Method and apparatus for multi-dimensional evaluation of endometrial function

Abstract

This invention provides a method and apparatus for multidimensional assessment of endometrial function. The method is based on endometrial transcriptome sequencing (Bulk RNA-seq) data, utilizing a reliable single-cell reference feature matrix to analyze cell composition and assessing epithelial cell composition through gene module scoring. In the Bulk RNA-seq data, the average expression level of each feature is calculated based on a priori set of immune and aging-related feature genes, and then standardized to obtain immune indicators and aging scores. The method and apparatus of this invention can assess endometrial function from three dimensions: endometrial cell composition, immune status, and aging level.

Description

Technical Field

[0001] This invention relates to a method and apparatus for assessing endometrial function, specifically a method and apparatus for assessing multidimensional endometrial function based on Bulk RNA sequencing of cell composition and immune-senescence, belonging to the field of assisted reproductive technology. Background Technology

[0002] Assisted reproductive technology (ART) has seen rapid development in recent decades to help infertile patients achieve pregnancy. However, the clinical pregnancy rate fluctuates only around 50%. While ovulation induction and embryo culture techniques have become increasingly sophisticated, and breakthroughs in preimplantation genetic testing (PGT) have ensured the availability of high-quality embryos through screening and selection, some patients still fail to achieve pregnancy even with high-quality embryo transfer. Successful pregnancy depends on embryo quality, endometrial receptivity, and the synergy between the embryo and the endometrium. Given the relative stability of embryo retrieval techniques, objectively quantifying the endometrial condition and using this information to guide embryo transfer has become a pressing clinical challenge.

[0003] Currently, clinical practice involves monitoring the uterus and its endometrial morphology to assess endometrial status. This includes observing endometrial thickness, morphology, and blood supply via ultrasound. Appropriate endometrial preparation, good endometrial thickness, and blood flow provide a favorable implantation environment for the embryo, while poor endometrial condition may prevent successful implantation. Endometrial thickness reflects the functional status of the endometrium and predicts endometrial receptivity. However, the relationship between endometrial thickness and ART pregnancy outcomes remains unclear. With ongoing clinical research, increasing evidence suggests that endometrial thickness alone has no predictive value for ART pregnancy outcomes. Some studies report that endometrial morphology classification and hemodynamics can also assess endometrial receptivity and predict pregnancy outcomes. In clinical ART, most centers use a multi-indicator approach. Despite this, some patients still experience recurrent implantation failure (RIF). These assessment or prediction methods suffer from the following problems: insufficient quantitative data leads to inaccurate results; and subjective judgments vary among examiners. With the rapid development of molecular biology and high-throughput sequencing technology, researchers have gradually turned their attention to the more microscopic world, and the receptivity of the endometrium has received attention from the perspective of cellular composition and transcriptome level.

[0004] At the mechanistic level, non-ciliated epithelium in the mid-secretory phase (including tubular, glandular, and secretory epithelium) directly participates in the process of embryo adhesion and invasion, and transmits paracrine signals to the mesenchyme to initiate decidualization. Insufficient proportions of this epithelial subset are closely related to microenvironmental dysregulation and implantation failure. Simultaneously, a balanced endometrial immune environment is also crucial for successful pregnancy. In endometrial tissue, immunosuppressive cells that promote angiogenesis and tissue repair (including Treg, Th2, uNK1, and M2) are relatively dominant, contributing to the establishment of local tolerance, vascular remodeling, and tissue repair. When the immune state leans towards pro-inflammatory, it is often associated with early pregnancy loss. Traditional imaging and single gene markers are insufficient to fully reflect these subtle changes in cell subsets and immune balance.

[0005] At the transcriptomic level, the "cellular age" of the endometrium can be assessed, specifically by evaluating cellular age in the following aspects: senescence-associated secretory phenotype (SASP) factors, DNA damage response and cell cycle inhibition, mitochondrial dysfunction and extracellular matrix remodeling gene reprogramming, etc. These changes collectively promote inflammatory senescence and weaken decidualization and receptivity.

[0006] Researchers have already used endometrial transcriptomics to perform endometrial receptivity testing (rsERT). This endometrial receptivity analysis (ERA) can identify a specific window of implantation (WOI) for patients with recurrent implantation failure, and is therefore widely used in clinical practice. However, even after intervention based on the implantation window, some patients still experience pregnancy failure.

[0007] Therefore, there is an urgent need for a quantifiable, reproducible endometrial assessment technique that is compatible with clinical procedures. Summary of the Invention

[0008] One object of the present invention is to provide an improved method for assessing endometrial function.

[0009] Another object of the present invention is to provide an apparatus for assessing endometrial function.

[0010] To meet clinical accessibility and cost constraints, this invention proposes replacing high-cost single-cell sequencing with Bulk RNA-seq deconvolution: cellular composition is analyzed using a reliable single-cell reference feature matrix based on traditional endometrial transcriptome data, and epithelial cell composition is assessed through gene module scoring (see...). Figure 1Meanwhile, in Bulk RNA-seq data, the average expression levels of different features were calculated based on a prior set of immune and aging-related feature genes and then standardized to obtain different immune indicators and aging scores.

[0011] Against this backdrop, this invention proposes an integrated assessment method for endometrial cell composition and immune-aging based on deconvolution and gene module scoring: without increasing the burden of clinical procedures and while remaining compatible with routine sampling processes, it uses BulkRNA-seq as input to directly output objective quantitative indicators reflecting the proportion of non-ciliated epithelium, immune balance, and the degree of accelerated endometrial aging. These indicators are used to interpret endometrial function and provide a basis for individualized intervention and embryo transfer decisions.

[0012] Specifically, on the one hand, the present invention provides a method for assessing endometrial function, the method comprising: Cellular component analysis: Using a deconvolution method, based on endometrial transcriptome sequencing (Bulk RNA-seq) data, and utilizing the endometrial immune cell or tissue-type single-cell feature matrix, the composition of endometrial single cells or the proportion of each cell type in the target sample is analyzed. Gene module scoring: In Bulk RNA-seq data, the average expression level of each feature was calculated based on the feature gene set related to immunity and aging, and then standardized to obtain immune indicators and aging scores respectively; Based on the above-mentioned cellular composition, immune indicators, and aging score, endometrial function was assessed.

