of cells from progenitor cells to various differentiated end cell types. In this way, the hierarchical structure in the cell differentiation process can be clearly revealed.

[0058] Step 3.3, using a sensitivity analysis method, the key gene and pathway identification unit perturbs the trajectory in the hyperbolic space, identifying genes and signal pathways that play a key role in the differentiation process.

[0059] The calculation formula of step 3.3 is:

[0060]

[0061] wherein, represents the sensitivity of gene to the differentiation trajectory (i.e., the dynamic transition path of cells from progenitor cells to end states). By calculating the sensitivity, it can be determined which genes play an important role in the cell differentiation process.

[0062] Pathway sensitivity analysis: for a signal pathway (containing multiple genes , ,..., ), the sensitivity of all genes in the pathway is comprehensively evaluated to assess its contribution to the trajectory :

[0063] wherein, reflects the overall influence of the signal pathway on the differentiation trajectory.

[0064] Embodiment 2:

[0065] Based on embodiment 1, the application embodiment 2 provides a more specific cell differentiation trajectory inference method based on deep hyperbolic manifold embedding, comprising:

[0066] Step 1, obtaining single-cell RNA sequencing data.

[0067] Step 2, introducing hyperbolic space embedding of negative curvature, extracting features of the single-cell RNA sequencing data through a deep feature extraction network, and mapping the extracted high-dimensional features to a low-dimensional hyperbolic space.

[0068] In step 2, the application embodiment also introduces a loss function module of hyperbolic embedding, which is used to ensure that high-dimensional single-cell data can be effectively mapped to a low-dimensional hyperbolic space and retain the hierarchical relationship and proximity structure of the data; the loss function module includes a reconstruction loss unit and a hyperbolic regularization loss unit.

[0069] Specifically, the reconstruction loss unit is used to calculate the reconstruction loss function, so that the distance between adjacent cells in the hyperbolic space remains as close as possible, which is achieved by the following way:

[0070]

[0071] wherein, denotes the weight between cell and cell , is the distance in the hyperbolic space, is the distance in the original space.

[0072] The hyperbolic regularization loss unit is used to calculate the hyperbolic regularization loss function to ensure the stability of the data in the hyperbolic space and the effectiveness of the embedding. The hyperbolic regularization loss function needs to regularize the embedding points to avoid their deviation from the boundary of the hyperbolic space; which is achieved by the following way:

[0073]

[0074] This loss term is used to constrain the length of the embedding point to be close to 1, so as to ensure that they are located within a reasonable range of the hyperbolic space.

[0075] The total loss function is the weighted sum of the above loss functions:

[0076]

[0077] wherein, , is the weight coefficient of the loss term, which is used to balance the influence of each part of the loss.

[0078] Step 3: Construct the transformation pathways of cells from progenitor cells to various terminal cell types by inferring trajectories in low-dimensional hyperbolic space, and use sensitivity analysis to evaluate key genes and pathways.

[0079] In summary, this invention provides an efficient and accurate method for inferring cell differentiation trajectories, which can effectively process high-dimensional single-cell data, capture the hierarchical structure and key genes in the cell differentiation process, and provide a powerful tool for cell biology research.

[0080] For example, in studying embryonic development, single-cell RNA sequencing data from different developmental stages can be input into the method of this invention. Cellular features are extracted using a deep feature extraction network and embedded into hyperbolic space. Then, differentiation trajectories between cells are inferred based on hyperbolic distance, constructing a differentiation tree of embryonic cells and clarifying the differentiation pathways of different cell types during development. Simultaneously, sensitivity analysis identifies genes and signaling pathways that play key roles in differentiation, providing important clues for revealing the molecular mechanisms of embryonic development.

[0081] For example, a single-cell RNA sequencing dataset of zebrafish larvae was used as input data. This dataset, derived from 5-day-old zebrafish larvae, comprises over 70,000 cells that underwent transcriptome sequencing after isolation and single-cell capture. Single-cell capture was performed using the 10X Genomics Chromium system combined with Droplet Microfluidics technology. Each cell was encapsulated in a microdroplet, lysed, and reverse transcribed to generate cDNA with cell-specific barcodes and unique molecular identifiers (UMIs).

