Omics data capture and reconstruction method based on JDLGR algorithm

Abstract

The invention discloses an omics data capture and reconstruction method based on a JDLGR algorithm, and belongs to the technical field of life science. The objective of the invention is to solve the problem of poor accuracy of a recovery result in reconstruction of existing omics data. According to the method, high-dimensional data and low-dimensional data about omics data are used for forming joint data Xc, and the Xc is used for training a joint dictionary Dc; and performing omics data recovery by using the trained joint dictionary Dc through low-dimensional omics data Y measured by an experiment, and in the recovery process, recovering omics information X * of high-dimensional data by using Y, or high-dimensional gene data blocks x * by using partitioned data y of Y, and taking the data blocks x * as a column vector of a high-dimensional data matrix X *, so as to finally obtain the omics information X * of the high-dimensional data.

Description

Technical Field

[0001] The present invention belongs to the field of life science technology, and specifically relates to a method for capturing and reconstructing multi-omics data of different cells. Background Art

[0002] With the advancement of life science technology, studying cell states based on multi-omics data has become a more accurate research method. By integrating different omics data such as genomics, transcriptomics, proteomics and metabolomics, a more comprehensive description of cell states can be provided. In terms of disease treatment, drugs are used to regulate the expression of genes, proteins and metabolites in cells to achieve multi-level regulation and thus achieve therapeutic goals. Therefore, fast, low-cost and accurate acquisition and sequencing of multi-omics data is a key technology to achieve this goal.

[0003] Common multi-omics data acquisition methods include gene sequencing, mass spectrometry, and nuclear magnetic resonance spectroscopy. For example, gene expression is measured by counting all RNA sequences in cells through gene sequencing technology (such as NGS), and the gene expression is calculated through data processing; proteomics uses mass spectrometry technology to identify and quantify protein structural information and the detailed composition of peptide chains; and metabolomics uses mass spectrometry or nuclear magnetic resonance spectroscopy to analyze the types and concentrations of all metabolites.

[0004] Current omics data sequencing methods mainly include traditional sequencing methods and dimensionality reduction observation methods. Traditional methods require sequencing of all omics data. This method needs to be sent to a professional company for testing. The process is complicated, the workload is huge, and it often takes several weeks, which is extremely inconvenient. In order to improve sequencing efficiency, dimensionality reduction observation methods were proposed. Dimensionality reduction observation reduces the amount of detection and shortens the detection time by downsampling the data. However, it is difficult to restore the sequencing results of the original data from the sequencing results of a small part of the data. At present, this type of method generally has problems such as low data credibility, complex experimental design process, inability to simultaneously retain the linear and nonlinear information of genetic data, and high cost, which limits the actual application scenarios of this method.

[0005] Traditional methods are time-consuming and costly, but single-sequencing methods are mature and have high accuracy; dimensionality reduction observation methods are time-consuming, but have low accuracy and complex algorithms. If you want to reduce sequencing costs and improve efficiency, you must reduce the amount of sequencing. Dimensionality reduction observation methods have already achieved significant results in the field of image processing. Therefore, the top priority is to find the essence of the inadaptability of dimensionality reduction observation methods in the biological field and its solution.

[0006] By summarizing a large amount of experimental experience, many problems faced by the above-mentioned dimensionality reduction observation method in the biological field are essentially caused by the high randomness of the biological environment, which is mainly reflected in the following two aspects:

[0007] (1) Biological stochasticity problem. Biological stochasticity refers to the randomness and uncertainty inherent in biological systems, which causes individuals to show differences from expected values ​​in gene expression, cell behavior, and disease response.

[0008] (2) High off-target rate problem. When using biosensors to detect substances in the biological environment through biochemical reactions, signal crosstalk may occur, such as the detection results of non-specific sensors containing substances that are not desired to be detected, and mutual interference between sensors, etc. These problems bring high uncertainty to sequencing.

[0009] The high randomness of biological problems places higher demands on algorithm robustness than in other fields. If mature algorithms from other fields are mechanically applied without considering the particularities of biological problems, the accuracy of the recovery results will be difficult to meet the standards. Summary of the invention

[0010] The present invention aims to solve the problem of poor accuracy of restoration results in the existing reconstruction of omics data.

[0011] A method for capturing and reconstructing omics data based on the JDLGR algorithm, using the trained joint dictionary D c Perform omics data recovery;

[0012] The joint dictionary D c The training process includes the following steps:

[0013] Using high-dimensional data X h , low-dimensional data X l Composition of joint data X c , high-dimensional data X h is a matrix, each column vector in the matrix represents the omics data contained in a sample, and each row vector represents the data of the same omics feature in different samples; low-dimensional data X l X is selected by random sampling h The total number of rows is extracted;

[0014] Using a high-dimensional dictionary D h , low-dimensional dictionary D l Composition of joint dictionary D c , for the joint dictionary D c Conduct training. The goals of the training process are as follows:

[0015]

[0016] in, is the Lagrange operator; α is the sparse representation vector;

[0017] The trained joint dictionary D cThe process of recovering omics data is based on the trained joint dictionary D c , recover the omics information X of high-dimensional data from the low-dimensional data Y of full sequencing * In this process, firstly based on formula (2), using D l Solve for α with Y * ; get α * Then use α * and D h Calculate X h * ;

[0018]

[0019] Among them, α * is a sparse representation vector;

[0020] The low-dimensional data Y of the whole sequencing is the low-dimensional omics data measured by experiments. The specific omics features contained in Y are combined with the training joint dictionary D c The low-dimensional data X used l of the same type and size.

[0021] Furthermore, high-dimensional data X h Based on the original omics data X, that is, the original omics data X is normalized and then decentralized, where the decentralized operation refers to X h The sample mean is subtracted from each sample.

[0022] Furthermore, the combined data X c , joint dictionary D c as follows:

[0023]

[0024] Among them, N and M are respectively h , X l The number of rows in the matrix.

[0025] Furthermore, for the joint dictionary D c The training process includes the following steps:

[0026] S121, Initialization D c and α;

[0027] Among them, D c After randomly assigning any non-zero value, decentralization and normalization operations are required; α is assigned to zero;

[0028] S122, the existing D c Substitute into formula (1), optimize the formula with α as the variable, and obtain the trained α;

[0029] S123, Substitute the existing α into formula (1), and use D c Optimize the formula for the variable and obtain the trained D c ;

[0030] S124, repeat S121-S123 until the number of cycles reaches the preset value, and finally obtain the trained joint dictionary D c .

[0031] Furthermore, the high-dimensional omics information X is recovered from the low-dimensional data Y of the whole sequencing. * The process includes the following steps:

[0032] S201, the low-dimensional dictionary component D l Substitute the low-dimensional data Y into formula (2), and use α * For the variable optimization formula (2), we can obtain the sparse representation vector α after training. * ;

[0033] S202, according to the relationship of formula (3), use α * and D h Solve and restore the data texture matrix X h ;

[0034] X h =D h α * (3)

[0035] S203, according to formula (4), X h Recovery centralization, obtain recovery data X h * :

[0036] X h * =X h +mean(Y) (4)

[0037] Among them, mean(Y) is the sample mean of the full sequencing low-dimensional data Y.

[0038] A method for capturing and reconstructing omics data based on the JDLGR algorithm, using the trained joint dictionary D c Perform omics data recovery;

[0039] The joint dictionary D c The training process includes the following steps:

[0040] Step 101: Obtain high-dimensional data X h , X his a matrix, each column vector in the matrix represents the omics data contained in a sample, and each row vector represents the data of the same omics feature in different samples;

[0041] For a high-dimensional data matrix X h For each column, cluster them by rows and arrange the results from high to low in terms of relevance;

[0042] Step 102: X h Each sample in the image, i.e. each column, is divided into n m×1 blocks of the same size, and each block is downsampled at the same position to obtain X l1 , X l2 ,......,X ln , the same position sampling means that the relative position of the sampling in each block is the same; these blocks formed after sampling are vertically spliced ​​as low-dimensional data blocks X l One column in , so that X l ;

[0043] Step 103: Use high-dimensional dictionary D h , low-dimensional dictionary D l Composition of joint dictionary D c , for the joint dictionary D c Conduct training. The goals of the training process are as follows:

[0044]

[0045] in, is the Lagrange operator; α is the sparse representation vector;

[0046] The trained joint dictionary D c The process of recovering omics data includes the following steps:

[0047] For Y, we start from the upper left corner and take m×1 blocks without overlap, which are recorded as y. The low-dimensional data Y of the whole sequencing is the low-dimensional omics data measured by experiments. The specific omics features contained in Y are consistent with the training joint dictionary D. c The low-dimensional data X used l The type and size are the same; for each y, the following processing is performed:

[0048] Step 201: Base D l and y, solve the optimization equation where α * is a sparse representation vector, is the Lagrange operator;

[0049] Step 202: Generate a high-dimensional data texture block x=D h α * , where α* is the result corresponding to the sparse representation vector obtained by the optimization equation in step 201, and x is the restored decentralized high-dimensional data block;

[0050] Step 203, calculating the high-dimensional data block x*=x+mean(y), where mean represents the mean;

[0051] Step 204: generate each data block x * As a high-dimensional data matrix X * A column vector of X * .

[0052] Furthermore, high-dimensional data X h Based on the original omics data X, that is, the original omics data X is normalized and then decentralized, where the decentralized operation refers to X h The sample mean is subtracted from each sample.

[0053] Furthermore, the combined data X c , joint dictionary D c as follows:

[0054]

[0055] Among them, N and M are respectively h , X l The number of rows in the matrix.

[0056] Furthermore, for the joint dictionary D c The training process includes the following steps:

[0057] Step 121: Initialize D c and α;

[0058] Among them, D c After randomly assigning any non-zero value in, decentralization and normalization operations are required; α is all assigned zero;

[0059] Step 122: Change the existing D c Substitute into formula (5), optimize the formula with α as the variable, and obtain the trained α;

[0060] Step 123: Substitute the existing α into equation (5) and use D c Optimize the formula for the variable and obtain the trained D c ;

[0061] Step 124: Repeat steps 121 to 123 until the number of cycles reaches a preset value, and finally obtain the trained joint dictionary D c .

[0062] