Biomarker combination and application thereof in predicting embryo quality

Abstract

The invention discloses a biomarker combination and application thereof in predicting embryo quality. Wherein the biomarker combination comprises small molecule non-coding RNA (Ribonucleic Acid) with nucleotide sequences as shown in SEQ ID NO.1-3; or micromolecule non-coding RNA as shown in SEQ ID NO.1-8; or micromolecular non-coding RNA as shown in SEQ ID NO.1-13. The invention also discloses a method for preparing the micromolecular non-coding The biomarker combination disclosed by the invention not only contains miRNA, but also contains multiple types of sncRNA such as tsRNA, snRNA, Y RNA, snoRNA and the like, and the biomarker combination has the characteristic of obviously distinguishing the embryo quality. When the biomarker combination is used for predicting the embryo quality, only an embryo culture solution is needed, and as a non-invasive sample collection operation, the shape and subsequent development of the embryo are not influenced. Meanwhile, the invention also provides a construction method and a prediction method of the embryo quality prediction model, and the final model has high sensitivity and high specificity, thereby providing guidance for embryo screening.

Description

Technical Field

[0001] The present invention relates to the field of biotechnology, and in particular to a biomarker combination and an application thereof in predicting embryo quality. Background Art

[0002] Infertility is a major health challenge facing the world, affecting approximately one in six people. Although assisted reproductive technology has been used for decades, the pregnancy rate and delivery rate per transplant are still approximately 35% and 25%, respectively. In order to improve the success rate of assisted reproductive technology and reduce the risk of babies with genetic diseases, scholars have made many efforts in embryo screening, including preimplantation genetic testing (PGT). However, PGT is an invasive detection method, which may affect the survival rate and implantation potential of the embryo and may cause long-term risks to offspring. Therefore, non-invasive embryo biopsy (niPGT) technology using human embryo culture medium (hSCM) for embryo screening came into being. The existing niPGT technology focuses more on the detection of embryo euploidy using extracellular DNA, but studies have shown that extracellular small non-coding RNA (sncRNA) in embryo culture medium, especially microRNA (miRNA), a class of sncRNA with a length of about 22 nucleotides, plays an important role in regulating embryo development and predicting embryo implantation potential.

[0003] MiRNA is stable in the extracellular environment and easy to detect, so it can be used as a biomarker to predict embryo implantation outcomes. To date, many papers have revealed the predictive potential of microRNA in embryo culture medium. These studies have used a variety of detection methods, including real-time reverse transcription quantitative polymerase chain reaction (RT-qPCR), droplet digital PCR, chips, next-generation sequencing (NGS), or a combination of these methods. However, the results of these studies are inconsistent or even contradictory, which may be due to the use of different detection strategies and techniques, the small content of microRNA in embryo culture medium, which leads to reduced sensitivity and accuracy, and the heterogeneity of individual samples. In order to overcome the problem of low detection sensitivity, scholars have tried to mix embryo culture medium from multiple embryo sources. This method can improve detection sensitivity and accuracy, however, this strategy cannot be used to prioritize individual embryos, that is, individual embryos cannot be screened. Another limitation of the current embryo culture medium research is that scholars mostly focus on microRNAs and ignore other sncRNAs, including fragments derived from transfer RNA (tsRNA). This type of sncRNA is very abundant in embryo culture medium, and more and more studies have revealed the biological functions and clinical significance of tsRNA in the extracellular and blood circulation systems, as well as its potential as a new non-invasive biomarker. In addition to tsRNA, there are also small nucleolar RNA (snoRNA), small nuclear RNA (snRNA) and Y RNA (a component of Ro60 ribonucleoprotein particles) in embryo culture medium. In bovine embryo culture medium, scholars have found that the expression of snoRNAs discharged during embryonic development varies at different stages. For example, the expression of snoRNAs reaches a peak at the 8-cell stage of the embryo, indicating that snoRNAs may be involved in the activation of maternal genes. In addition, scholars have found that snoRNAs bind to miRNAs, thereby affecting the expression of the ZNF23 gene, leading to the proliferation, metastasis and invasion of endometrial cells, thereby achieving the result of regulating endometrial receptivity. Therefore, it is necessary to conduct a more comprehensive analysis of sncRNAs in the culture medium of individual embryos and further explore the application of all sncRNAs as biomarkers in embryo screening.

[0004] For the field of assisted reproductive technology, a highly sensitive and highly specific early embryo quality prediction method can effectively screen high-quality embryos, which is very necessary for improving the success rate of transplantation and pregnancy rate. As a non-invasive and non-invasive sample type, the sncRNA composition information of the embryo culture medium can be used as a carrier for predicting embryo quality. Existing prediction methods only focus on miRNA, but the inventors have found that miRNA only occupies a very small part, about 1%, and most of the remaining tsRNA and other sncRNA through a comprehensive analysis of the composition of sncRNA in the embryo culture medium. Therefore, it is necessary to fully utilize sncRNA information to construct a prediction model. At the same time, there is currently no report on the process and method of constructing a model for predicting early embryo quality using all types of sncRNA information. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the purpose of the present invention is to provide a biomarker combination and its application in predicting embryo quality, aiming to solve the problem that the existing embryo quality prediction methods have limited sensitivity and accuracy and cannot screen single embryos.

[0006] The technical solution of the present invention is as follows:

[0007] In a first aspect of the present invention, a biomarker combination is provided, wherein the biomarker combination comprises a small non-coding RNA having a nucleotide sequence as shown in SEQ ID NO. 1-3.

[0008] Specifically, the present invention provides a biomarker combination for predicting embryo quality, which includes three miRNAs derived from blastocyst period embryo culture fluid: hsa-miR-345-5p (SEQ ID NO: 1), hsa-miR-202-3p (SEQ ID NO: 2) and hsa-miR-12136 (SEQ ID NO: 3).

[0009] In a second aspect of the present invention, a biomarker combination is provided, wherein the biomarker combination comprises a small non-coding RNA having a nucleotide sequence as shown in SEQ ID NO. 1 to 8.

[0010] Specifically, based on the first aspect, the present invention further provides a biomarker combination for predicting embryo quality, the biomarker combination comprising 4 miRNAs, 1 tRNA, 1 snRNA, 1 Y RNA and 1 snoRNA derived from embryo culture medium during the blastocyst period. Among them, miRNAs are: hsa-miR-345-5p (SEQ ID NO: 1), hsa-miR-202-3p (SEQ ID NO: 2), hsa-miR-12136 (SEQ ID NO: 3) and hsa-miR-128-3p (SEQ ID NO: 4); tRNA is: tRNA-Ile-AAT-6-1 (SEQ ID NO: 5); snRNA is: RNU4-82P (SEQ ID NO: 6); Y RNA is: Y_RNA.142 (SEQ ID NO: 7); snoRNA is: SNORD65 (SEQ ID NO: 8).

[0011] The third aspect of the present invention provides a biomarker combination, wherein the biomarker combination includes a small non-coding RNA having a nucleotide sequence as shown in SEQ ID NO.1-13.

[0012] Specifically, based on the first aspect, the present invention further provides a biomarker combination for predicting embryo quality, which includes 4 miRNAs, 1 tRNA, 1 snRNA, 2 Y RNAs and 5 snoRNAs derived from blastocyst period embryo culture fluid. Among them, miRNA is: hsa-miR-345-5p (SEQ ID NO: 1), hsa-miR-202-3p (SEQ ID NO: 2), hsa-miR-12136 (SEQ ID NO: 3) and hsa-miR-128-3p (SEQ ID NO: 4); tRNA is: tRNA-Ile-AAT-6-1 (SEQ ID NO: 5); snRNA is: RNU4-82P (SEQ ID NO: 6); Y RNA is: Y_RNA.142 (SEQ ID NO: 7) and Y_RNA.449 (SEQ ID NO: 9); snoRNA is: SNORD65 (SEQ ID NO: 8), SNORD58C (SEQ ID NO: 10), SNORD88C (SEQ ID NO: 11), SNORD68 (SEQ ID NO: 12) and SNORD43 (SEQ ID NO: 13).

[0013] A fourth aspect of the present invention provides a use of a biomarker combination as described in any one of the above items in the preparation of a kit for predicting embryo quality.

[0014] In a fifth aspect of the present invention, a reagent for detecting a biomarker combination is provided, wherein the reagent is used to detect the expression level of the biomarker combination as described in any one of the above items.