[0082] like Figure 3 As shown, Figure 3 This is a cell differentiation tree generated on a single-cell RNA sequencing dataset of zebrafish larvae using the method of this invention. The scale bar in the lower left corner represents the ratio of branch length to differentiation distance.

[0083] It should be noted that the parts in this embodiment that are the same as or similar to those in Embodiment 1 can be referred to each other, and will not be repeated in this application.

[0084] Example 3:

[0085] Building upon Examples 1 and 2, Example 3 of this application provides a cell differentiation trajectory inference system based on deep hyperbolic manifold embedding, such as... Figure 4 As shown, it includes:

[0086] The acquisition module is used to acquire single-cell RNA sequencing data;

[0087] An extraction module is configured to introduce hyperbolic space embedding with negative curvature, extract features from the single-cell RNA sequencing data by a deep feature extraction network, and map the extracted high-dimensional features to a low-dimensional hyperbolic space;

[0088] A construction module is configured to construct transition paths of cells from progenitor cells to various differentiated end cell types by trajectory inference in the low-dimensional hyperbolic space, and evaluate key genes and pathways by using a sensitivity analysis method.

[0089] Specifically, the system provided in the embodiment corresponds to the method provided in embodiments 1 and 2, and thus, the same or similar parts in the embodiments can be referred to each other, and will not be described herein again.

Claims

1. A method for inferring cell differentiation trajectories based on deep hyperbolic manifold embedding, characterized in that, include: Step 1: Obtain single-cell RNA sequencing data; Step 2: Introduce hyperbolic space embedding with negative curvature, extract features from the single-cell RNA sequencing data using a deep feature extraction network, and map the extracted high-dimensional features to a low-dimensional hyperbolic space; in Step 2, the deep feature extraction network consists of multiple fully connected layers and activation functions; the last layer of the deep feature extraction network consists of a hyperbolic mapping layer and a hyperbolic activation function; Step 3: Construct the transformation pathways of cells from progenitor cells to various terminal cell types by inferring trajectories in low-dimensional hyperbolic space, and use sensitivity analysis to evaluate key genes and pathways. Step 3 includes: Step 3.1: Calculate the hyperbolic distance between cells; Step 3.2: Based on the hyperbolic distance between cells, construct a cell differentiation tree using a hierarchical clustering algorithm to clarify the transformation path of cells from progenitor cells to various terminal cell types; Step 3.3: Using sensitivity analysis, the trajectories in hyperbolic space are perturbed to identify genes and signaling pathways that play a key role in the differentiation process; The calculation formula for step 3.3 is as follows: in, Indicates gene On the differentiation trajectory Sensitivity.

2. The cell differentiation trajectory inference method based on deep hyperbolic manifold embedding according to claim 1, characterized in that, Step 2 also includes: Calculate a reconstruction loss function, which is used to keep the distance between neighboring cells in the low-dimensional hyperbolic space close.

3. The method for inferring cell differentiation trajectories by deep hyperbolic manifold embedding according to claim 2, characterized in that, Step 2 also includes: Calculate the hyperbolic regularization loss function, which is used to regularize the embedding points to prevent them from deviating from the boundary of the hyperbolic space.

4. A cell differentiation trajectory inference system based on deep hyperbolic manifold embedding, characterized in that, For performing the method according to any one of claims 1 to 3, comprising: The acquisition module is used to acquire single-cell RNA sequencing data; The extraction module is used to introduce hyperbolic space embedding with negative curvature, extract features from the single-cell RNA sequencing data through a deep feature extraction network, and map the extracted high-dimensional features to a low-dimensional hyperbolic space. The module is designed to construct the transition pathways of cells from progenitor cells to various terminal cell types by inferring trajectories in low-dimensional hyperbolic space, and to evaluate key genes and pathways using sensitivity analysis methods.

5. A computer storage medium, characterized in that, The computer storage medium stores a computer program; when the computer program is run on the computer, it causes the computer to perform the method described in any one of claims 1 to 3.

6. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the method as described in any one of claims 1 to 3.

Images