[0013] According to some specific embodiments of the present invention, in the method for assessing endometrial function of the present invention, the single-cell composition of the endometrium includes one or more of the following cell types: endometrial ciliated epithelial cells, non-ciliated epithelial cells, mesenchymal cells, fibroblasts, endothelial cells, lymphocytes, etc.

[0014] According to some specific embodiments of the present invention, the immune indicators in the method for assessing endometrial function of the present invention include, but are not limited to, one or more of the following: M1 / M2 immune balance index, uNK1 / uNK3 immune balance index, Th1 / Th2 immune balance index, and Treg / Th17 immune balance index.

[0015] According to some specific embodiments of the present invention, in the method for assessing endometrial function of the present invention, the aging score includes, but is not limited to, scores representing the degree of cellular aging obtained from genes related to pathways such as cellular aging-related secretory phenotypic factors, DNA damage response and cell cycle inhibition, mitochondrial dysfunction and extracellular matrix remodeling gene reprogramming.

[0016] According to some specific embodiments of the present invention, the cellular component analysis process in the method for assessing endometrial function of the present invention includes: Constructing a target single-cell feature expression matrix: The single-cell feature expression matrix is ​​derived from self-tested or publicly available single-cell transcriptome sequencing data of endometrial tissue. Based on this data analysis, the cell types contained in the tissue and the genes specifically expressed by each cell type are identified. Then, a cell-specific gene expression matrix S is constructed, where S is a gene × cell type matrix. The expression data obtained from bulk transcriptome sequencing of endometrial tissue transcriptome nucleic acid (RNA) samples were used to construct a bulk expression matrix B, where B is a gene × sample matrix; The expression levels of multiple characteristic genes involved in matrix S are extracted from matrix B to form the tissue Bulk transcriptome characteristic gene expression matrix E, where E is a gene × sample matrix; Set up a matrix R to calculate the proportion of each single cell, where R is a sample × cell type matrix, and use the deconvolution method to analyze and obtain the cell proportion.

[0017] According to some specific embodiments of the present invention, in the method for assessing endometrial function: A cell-specific gene expression matrix S is defined, with a size of I×J, where I is the number of genes and J is the number of cell types. Rows represent genes, columns represent cell types, and values ​​represent the expression level of a gene in a specific cell type. ij The expression value of the i-th characteristic gene in the j-th cell type is recorded. The bulk expression matrix B consists of rows representing genes, columns representing samples, and values ​​representing gene expression levels. The Bulk transcriptome characteristic gene expression matrix E is of size I×N, where I is the number of characteristic genes and N is the number of samples. in Let the expression value of the i-th characteristic gene in the n-th sample be denoted as ; The matrix R representing the proportion of each single cell type is of size N×J, where N is the number of samples in matrix E, J is the number of single cell types in matrix S, and R... nj The percentage of the j-th cell type in the n-th sample tissue is an unknown value that needs to be solved. S R=E. Solving the equation using the deconvolution method yields R', and the value in R' is the cell proportion obtained from the deconvolution analysis.

[0018] According to some specific embodiments of the present invention, the deconvolution analysis process in the method for assessing endometrial function of the present invention includes: ① Take the expression value of the I characteristic gene of the nth sample from the tissue bulk transcriptome characteristic gene expression matrix E, and let it be the array SampleE. Then:

[0019] ② For the nth sample, the bulk expression value of each gene is equal to the total expression value of multiple single-cell clusters. Thus, each characteristic gene will have an equation. Taking the i-th equation as an example, the formula is as follows:

[0020] ③ For the nth sample, there are I equations in the above equations, forming a system of equations. Support vector regression analysis is then used to deduce the most reliable R. nj The values ​​are used to form the final matrix of the relative content of each endometrial single cell in each sample.

[0021] According to some specific embodiments of the present invention, the method for assessing endometrial function further includes: Based on clinical information of the samples, the correlation between the composition of intima single cells and clinical observation indicators was observed in groups of samples. Setting an appropriate threshold for the percentage of single cells in the endometrium can predict clinical observation indicators.

[0022] According to some specific embodiments of the present invention, the gene module scoring process in the method for assessing endometrial function of the present invention includes: Identify pro-inflammatory / anti-inflammatory immune subsets (Treg, Th17, Th1, Th2, uNK1, uNK3, M1, M2) and the set of transcriptional signature genes associated with cellular senescence; For each sample to be tested, after completing the routine normalization, the mean expression value (mean_i) of each feature gene set, the mean expression value (mean_all) of the whole gene set, and the standard deviation (sd_all) are calculated, and the dimensionless module score is obtained by standardizing according to the in-sample z-squared. Z_i= (mean_i – mean_all) / sd_all Intra-sample z-normalization is performed on the mean of each feature to naturally eliminate the influence of differences in sequencing depth / overall expression amplitude, resulting in dimensionless scores that can be compared across batches. The scores from the above modules are further used to construct the immune balance index and aging score.

[0023] According to some specific embodiments of the present invention, the method for assessing endometrial function of the present invention includes an assessment of three dimensions: endometrial cell composition, immune status, and aging level.

[0024] On the other hand, the present invention also provides an apparatus for evaluating endometrial function, the apparatus comprising a detection unit and a data analysis unit, wherein: The detection unit is used to detect samples from individuals and obtain endometrial transcriptome sequencing data; The data analysis unit is used to analyze and process the obtained endometrial transcriptome sequencing data.

[0025] According to some specific embodiments of the present invention, in the device for assessing endometrial function, the examination unit includes, but is not limited to, any feasible instruments / reagents for transcriptome sequencing in the prior art. In some specific embodiments of the present invention, RNA is extracted from endometrial tissue samples using an RNA extraction kit, amplified and constructed into a library after reverse transcription, and then sequenced to obtain transcriptome sequencing data.