[0015] In a preferred technical solution, the reagent is a primer and / or a probe. It can be understood that the primer and / or the probe is designed based on the above sncRNA nucleotide sequence SEQ ID NO: 1 to SEQ ID NO: 13.

[0016] A sixth aspect of the present invention provides a use of a reagent as described in any one of the above items in the preparation of a kit for predicting embryo quality.

[0017] In a preferred technical solution, the kit further comprises a reverse transcription reaction solution, the main components of which are RNA reverse transcriptase, reverse transcription primer, buffer, DTT, dNTPs, and RNase inhibitor; and / or, the kit further comprises a PCR reaction solution, the main components of which are DNA polymerase, buffer, upstream and downstream primers, and dNTPs; and / or, the kit further comprises a ROX reference dye and nuclease-free water; and / or, the kit further comprises a standard. And the detection object of the kit is an embryo culture solution. Specifically, the detection object of the kit is an embryo culture solution collected before the embryo develops to the blastocyst stage / or the embryo stagnates in the blastocyst stage. More specifically, the embryo culture process refers to the operation of transferring the embryo (Day 0) after fertilization to the morula stage (Day 3) to continue the culture in a new embryo culture solution droplet, and finally collecting the embryo culture solution droplet on Day 5 or 6. This step depends on the degree of embryo development. If the embryo develops normally, it is collected before the embryo hatches, otherwise it is collected on Day 6.

[0018] A seventh aspect of the present invention provides a method for constructing an embryo quality prediction model, the method comprising:

[0019] Data acquisition: including sample information data and experimental data; wherein, the sample information data uses the embryo quality as a grouping label, and the embryos are scored according to the Gardner blastocyst scoring system. Blastocysts with a blastocyst cavity score ≥ 3, and the number and arrangement of the inner cell mass cells are grade B or above, and the number and structure of the trophoblast cells are grade B or above are defined as high-quality embryos, otherwise they are low-quality embryos; the experimental data are the expression amount data of biomarkers in the embryo culture medium; wherein, the nucleotide sequence of the biomarker is shown in SEQ ID NO.1-3, or the nucleotide sequence of the biomarker is shown in SEQ ID NO.1-8, or the nucleotide sequence of the biomarker is shown in SEQ ID NO.1-13;

[0020] Data analysis: standardizing and normalizing the expression data to obtain processed data;

[0021] Model training and optimization: The processed data is input into a machine learning model for training, and the model parameters are adjusted through multiple cycles to obtain an embryo quality prediction model.

[0022] A preferred technical solution, the method for constructing an embryo quality prediction model based on the expression level of biomarkers in embryo culture medium specifically comprises:

[0023] Data acquisition: including sample information data and experimental data of each sample. Among them, the sample information data uses the embryo quality as the grouping label, and the embryos on the fifth day after fertilization (Day 5) are scored according to the Gardner blastocyst scoring system. The blastocyst cavity score is ≥3, and the number and arrangement of the inner cell mass cells are grade B or above, and the number and structure of the trophoblast cells are grade B or above. The blastocyst is defined as a high-quality embryo, otherwise it is a low-quality embryo; the experimental data is the expression amount data of the biomarker in the embryo culture medium; wherein, the nucleotide sequence of the biomarker is shown in SEQ ID NO.1-3, or the nucleotide sequence of the biomarker is shown in SEQ ID NO.1-8, or the nucleotide sequence of the biomarker is shown in SEQ ID NO.1-13;

[0024] Data analysis: After data acquisition, the expression data is standardized and normalized to obtain processed data so that the expression levels of biomarkers between different samples are comparable;

[0025] Model training and optimization: The processed data is input into a machine learning model for training, and the model parameters are adjusted through multiple cycles to obtain an embryo quality prediction model. Preferably, the machine learning model is a supervised learning model. Among them, the model parameters refer to the set of parameters that can be optimized and adjusted corresponding to the adopted model. The purpose of optimization and adjustment is to improve the model prediction effect, including sensitivity, accuracy, precision, area under the receiver operating characteristic curve (ROC) (AUC), etc.

[0026] An eighth aspect of the present invention provides an embryo quality prediction model, wherein the embryo quality prediction model is constructed by the method described in the seventh aspect.

[0027] A ninth aspect of the present invention provides a method for predicting embryo quality, the method comprising:

[0028] Data acquisition: obtaining the expression data of the biomarker in the embryo culture medium; wherein the nucleotide sequence of the biomarker is as shown in SEQ ID NO.1-3, or the nucleotide sequence of the biomarker is as shown in SEQ ID NO.1-8, or the nucleotide sequence of the biomarker is as shown in SEQ ID NO.1-13;

[0029] Data analysis: After acquisition, the expression data is standardized and normalized to obtain processed data;

[0030] Identification: inputting the processed data into the embryo quality prediction model as described in the eighth aspect to obtain a prediction result of the embryo quality in the embryo culture medium;

[0031] Output: Output the recognition results.

[0032] Specifically, the prediction result is whether the target object belongs to the embryo culture fluid derived from "high-quality embryos" or "low-quality embryos", or the prediction result is the probability that the target object is the embryo culture fluid derived from "high-quality embryos" or "low-quality embryos", and then the embryo quality of the target object is determined by comparing the probabilities.

[0033] It should be noted that the samples obtained in the ninth aspect are distinguished from the samples in the construction method of the seventh aspect. The two types of samples should have no intersection, and the same sample cannot be used in both the construction method of the seventh aspect and the prediction method of the ninth aspect. Here, in order to distinguish, we refer to all samples used in the construction method of the seventh aspect as "training set" samples, and the samples used in the prediction method of the ninth aspect as "test set" samples. It can be understood that the samples used in the construction method of the seventh aspect can be further divided into "training set" samples and "validation set" samples.

[0034] Beneficial effects: The present invention provides a biomarker combination and its application in predicting embryo quality. Compared with the prior art, the biomarker combination of the present invention is used to predict embryo quality, and only embryo culture medium is required. This is a non-invasive test and will not have an adverse effect on the embryo. In addition, the biomarker combination of the present invention not only includes miRNA, but also includes various types of sncRNA such as tsRNA, snRNA, Y RNA, snoRNA, etc., which is different from other common small molecule non-coding RNA markers. The present invention fully explores the composition of all types of sncRNA in the embryo culture medium, and through strict feature screening, the final biomarker combination has obvious characteristics that can distinguish embryo quality, and the embryo quality prediction model trained using the biomarker combination has better accuracy, sensitivity, specificity and AUC value. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a schematic diagram of the process of screening a characteristic sncRNA set for predicting embryo quality in Example 1 of the present invention.

[0036] Figure 2 It is a schematic diagram of the differential expression analysis results of sncRNA.

[0037] Figure 3 It is the statistical result of the out-of-bag error rate of sets 1 to 3 in the random forest model training process in Example 1.

[0038] Figure 4 It is the statistical result of multiple indicators of sets 1 to 3 in Example 1 in the random forest model test set.

[0039] Figure 5 It is the AUC result of sets 1 to 3 in Example 2 in the logistic regression model.

[0040] Figure 6 It is the AUC result of sets 1 to 3 in Example 3 in the Bayesian logistic regression model.

[0041] Figure 7 It is the prediction result of the model constructed by using the hsa-miR-202-3p, hsa-miR-345-5p, and hsa-miR-12136 expression data obtained by RT-qPCR detection in Example 4. DETAILED DESCRIPTION

[0042] The present invention provides a biomarker combination and its application in predicting embryo quality. In order to make the purpose, technical scheme and effect of the present invention clearer and more specific, the present invention is further described in detail below.

[0043] Embodiment 1:

[0044] 1. Materials

[0045] 1.1 Research subjects:

[0046] The inventors collected 185 human blastocyst culture fluids from 37 assisted reproductive cycles visiting the Assisted Reproductive Center of the Shenzhen Hospital of the University of Hong Kong from January 2022 to November 2022. The study was approved by the Ethics Committee (No.hkuszh2022262) and the patients' informed consent was obtained. The embryos all came from intracytoplasmic sperm injection (ICSI) cycles. When the embryos were fertilized and cultured to day 3 (Day 3), the medium was changed and the embryos were transferred to about 20 μL of fresh G-2PLUS (Vitrolife, Sweden) blastocyst culture medium and cultured until day 5 (Day 5) or day 6 (Day 6) (depending on the development of the blastocyst), and then 10 μL of culture medium was collected, transferred to a PCR tube and stored at -80 °C for use. According to the Gardner blastocyst scoring system, the development of blastocysts on Day 5 or Day 6 after fertilization was scored. Blastocysts with a blastocyst cavity score ≥ 3, inner cell mass cell number and cell arrangement of grade B or above, and trophoblast cell number and structure of grade B or above were defined as “high-quality” embryos, and embryos below the above standards were defined as “low-quality” embryos.

[0047] 2. Methods

[0048] 2.1 sncRNA library construction and sequencing

[0049] Total RNA extraction: 10 μL of embryo culture medium was mixed with 100 μL of RNeasy Mini Kit lysis (RTL) buffer (Qiagen, Germany) to extract total RNA. After vortexing, 160 μL of anhydrous ethanol (Macklin, China) and 2 μL of coprecipitant (Merck Millipore, Germany) were added. The mixture was incubated at -20°C overnight or for two days. After incubation, centrifugation was performed at 16,000 × g for 20 minutes at 4°C. The supernatant was discarded, and the precipitate was washed twice with 1 ml of fresh 80% ethanol and air dried for 5-6 minutes, then resuspended in 7 μL of RNase-free water and stored at -80°C for use.

[0050] sncRNA sequencing library construction and sequencing: 3 μL of total RNA from each sample was used to prepare cDNA library in a PCR tube. All steps were performed using Veriti TM(Applied Biosystems, USA) thermal cycler. A customized 8bp sequence identification code (UMI, IDT Technologies, Singapore) was attached to each sample in the second round of PCR step. Afterwards, the cDNA library was purified separately using AMPure XP magnetic beads (Beckman Coulter Inc., USA) at a ratio of 1:1 between magnetic beads and samples, and quantified using Qubit 4 Fluorometer (Invitrogen, USA) and Qubit 1X dsDNA HS Detection Kit (Invitrogen, USA). All purified samples were mixed according to concentration. The double-stranded library was denatured, circularized and digested using the MGIEasy Universal Library Conversion Kit (App-A) (Shenzhen BGI Intelligent Manufacturing Technology Co., Ltd., China) to obtain single-stranded circular DNA, and then DNA amplification was performed using rolling circle amplification (RCA) and DNA nanoball (DNB) technology. After DNB purification, quantification, and identification, the sncRNA library was sent to Shenzhen Geneplus for sequencing on the MGISEQ-2000RS platform (Shenzhen BGI Intelligent Manufacturing Technology Co., Ltd., China), with a sequencing length of 1 × 51 bp.

[0051] 2.2 Bioinformatics analysis

[0052] The sequencing data were analyzed using conventional upstream bioinformatics processes, as follows: UMI sequences were extracted using the UMI-tools (v1.1.2) tool, followed by the removal of adapter fragments using the Cutadapt tool (v3.5). The sequencing fragments were annotated to the human reference genome (GRCh38) using STAR (v2.5.1b), and then the redundant fragments caused by PCR amplification were removed using UMI-tools. The coverage of the sequencing fragments of each sncRNA was calculated using the calculation sncRNA coverage tool in the single-cell sncRNA analysis process, as described in Reference 1. During the calculation, the information of sncRNA was obtained from the miRbase (v22.1), GtRNAdb (release 21), and GENCODE (v42) databases.

[0053] Document 1: Hagemann-Jensen M, Abdullayev I, Sandberg R, Faridani OR.Small-seq for single-cell small-RNA sequencing.Nat Protoc.2018Oct;13(10):2407-2424.doi:10.1038 / s41596-018-0049-y.PMID:30250291.

[0054] 2.3 Construction and validation of embryo quality prediction model

[0055] The inventors collected embryo culture fluid from 68 "high-quality" embryos and 54 "low-quality" embryos as training set samples, and then performed sncRNA library construction, quality inspection, and RNA-Seq sequencing on each sample. At the same time, the inventors collected embryo culture fluid from an additional 32 "high-quality" embryos and 31 "low-quality" embryos as test set samples, and performed sncRNA library construction, quality inspection, and RNA-Seq sequencing on each sample. It should be noted that the test set samples are not used for model construction, and the training set samples are not used for model effect evaluation.

[0056] Model building is divided into the following steps:

[0057] A) Data acquisition: Integrate the sequencing fragment coverage of the training set samples and integrate them into a format where each sncRNA coverage is a row and each sample is a column. At the same time, integrate the embryo quality information of the training set samples as sample grouping information.

[0058] B) Data analysis: The edgeR software (v3.38.4) in R language (4.2.0.0) was used to perform differential expression analysis according to the sample grouping information, and the expression matrix normalized by the weighted truncated mean method (TMM) of the M-value, the mean within the group, the differential expression multiple between the groups, and the statistical probability p value were output. When the p value between the two groups (p-adjusted after BH method correction) ≤ 0.05 and |log2 (difference multiple) | ≥ 1, the sncRNA was considered to be differentially expressed between the two groups.

[0059] C) Feature screening: The differentially expressed sncRNA set in the above B) data analysis step constitutes a preliminary feature screening set 1. Further, sncRNAs with higher expression are screened from set 1 to form set 2, so that the sncRNAs in set 2 satisfy that in the training sample set, the number of samples with an expression level of 4 or more of the sncRNA accounts for more than 15%. Further, the optimal subset set 3 is screened from set 2 using the least absolute shrinkage and selection algorithm (LASSO).

[0060] D) Model training and optimization: The expression matrix after sample normalization was used as input, and the random forest classification model (v4.7-11) was used to build models for the three sets in C) feature screening. The hyperparameter tuning of the random forest model includes ntree (the number of single decision trees in the forest) and mtry (the number of features randomly sampled at each node). The tuning process uses 5-fold cross-validation to improve the generalization ability, and the final optimal hyperparameter combination is used as the hyperparameter of the model for model building. The model building process uses 10-fold cross-validation, and the optimal model in the 10-fold cross-validation is finally used as the final model. At the same time, a 10-fold cross-validation model is constructed through 500 random cycles, and various model evaluation indicators of 500 random cycles are counted.

[0061] The overall process steps are as follows Figure 1 shown.

[0062] 3. Prediction model results

[0063] After the sncRNA coverage data acquisition and normalized expression matrix generation as described above, the 63 test set samples were passed into the above-mentioned optimized model as input information. The prediction result matrix mainly has the information of each sample as a row, and the probability of the sample being judged as a "high-quality" embryo or a "low-quality" embryo as a column. When the probability of a "high-quality" embryo of the sample is greater than the probability of a "low-quality" embryo, the sample is considered to be predicted as a "high-quality" embryo, and vice versa.

[0064] The three feature screening sets contained 39, 13, and 8 sncRNAs, respectively. Set 1 contained 39 sncRNAs that were significantly expressed in the differential analysis ( Figure 2 The evaluation values ​​of the three final model evaluation results constructed in 500 random cycles are as follows Figure 3 and Figure 4 As shown in . Among them, set 1 achieved an accuracy of 0.67, sensitivity of 0.82, specificity of 0.56 and AUC value of 0.68 in the test set samples; set 2 achieved an accuracy of 0.78, sensitivity of 0.86, specificity of 0.76 and AUC value of 0.84 in the test set samples; set 3 achieved an accuracy of 0.74, sensitivity of 0.85, specificity of 0.70 and AUC value of 0.81 in the test set samples. Set 2 performed better in the test set samples.

[0065] The sequences of sncRNAs contained in the three sets mentioned above are shown in Table 1. In Table 1, the naming of sncRNAs is based on the naming rules of the three databases: miRbase v22.1, GtRNAdb release 21, and GENCODE v42.

[0066] Table 1. Sequences of sncRNAs involved in the present invention

[0067]

[0068]

[0069]

[0070]

[0071]

[0072]

[0073] Embodiment 2:

[0074] The difference between Example 2 and Example 1 is that the model training and optimization process are different, and the rest are the same. The specific differences are:

[0075] Model training and optimization: The expression matrix after sample normalization was used as input, and a logistic regression model (glm v4.1.2) was used for 10-fold cross validation to build a model for predicting embryo quality. The best model after 10-fold cross validation was selected as the final model. Since the model had no hyperparameters to adjust, no hyperparameter tuning was performed, and no 500 random cycle model construction was performed.

[0076] Prediction model results: Three different logistic regression models were constructed using three feature selection sets. The performance of these three models in the test set is as follows: Figure 5 As shown in the figure, set 1 achieved an accuracy of 0.52, a sensitivity of 0.81, a specificity of 0.23, and an AUC value of 0.52 in the test set samples; set 2 achieved an accuracy of 0.83, a sensitivity of 0.84, a specificity of 0.84, and an AUC value of 0.845 in the test set samples; set 3 achieved an accuracy of 0.82, a sensitivity of 0.87, a specificity of 0.77, and an AUC value of 0.84 in the test set samples. Set 2 performed better in the test set samples.

[0077] Embodiment 3:

[0078] The difference between Example 3 and Example 2 is that the models used are different, and the rest are the same. The specific differences are:

[0079] The Bayesian logistic regression model (arm v1.13-1) was used to build the model, and the prediction results were as follows: Figure 6As shown in the figure, set 1 achieved an accuracy of 0.50, sensitivity of 0.78, specificity of 0.23 and AUC value of 0.52 in the test set samples; set 2 achieved an accuracy of 0.84, sensitivity of 0.87, specificity of 0.84 and AUC value of 0.842 in the test set samples; set 3 achieved an accuracy of 0.82, sensitivity of 0.88, specificity of 0.77 and AUC value of 0.84 in the test set samples. Set 2 performed better in the test set samples.

[0080] Embodiment 4:

[0081] In this example, the inventors collected 16 embryo culture fluid samples from the Department of Obstetrics and Gynecology, Li Ka Shing Faculty of Medicine, University of Hong Kong, all of which were ICSI sources. The embryo culture, culture fluid collection process and embryo quality judgment criteria are as described in Example 1. The study was reviewed and approved by the Hong Kong University and Hospital Authority Hong Kong Island West Hospital Cluster Research Ethics Committee (UW 23-311) and informed consent was obtained from the patients.

[0082] 1. Extraction of sncRNA from embryo culture medium: Add 0.8 μL UniSp6 RNA spike-in as a reference gene to each sample, extract total RNA from embryo culture medium according to the operating procedures of RNA extraction kit (miRNeasy Serum / Plasma Advanced Kit, Qiagen), and immediately reverse transcribe the extracted RNA into cDNA using reverse transcription kit (miRCURY LNA RT Kit, Qiagen). The reverse transcription operation is as follows 2.

[0083] 2. RT-PCR detection of the relative expression levels of hsa-miR-202-3p, hsa-miR-345-5p and hsa-miR-12136 in embryo culture medium: using the tailing PCR method and universal reverse transcription primers, the upstream and downstream primers of the PCR reaction were purchased (miRCURYLNA miRNA PCR Assay, Qiagen, Table 1).

[0084] Table 1. Primer sequences:

[0085] sncRNA PCR upstream and downstream primer sequence numbers (Qiagen) hsa-miR-202-3p GeneGlobe ID-YP00205990 hsa-miR-345-5p GeneGlobe ID-YP00206006 hsa-miR-12136 GeneGlobe ID-YP02126973 Reference sequence UniSp6 Spike in GeneGlobe ID-YP00203954

[0086] A) Reverse transcription to synthesize cDNA: The reaction system is configured according to the instructions of the miRCURY LNA RT Kit.

[0087]

[0088] Reverse transcription reaction program settings:

[0089] temperature time Number of cycles 42℃ 60min 1 95℃ 5min 1 4℃ Forever 1

[0090] The obtained cDNA was diluted 3 times and used as a template for amplification reaction.

[0091] B) RT-qPCR: Quantitative PCR detection of the prepared cDNA

[0092] The qPCR amplification system used is as follows:

[0093]

[0094] qPCR reaction program settings:

[0095]

[0096]

[0097] 3. Quantitative data analysis: RT-PCR results are expressed as Ct values, △Ct = Ct (target gene) - Ct (reference gene), and the average △Ct value of the high-quality embryos as the control group is calculated. Then, the △△Ct of each sample is calculated as △Ct (sample) - △Ct (average value of the control group). -△△Ct Calculate the relative expression of sncRNA in the sample.

[0098] 4. The embryo culture medium sample contained 8 "high-quality" embryos and 8 "low-quality" embryos. A prediction model was constructed by 3-fold cross validation using the logistic regression model (glm v4.1.2) described in Example 2, and the optimal model of the 3-fold cross validation was used as the final model.

[0099] 5. The relative expression levels of hsa-miR-202-3p, hsa-miR-345-5p, and hsa-miR-12136 in embryo culture medium (-△△Ct) can be calculated by constructing a comprehensive evaluation value formula based on the logistic regression model: Logit = 1.038*(-△△Ct hsa-miR-202-3p )+1.055*(-△△Ct hsa-miR-345-5p )+2.16*(-△△Ct hsa-miR-12136 )

[0100] 6. Calculate the comprehensive evaluation value of each test set sample according to the formula in 5. and construct the receiver operating characteristic curve (ROC) graph, as shown in Figure 7 As shown in Figure 2, the optimal threshold value was determined to be 26.97 based on the Youden index.

[0101] 7. Therefore, the judgment standard of the embryo quality model based on RT-qPCR is to calculate the comprehensive evaluation value of the sample according to the formula in 5. When the evaluation value Logit ≥ 26.97, the sample is judged as a "low-quality" embryo. Conversely, when Logit < 26.97, the sample is judged as a "high-quality" embryo.

[0102] It can be seen from the above examples that the use of second-generation sequencing data to screen suitable small non-coding RNA embryo quality biomarkers, and the use of RT-qPCR to verify the prediction effect of the biomarker set, can well confirm that these sncRNAs are closely related to embryo quality. The present invention provides a method for constructing an embryo quality prediction model using second-generation sequencing and RT-qPCR data, and a group of valuable sncRNAs are obtained by this method.

[0103] The above is a preferred embodiment of the present invention, and is not intended to limit the scope of the present application. For those skilled in the art, any modification, replacement, and embellishment made without departing from the principles of the present application should be included in the protection scope of the present invention.

Claims

1. A biomarker combination, characterized in that: The biomarker combination includes small non-coding RNAs with nucleotide sequences as shown in SEQ ID NOs. 1 to 3.

2. A biomarker combination, characterized in that: The biomarker combination includes small non-coding RNAs with nucleotide sequences as shown in SEQ ID NOs. 1 to 8.

3. A biomarker combination, characterized in that: The biomarker combination includes small non-coding RNAs with nucleotide sequences as shown in SEQ ID NOs. 1 to 13.

4. Use of the biomarker combination according to any one of claims 1 to 3 in preparing a kit for predicting embryo quality.

5. A reagent for detecting a combination of biomarkers, characterized in that: The reagent is used to detect the expression level of the biomarker combination according to any one of claims 1 to 3.

6. The reagent according to claim 5, characterized in that The reagents are primers and / or probes.

7. Use of the reagent according to any one of claims 5 to 6 in preparing a kit for predicting embryo quality.

8. A method for constructing an embryo quality prediction model, characterized in that: The construction method comprises: Data acquisition: including sample information data and experimental data; wherein, the sample information data uses the embryo quality as a grouping label, and the embryos are scored according to the Gardner blastocyst scoring system. Blastocysts with a blastocyst cavity score ≥ 3, and the number and arrangement of the inner cell mass cells are grade B or above, and the number and structure of the trophoblast cells are grade B or above are defined as high-quality embryos, otherwise they are low-quality embryos; the experimental data are the expression amount data of biomarkers in the embryo culture medium; wherein, the nucleotide sequence of the biomarker is shown in SEQ ID NO.1-3, or the nucleotide sequence of the biomarker is shown in SEQ ID NO.1-8, or the nucleotide sequence of the biomarker is shown in SEQ ID NO.1-13; Data analysis: standardizing and normalizing the expression data to obtain processed data; Model training and optimization: The processed data is input into a machine learning model for training, and the model parameters are adjusted through multiple cycles to obtain an embryo quality prediction model.

9. An embryo quality prediction model, characterized in that: The embryo quality prediction model is constructed by the method according to claim 8.

10. A method for predicting embryo quality, characterized in that: The prediction method comprises: Data acquisition: obtaining the expression data of the biomarker in the embryo culture medium; wherein the nucleotide sequence of the biomarker is as shown in SEQ ID NO.1-3, or the nucleotide sequence of the biomarker is as shown in SEQ ID NO.1-8, or the nucleotide sequence of the biomarker is as shown in SEQ ID NO.1-13; Data analysis: After acquisition, the expression data is standardized and normalized to obtain processed data; Identification: inputting the processed data into the embryo quality prediction model as claimed in claim 9 to obtain a prediction result of the embryo quality in the embryo culture medium; Output: Output the recognition results.

Images