[0026] According to some specific embodiments of the present invention, in the device for assessing endometrial function, the data analysis unit performs analysis and processing, including: The method described in any of the foregoing schemes of this invention is used to analyze the endometrial single-cell composition or the proportion of each type of cell, immune indicators and aging fraction in the target sample to assess endometrial function.

[0027] According to some specific embodiments of the present invention, in the device for assessing endometrial function of the present invention, the data analysis unit includes a cell component analysis module and a gene scoring module, wherein the cell component analysis module is used to analyze the endometrial single cell composition or the proportion of each type of cell in the target sample, and the gene scoring module is used to calculate the average expression level of different features based on the set of immune and aging-related characteristic genes in Bulk RNA-seq data and then standardize it to obtain immune indicators and aging scores of different features.

[0028] According to some specific embodiments of the present invention, in the device for evaluating endometrial function, the data analysis unit includes a cell component analysis module and a gene scoring module, wherein the cell component analysis module includes a target single-cell feature expression matrix construction module, a bulk expression matrix construction module, and a bulk transcriptome feature gene expression matrix E construction module; a deconvolution analysis module, wherein: The target single-cell feature expression matrix construction module is used to construct the cell-specific gene expression matrix S; The bulk representation matrix construction module is used to construct the bulk representation matrix B. The Bulk transcriptome characteristic gene expression matrix E construction module is used to construct the Bulk transcriptome characteristic gene expression matrix E; The deconvolution parsing module is used to parse the cell proportions based on the cell-specific gene expression matrix S and the Bulk transcriptome characteristic gene expression matrix E.

[0029] According to some specific embodiments of the present invention, in the device for evaluating endometrial function of the present invention, the cell component analysis module analyzes the endometrial single-cell composition or the proportion of each type of cell in the target sample in the following manner: Constructing a target single-cell feature expression matrix: The single-cell feature expression matrix is ​​derived from self-tested or publicly available single-cell transcriptome sequencing data of endometrial tissue. Based on this data analysis, the cell types contained in the tissue and the genes specifically expressed by each cell type are identified. Then, a cell-specific gene expression matrix S is constructed, where S is a gene × cell type matrix. Expression data obtained from bulk transcriptome sequencing of endometrial tissue transcriptome nucleic acid (RNA) samples at a specific time period were used to construct a bulk expression matrix B, where B is a gene × sample matrix; The expression levels of multiple characteristic genes involved in matrix S are extracted from matrix B to form the tissue Bulk transcriptome characteristic gene expression matrix E, where E is a gene × sample matrix; Set up a matrix R to calculate the proportion of each single cell, where R is a sample × cell type matrix, and use the deconvolution method to analyze and obtain the cell proportion.

[0030] According to some specific embodiments of the present invention, in the device for assessing endometrial function of the present invention, during the process of the cell component analysis module resolving the endometrial single-cell composition or the proportion of each type of cell in the target sample: A cell-specific gene expression matrix S is defined, with a size of I×J, where I is the number of genes and J is the number of cell types. Rows represent genes, columns represent cell types, and values ​​represent the expression level of a gene in a specific cell type. ij The expression value of the i-th characteristic gene in the j-th cell type is recorded. The bulk expression matrix B consists of rows representing genes, columns representing samples, and values ​​representing gene expression levels. The Bulk transcriptome characteristic gene expression matrix E is of size I×N, where I is the number of characteristic genes and N is the number of samples. in Let the expression value of the i-th characteristic gene in the n-th sample be denoted as ; The matrix R representing the proportion of each single cell type is of size N×J, where N is the number of samples in matrix E, J is the number of single cell types in matrix S, and R... nj The percentage of the j-th cell type in the n-th sample tissue is an unknown value that needs to be solved. S R=E. Solving the equation using the deconvolution method yields R', and the value in R' is the cell proportion obtained from the deconvolution analysis.

[0031] According to some specific embodiments of the present invention, in the device for assessing endometrial function, the process of obtaining the cell ratio by the deconvolution analysis module includes: ① Take the expression value of the I characteristic gene of the nth sample from the tissue bulk transcriptome characteristic gene expression matrix E, and let it be an array SampleE. Then:

[0032] ② For the nth sample, the bulk expression value of each gene is equal to the total expression value of multiple single-cell clusters. Thus, each characteristic gene will have an equation. Taking the i-th equation as an example, the formula is as follows:

[0033] ③ For the nth sample, there are I equations in the above equations, forming a system of equations. Support vector regression analysis can be used to deduce the most reliable R. nj The values ​​are used to form the final matrix of the relative content of each endometrial single cell in each sample.

[0034] According to some specific embodiments of the present invention, the device for assessing endometrial function may further include a clinical observation indicator module, which is used to observe the correlation between the endometrial single-cell composition and clinical observation indicators of samples based on sample clinical information, and to set an appropriate endometrial single-cell percentage threshold to predict clinical observation indicators.

[0035] According to some specific embodiments of the present invention, in the device for assessing endometrial function, the gene scoring module is used to obtain immune indicators and aging scores in the following manner: Identify pro-inflammatory / anti-inflammatory immune subsets (Treg, Th17, Th1, Th2, uNK1, uNK3, M1, M2) and the set of transcriptional signature genes associated with cellular senescence; For each sample to be tested, after completing the routine normalization, the mean expression value (mean_i) of each feature gene set, the mean expression value (mean_all) of the whole gene set, and the standard deviation (sd_all) are calculated, and the dimensionless module score is obtained by standardizing according to the in-sample z-squared. Z_i= (mean_i – mean_all) / sd_all Intra-sample z-normalization is performed on the mean of each feature to naturally eliminate the influence of differences in sequencing depth / overall expression amplitude, resulting in dimensionless scores that can be compared across batches. The scores from the above modules are further used to construct the immune balance index and aging score.

[0036] According to some specific embodiments of the present invention, the immune balance index in the device for assessing endometrial function includes, but is not limited to, one or more of the following: M1 / M2 immune balance index, uNK1 / uNK3 immune balance index, Th1 / Th2 immune balance index, and Treg / Th17 immune balance index. In some specific embodiments of the present invention, the immune balance index can be used to assess clinical pregnancy rate. An immune balance index indicating an excessive pro-inflammatory state is associated with a decreased clinical pregnancy rate.

[0037] According to some specific embodiments of the present invention, in the device for assessing endometrial function, the aging score includes, but is not limited to, scores representing the degree of cellular aging obtained from genes related to pathways such as cellular aging-related secretory phenotypic factors, DNA damage response and cell cycle inhibition, mitochondrial dysfunction, and extracellular matrix remodeling gene reprogramming. In some specific embodiments of the present invention, the aging score can identify individuals with "accelerated endometrial aging" and can be used to assess clinical pregnancy rates. An increased aging score corresponds to a decreased clinical pregnancy rate.

[0038] On the other hand, the present invention also provides a non-transitory computer-readable storage medium storing computer instructions, which, when executed by a processor, achieve the following: using the method described in any of the foregoing schemes of the present invention to parse the endometrial single cell content or the proportion of each type of cell, immune indicators, and aging fraction in the target sample, and further evaluate endometrial function.

[0039] On the other hand, the present invention also provides a computer device, which includes a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to achieve: using the method described in any of the foregoing schemes of the present invention to analyze the endometrial single cell content or the proportion of each type of cell, immune indicators and aging fraction in the target sample, and further evaluate endometrial function.

[0040] The present invention provides a method and apparatus for assessing endometrial function that simultaneously covers three key biological axes: (1) epithelial integrity and subset composition; (2) immune homeostasis / pro-inflammatory imbalance; and (3) endometrial transcriptomic senescence. The present invention proposes replacing costly single-cell sequencing with BulkRNA-seq deconvolution: cellular composition is analyzed using a reliable single-cell reference feature matrix based on traditional endometrial transcriptomic data, and epithelial cell composition is assessed through gene module scoring. Simultaneously, in BulkRNA-seq data, the average expression levels of different features are calculated based on a priori immune and senescence-related feature gene sets, and then standardized to obtain different immune indicators and senescence scores. The technology of the present invention meets clinical accessibility requirements and reduces costs. Attached Figure Description

[0041] Figure 1 This is a schematic diagram of the endometrial single-cell analysis process in this invention.

[0042] Figure 2 This invention presents an example of immune cell analysis results for endometrial tissue. The proportion of each immune cell in the endometrium was analyzed based on a matrix of 22 immune cell features and Bulk RNAseq data. The horizontal axis represents each sample, and the vertical axis represents the relative content of each single cell. Different colors indicate the corresponding immune cell types, as detailed in the legend on the right side of the figure.

[0043] Figure 3 The low percentage of non-ciliated epithelial cells in Example 2 suggests impaired endometrial function.

[0044] Figure 4 The threshold for the proportion of non-ciliated cells is selected in Example 2.

[0045] Figure 5 The results of the independent validation set in Example 2 are displayed.

[0046] Figure 6 The average pregnancy rate in the M1 excessive pro-inflammatory group was significantly lower than that in the macrophage immune balance group.

[0047] Figure 7 The average pregnancy rate in the uNK over-pro-inflammatory group was significantly lower than that in the uNK cell immune balance group.

[0048] Figure 8 The average pregnancy rate in the Th1 over-inflammatory group was significantly lower than that in the T helper cell immune balance group.

[0049] Figure 9 The average pregnancy rate in the Th17 over-pro-inflammatory group was significantly lower than that in the Treg and Th17 cellular immune balance groups.

[0050] Figure 10 This demonstrates the characteristics of endometrial cell senescence in the development dataset of Example 6.

[0051] Figure 11 This shows the characteristics of endometrial cell senescence in the validation dataset of Example 6.

[0052] Figure 12 The study showed that the cGAS-STING pathway was enhanced and the mitochondrial fusion signaling pathway was weakened in the accelerated aging group. Detailed Implementation

[0053] To provide a clearer understanding of the technical features, objectives, and beneficial effects of the present invention, the technical solution of the present invention will now be described in detail with reference to specific embodiments and accompanying drawings. It should be understood that these examples are for illustrative purposes only and are not intended to limit the scope of the invention. Various changes and / or modifications that are readily conceived by those skilled in the art within the spirit and scope of the present invention are considered to be covered within the protection scope of the present invention.

[0054] In the embodiments, all original reagent materials are commercially available, and experimental methods without specific conditions are conventional methods and conditions well known in the field, or according to the conditions recommended by the instrument manufacturer.

[0055] This invention provides an integrated assessment method for endometrial cell composition and immune-senescence based on deconvolution and gene module scoring, replacing costly single-cell sequencing with Bulk RNA-seq deconvolution: Cell composition is analyzed using a reliable single-cell reference feature matrix on traditional endometrial transcriptome data, and epithelial cell composition is assessed through gene module scoring (e.g., ...). Figure 1 (As shown). Meanwhile, in the Bulk RNA-seq data, the average expression levels of different features were calculated based on a prior set of immune and aging-related feature genes and then standardized to obtain different immune indicators and aging scores.

[0056] The specific technical solution is as follows: 1) Required sample: Endometrial tissue sample from a specific period; 2) Cell component analysis: i. Constructing the target single-cell feature expression matrix (one-time construction): The single-cell feature expression matrix is ​​derived from self-tested or publicly available single-cell transcriptome sequencing data of specific tissues. Based on this data analysis, the cell types contained in the tissue and the genes specifically expressed by each cell type are identified. For example, differentially expressed genes between different cell types are identified based on the log value of the differential expression fold change > 2 and FDR < 0.05 (if there are more than 100, the top 50 can be selected). Then, a cell-specific gene expression matrix is ​​constructed, called the feature matrix S (genes × cell types, matrix size is set as I × J, where I is the number of genes and J is the number of cell types, e.g. Figure 1 In the single-cell feature matrix S, the rows represent genes, the columns represent cell types, and the values ​​represent the expression levels of genes in specific cell types. ij The expression value of the i-th characteristic gene in the j-th cell type is recorded. ii. Expression data obtained by bulk transcriptome sequencing of endometrial tissue transcriptome nucleic acid (RNA) samples at a specific time is called bulk expression matrix B (gene × sample). In the matrix, the rows are genes, the columns are samples, and the values ​​are gene expression levels. In practice, the genes in matrix B can be detected using routine RNAseq whole transcriptome sequencing (approximately 20,000 genes), and the samples are the samples to be studied (for example, in Example 1, the samples are endometrial tissue samples from the mid-secretionary phase of the menstrual cycle). iii. Extract the expression levels of the I characteristic genes involved in matrix S from matrix B to form a new matrix E (genes × samples, size I × N, where I is the number of characteristic genes and N is the number of samples, e.g.) Figure 1 Bulk transcriptome characteristic gene expression matrix (E), E in Let the expression value of the i-th characteristic gene in the n-th sample be denoted as ; iv. Define a matrix of unknown variables to be calculated, called the variable matrix R (sample × cell type, size N × J, where N is the number of samples in matrix E, and J is the number of single cell types in matrix S, such as...). Figure 1 The matrix R to be solved for the proportion of each single cell is to be solved. nj The percentage of the j-th cell type in the n-th sample tissue is an unknown value that needs to be solved. v. Based on the above steps, it can be known that S R=E. Solving the equation using deconvolution yields T', where the values ​​in T' represent the cell proportions obtained through deconvolution analysis. The specific steps are as follows: If we extract the expression values ​​of the I characteristic genes of the nth sample from the Bulk transcriptome characteristic gene expression matrix E, and denote the array SampleE, then:

[0057] For the nth sample, the bulk expression value of each gene is equal to the total expression value of multiple single-cell clusters. Thus, each feature gene will have an equation. Taking the i-th equation as an example, the formula is as follows:

[0058] For the nth sample, there are I equations in the above equations, which form a system of equations. The cell percentage can be deduced by using support vector regression analysis. vi. Based on clinical information of the samples, observe the correlation between the composition of single cells in the intima of the samples and clinical observation indicators in groups; vii. Set an appropriate threshold for the percentage of single endometrial cells to predict clinical observation indicators (indicators that are correlated with clinical pregnancy outcomes are appropriate observation indicators); 3) Gene module scoring: Based on prior knowledge derived from existing literature and single-cell evidence, pro-inflammatory / anti-inflammatory immune subsets (Treg, Th17, Th1, Th2, uNK1, uNK3, M1, M2) and sets of transcriptional signature genes related to cellular senescence were identified. For each sample, after routine normalization, the mean expression value (mean_i) of each signature gene set, as well as the mean expression value (mean_all) and standard deviation (sd_all) of the entire gene set, were calculated, and dimensionless module scores were obtained by in-sample z-normalization. Z_i= (mean_i – mean_all) / sd_all Intra-sample z-normalization is performed on the mean of each feature to naturally eliminate the influence of differences in sequencing depth / overall expression amplitude, resulting in a dimensionless score that can be compared across batches. The higher the score, the stronger the feature.

[0059] The module scores mentioned above are further used to construct the immune balance index and aging score. Specifically, the z-normalization method can be used to calculate module scores for different immune cell subset characteristic gene sets and cellular senescence characteristic gene sets.

[0060] This invention has tested over 2000 endometrial samples with clinical information. For each sample, deconvolution algorithm and gene module scoring algorithm were used to evaluate endometrial receptivity in three dimensions: endometrial cell composition, immune status, and aging level.

[0061] Example 1: Assessment of endometrial function by the proportion of unciliated epithelial cells in the endometrium

[0062] This embodiment provides an example of applying the technology of the present invention to assess endometrial function using the proportion of non-ciliated epithelial cells in the endometrium.

[0063] 1) Sample source: Endometrial tissue was collected during the mid-secretionary phase of the menstrual cycle, and clinical pregnancy outcomes of embryo transfer were collected for the corresponding patients; 2) RNA was extracted from endometrial tissue samples, amplified and constructed into a library after reverse transcription (using the Yikang Platinum Kit), and sequenced to obtain the transcriptome expression matrix E, where columns are samples, rows are genes, and values ​​are the TPM expression values ​​of genes. 3) Data quality control requires 2M raw reads, an exon percentage greater than 30%, and all sample test data must meet the quality control conditions; 4) Publicly available data from endometrial single-cell transcriptome sequencing (Wang et al.). Nat Medicine(2020) Analysis confirmed six major types of single cells in the endometrium, including ciliated epithelial cells, non-ciliated epithelial cells, mesenchymal cells, fibroblasts, endothelial cells, and lymphocytes. A single cell characteristic matrix, called the characteristic matrix S, was constructed, which includes six endometrial single cell subpopulations. The rows are 70+ characteristic genes, the columns are the six endometrial single cell subpopulations, and the values ​​are the relative expression levels (TPM) of genes in a certain single cell type. 5) Assume the matrix R represents the proportion of the six types of single cells in the sample, with rows representing the six types of endometrial tissue cells, columns representing sample numbers, and values ​​representing the relative proportion of each endometrial tissue single cell type, which are unknown variables. Then S R=E. Solving the equation using the deconvolution method yields R', where the value represents the relative cell content of the six single-cell types of endometrial tissue obtained through deconvolution analysis. Figure 2 ; 6) After analyzing the proportion of the six types of cells in the endometrium, a correlation analysis was conducted between them and the clinical pregnancy rate; 7) In this embodiment, based on 291 samples, the proportions of six cell types were analyzed. The distribution of the proportions of the six cell types in the failed and successful pregnancy groups was analyzed. Significant differences were indicated by a significant correlation. The results showed a significant correlation between the proportion of non-ciliated epithelial cells and clinical pregnancy outcomes. A low proportion of non-ciliated epithelial cells suggests impaired endometrial function. The proportion of non-ciliated epithelial cells in patients was divided into seven different intervals from low to high (<45%, 45%~50%, 50%~55%, 55%~60%, 60%~65%, 65%~70%, >70%), with implantation rates of 27.3%, 30.0%, 37.5%, 35.3%, 44.7%, 56.1%, and 58.8%, respectively. A trend was observed where a higher proportion of non-ciliated epithelial cells correlated with a higher implantation rate (p=0.039). Figure 3 As shown.

[0064] 8) By controlling the thresholds using two indicators that maximize and significantly increase the difference in pregnancy rate between groups and have the best explanatory power for the population, a 65% proportion of non-ciliated epithelial cells is defined as meeting the criteria. Figure 4 As shown.

[0065] In this embodiment, the low percentage of non-ciliated epithelial cells was validated on a completely independent validation set, indicating that a low percentage of non-ciliated epithelial cells suggests impaired endometrial function. When the percentage of non-ciliated epithelial cells is below 65%, both clinical pregnancy and continued pregnancy outcomes show a decreasing trend. Figure 5 As shown.

[0066] Example 2: M1 / M2 Immune Balance Index

[0067] This embodiment provides an example of applying the technology of the present invention to assess endometrial function using the M1 / M2 immune balance index.

[0068] Bulk RNA-seq was performed on 1384 endometrial tissue samples from three independent datasets, with clinical pregnancy (intrauterine gestational sac at 6–7 weeks) as the analytical endpoint. 500 samples were randomly selected as the development set.

[0069] Based on prior knowledge, two gene modules, M1 and M2, were identified in macrophages. For each sample, standard normalization and intra-sample z-normalization were performed first, and then the M1 module score (M1) and the M2 module score (M2) were calculated separately. M1 / M2 was defined as log10(M1) – log10(M2), with a larger M1 / M2 value representing an excessively pro-inflammatory state. A threshold was selected on the development set using M1 / M2 as a continuous indicator.

[0070] K-fold cross-validation was performed on the development set with 50 repetitions: each time the data was divided into k subsets, and the ROC was plotted on the subsets using M1 / M2 against "clinical pregnancy failure". The threshold maximizing Youden's J was selected under the constraint of sensitivity ≥80% (identifying adverse outcomes as positive). The 50 results from the 50 repetitions were then analyzed. The median of k thresholds is used as the locking threshold T. M1 / M2 This is then applied once in subsequent validation sets without further parameter tuning. In each validation set, the parameter setting is M1 / M2 > T. M1 / M2 The results were classified as M1_Skewed (excessive pro-inflammatory) or Mac_Balanced (immune balance), and the clinical pregnancy rate differences between the two groups were compared and the effect size was calculated.

[0071] See results Figure 6 In three datasets from three different centers, the mean pregnancy rate in the M1 hyperinflammatory group was significantly lower than that in the macrophage immune balance group, with a clinical pregnancy rate decrease of 18-38% in the hyperinflammatory group.

[0072] Example 3: uNK1 / uNK3 immune balance index

[0073] Bulk RNA-seq of 1384 endometrial tissue samples from three independent sources was used, with clinical pregnancy (intrauterine gestational sac at 6–7 weeks) as the analytical endpoint. 500 samples (same as in Example 2) were randomly selected as the development set.

[0074] Based on prior knowledge, two gene modules, uNK1 (dNK1) and uNK3 (dNK3), were identified as tissue-resident genes. For each sample, standard normalization and intra-sample z-standardization were performed first, and then the uNK1 module score (uNK1) and the uNK3 module score (uNK3) were calculated separately. uNK1 / uNK3 was defined as log10(uNK1) – log10(uNK3), with a smaller uNK1 / uNK3 value representing an excessively pro-inflammatory state. A threshold was selected on the development set using uNK1 / uNK3 as a continuous indicator.

[0075] K-fold cross-validation was performed on the development set with 50 repetitions: each time the data was divided into k subsets, and the ROC was plotted on the subsets using uNK1 / uNK3 against "clinical pregnancy failure". The threshold maximizing Youden's J was selected under the constraint of sensitivity ≥80% (identifying adverse outcomes as positive). The 50 repetitions were then used to generate the 50... The median of k thresholds is used as the locking threshold T. uNK1 / uNK3 This will be applied once in subsequent validation sets without further parameter tuning. In each validation set, the parameters are set according to uNK1 / uNK3 < T. uNK1 / uNK3 The results were classified as uNK3_Skewed (excessive pro-inflammatory) or uNK_Balanced (immune balance), and the clinical pregnancy rate differences between the two groups were compared and the effect size was calculated.

[0076] See results Figure 7 In three datasets from three different centers, the mean pregnancy rate in the uNK over-pro-inflammatory group was significantly lower than that in the uNK cell immune balance group, with a clinical pregnancy rate decrease of 19-27% in the over-pro-inflammatory group.

[0077] Example 4: Th1 / Th2 Immune Balance Index

[0078] Bulk RNA-seq was performed on 1384 endometrial tissue samples from three independent datasets, with clinical pregnancy (intrauterine gestational sac at 6–7 weeks) as the analytical endpoint. 500 samples were randomly selected as the development set.

[0079] Based on prior knowledge, two gene modules, Th1 and Th2, for helper T cells were identified. For each sample, routine normalization and intra-sample z-standardization were performed first, and then the Th1 module score (Th1) and the Th2 module score (Th2) were calculated separately. Th1 / Th2 was defined as log10(Th1) – log10(Th2), with a larger Th1 / Th2 value representing an excessively pro-inflammatory state. Thresholds were selected on the development set using Th1 / Th2 as a continuous indicator.

[0080] K-fold cross-validation was performed on the development set with 50 repetitions: each time the data was divided into k subsets, and the ROC was plotted on the subsets using Th1 / Th2 for "clinical pregnancy failure". The threshold maximizing Youden's J was selected under the constraint of sensitivity ≥80% (identifying adverse outcomes as positive). The 50 results from the 50 repetitions were then analyzed. The median of k thresholds is used as the locking threshold T. Th1 / Th2 This parameter is applied once in subsequent validation sets without further parameter tuning. In each validation set, the parameter is set according to Th1 / Th2 > T. Th1 / Th2 The results were categorized as Th1_Skewed (excessive pro-inflammatory) or Th_Balanced (immune balance), and the clinical pregnancy rates of the two groups were compared and the effect size was calculated.

[0081] See results Figure 8 In three datasets from three different centers, the mean pregnancy rate in the Th1 over-pro-inflammatory group was significantly lower than that in the T helper cell immune balance group, with a clinical pregnancy rate decrease of 24-27% in the over-pro-inflammatory group.

[0082] Example 5: Treg / Th17 immune balance index

[0083] Bulk RNA-seq was performed on 1384 endometrial tissue samples from three independent datasets, with clinical pregnancy (intrauterine gestational sac at 6–7 weeks) as the analytical endpoint. 500 samples were randomly selected as the development set.

[0084] Based on prior knowledge, two gene modules were identified: regulatory T cells (Treg) and helper T cell 17 (Th17). For each sample, standard normalization and intra-sample z-normalization were performed first, and then the Treg module score (Treg) and the Th17 module score (Th17) were calculated separately. Treg / Th17 was defined as log10(Treg) – log10(Th17), with a smaller Treg / Th17 value representing an excessively pro-inflammatory state. A threshold was selected on the development set using Treg / Th17 as a continuous indicator.

[0085] K-fold cross-validation was performed on the development set with 50 repetitions: each time the data was divided into k subsets, and the ROC curve for "clinical pregnancy failure" was plotted on the subsets using Treg / Th17. The threshold maximizing Youden's J was selected under the constraint of sensitivity ≥80% (identifying adverse outcomes as positive). The 50 results from the 50 repetitions were then analyzed. The median of k thresholds is used as the locking threshold T. Treg / Th17This parameter is applied once in subsequent validation sets without further parameter tuning. In each validation set, the parameter is set according to Treg / Th17 < T. Treg / Th17 The group was classified as Th17_Skewed (excessive pro-inflammatory) or Th17_Treg_Balanced (immune balance). The clinical pregnancy rate difference between the two groups was compared and the effect size was calculated.

[0086] See results Figure 9 In three datasets from different centers, the mean pregnancy rate in the Th17 hyperinflammatory group was significantly lower than that in the Treg and Th17 cellular immune balance groups, with a clinical pregnancy rate decrease of 22-25% in the hyperinflammatory group.

[0087] Example 6: Endometrial cell senescence scoring

[0088] Bulk RNA-seq was performed on 285 endometrial tissue samples with age information from two independent sources. The analytical endpoint was clinical pregnancy (intrauterine gestational sac at 6–7 weeks). Ninety samples were randomly selected as the development set.

[0089] Based on prior knowledge, gene modules related to cellular senescence were identified (including genes related to cellular senescence-related secretory phenotypic factors, DNA damage response and cell cycle inhibition, mitochondrial dysfunction, and extracellular matrix remodeling gene reprogramming). For each sample, conventional normalization and intra-sample z-normalization were performed to obtain a senescence score; a higher score indicates a greater degree of cellular senescence. Using the cellular senescence score as a continuous indicator, a threshold was selected on the development set.

[0090] K-fold cross-validation was performed on the development set with 50 repetitions: each time the data was divided into k subsets, and the ROC was plotted on the subsets for "clinical pregnancy failure" using cell senescence scoring. The threshold maximizing Youden's J was selected under the constraint of sensitivity ≥80% (identifying adverse outcomes as positive). The 50 values ​​generated from the 50 repetitions were then used to perform the cross-validation. The median of k thresholds is used as the locking threshold T. ageing This method is applied once in subsequent validation sets without further parameter tuning. In each validation set, the aging score is greater than T. ageing Those diagnosed with accelerated endometrial aging were classified as the normal group, while those with normal endometrial aging were classified as the normal group. The clinical pregnancy rates of the two groups were compared and the effect size was calculated.

[0091] like Figure 10As shown, in the development dataset, the chronological ages of the successful and failed pregnancies were similar (p=0.39), while the aging score of the failed pregnancy group was significantly higher (p=0.023). After stratification by threshold, the clinical pregnancy rate of the accelerated aging subgroup was 12.5% ​​(2 / 16), significantly lower than the 65.4% (34 / 52, Fisher p=3.4×10⁻⁶) of the normal group. -4 Furthermore, this was accompanied by a significant decrease in the proportion of non-ciliated epithelium (p=0.012) and reduced PGR transcription (p=0.0041), suggesting impaired endometrial epithelial composition and progesterone response. In external validation sets, such as... Figure 11 Similarly, no difference was observed in age (p=0.69), but the pregnancy rate in the accelerated aging group was only 30.2% (13 / 43), lower than the 48.6% (85 / 175, p=0.0395) in the normal group, and the proportion of non-ciliated epithelium was significantly reduced (p=2×10). -7 ) and decreased PGR expression (p=4.3×10 -6 ).

[0092] Figure 12 The results showed that the cGAS–STING activation pathway was upregulated in the accelerated aging group (p=7.4×10⁻⁶). -5 The downregulation of mitochondrial fusion genes (MFN1 / 2, OPA1) and mitochondrial fusion genes (MFN1 / 2, OPA1) (p=0.019) was positively correlated, which is consistent with the mechanism chain of "inflammatory aging - impaired mitochondrial function - restricted decidualization".

[0093] In summary, without the confounding effect of chronological age, cellular senescence scoring based on prior genetic modules can reliably identify individuals with "accelerated endometrial aging" and has been consistently associated with a significant decrease in clinical pregnancy rates in both the development and validation cohorts. Furthermore, it provides mechanistic evidence consistent with inflammation, mitochondrial, and progesterone signaling. This score can serve as a basis for clinical stratification and personalized interventions (such as epithelial support / mitochondrial function support and targeted immune regulation).

[0094] Various modifications and variations to the methods, evaluation apparatus, and uses of the present invention will be apparent to those skilled in the art without departing from the scope and spirit of the invention. Although the invention has been described in conjunction with specific embodiments, it will be understood that further modifications are possible, and the claimed invention should not be unduly limited to such specific embodiments. In fact, various modifications to the described methods for carrying out the invention that are apparent to those skilled in the art are intended to fall within the scope of the invention.

Claims

1. A method for assessing endometrial function, the method comprising: Cellular component analysis: Using a deconvolution method, based on endometrial transcriptome sequencing (Bulk RNA-seq) data, and utilizing the endometrial immune cell or tissue-type single-cell feature matrix, the composition of endometrial single cells or the proportion of each cell type in the target sample is analyzed. Gene module scoring: In Bulk RNA-seq data, the average expression level of each feature was calculated based on the feature gene set related to immunity and aging, and then standardized to obtain immune indicators and aging scores respectively; Based on the above-mentioned cellular composition, immune indicators, and aging score, endometrial function was assessed.

2. The method according to claim 1, wherein, The process of analyzing cellular components includes: Constructing a target single-cell feature expression matrix: The single-cell feature expression matrix is ​​derived from self-tested or publicly available single-cell transcriptome sequencing data of endometrial tissue. Based on this data analysis, the cell types contained in the tissue and the genes specifically expressed by each cell type are identified. Then, a cell-specific gene expression matrix S is constructed, where S is a gene × cell type matrix. The expression data obtained from bulk transcriptome sequencing of endometrial tissue transcriptome nucleic acid (RNA) samples were used to construct a bulk expression matrix B, where B is a gene × sample matrix; The expression levels of multiple characteristic genes involved in matrix S are extracted from matrix B to form the tissue Bulk transcriptome characteristic gene expression matrix E, where E is a gene × sample matrix; Set up a matrix R to calculate the proportion of each single cell, where R is a sample × cell type matrix, and use the deconvolution method to analyze and obtain the cell proportion.

3. The method according to claim 2, wherein: The cell-specific expression gene matrix S is set as I x J, I is the number of genes, J is the number of cell types, the row of the matrix is a gene, the column is a cell type, and the value is the expression amount of the gene in a specific cell type, S ij is the expression value of the ith characteristic gene in the jth cell type. The bulk expression matrix B consists of rows representing genes, columns representing samples, and values ​​representing gene expression levels. The Bulk transcriptome characteristic gene expression matrix E is of size I×N, where I is the number of characteristic genes and N is the number of samples. in Let the expression value of the i-th characteristic gene in the n-th sample be denoted as ; The matrix R representing the proportion of each single cell type is of size N×J, where N is the number of samples in matrix E, J is the number of single cell types in matrix S, and R... nj The percentage of the j-th cell type in the n-th sample tissue is an unknown value that needs to be solved. S R=E. Solving the equation using the deconvolution method yields R', and the value in R' is the cell proportion obtained from the deconvolution analysis.

4. The method according to claim 3, wherein, The deconvolution parsing process includes: ① Take the expression value of the I characteristic gene of the nth sample from the tissue bulk transcriptome characteristic gene expression matrix E, and let it be the array SampleE. Then: ② For the nth sample, the bulk expression value of each gene is equal to the total expression value of multiple single-cell clusters. Thus, each characteristic gene will have an equation. Taking the i-th equation as an example, the formula is as follows: ③ For the nth sample, there are I equations in the above equations, forming a system of equations. Support vector regression analysis is then used to deduce the most reliable R. nj The values ​​are used to form the final matrix of the relative content of each endometrial single cell in each sample.

5. The method according to any one of claims 2-4, further comprising: Based on clinical information of the samples, the correlation between the composition of intima single cells and clinical observation indicators was observed in groups of samples. Setting an appropriate threshold for the percentage of single cells in the endometrium can predict clinical observation indicators.

6. The method according to claim 1, wherein, The gene module scoring process includes: Identify pro-inflammatory / anti-inflammatory immune subsets (Treg, Th17, Th1, Th2, uNK1, uNK3, M1, M2) and the set of transcriptional signature genes associated with cellular senescence; For each sample to be tested, after completing the routine normalization, the mean expression value (mean_i) of each feature gene set, the mean expression value (mean_all) of the whole gene set, and the standard deviation (sd_all) are calculated, and the dimensionless module score is obtained by standardizing according to the in-sample z-squared. Z_i= (mean_i – mean_all) / sd_all Intra-sample z-normalization is performed on the mean of each feature to naturally eliminate the influence of differences in sequencing depth / overall expression amplitude, resulting in dimensionless scores that can be compared across batches. The scores from the above modules are further used to construct the immune balance index and aging score.

7. The method according to any one of claims 1-6, wherein, The assessment of endometrial function includes evaluation of three dimensions: endometrial cell composition, immune status, and aging level.

8. A device for assessing endometrial function, the device comprising a detection unit and a data analysis unit, wherein: The detection unit is used to detect samples from individuals and obtain endometrial transcriptome sequencing data; The data analysis unit is used to analyze and process the obtained endometrial transcriptome sequencing data.

9. The apparatus according to claim 8, characterized in that, When the data analysis unit performs analysis and processing, it includes: The method described in any one of claims 1-7 is used to analyze the endometrial single-cell composition or the proportion of each type of cell, immune indicators, and aging fraction in the target sample to assess endometrial function.

10. A non-transitory computer-readable storage medium storing computer instructions, which, when executed by a processor, perform the following: resolving the endometrial single-cell content or the proportion of each type of cell, immune indicators, and aging fraction in a target sample using the method described in any one of claims 1-7, and further evaluating endometrial function.

11. A computer device comprising a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, the processor executing the computer instructions to: analyze the endometrial single cell content or the proportion of various cell types, immune indicators and aging fraction in a target sample using the method of any one of claims 1-7, and further evaluate endometrial function.

Images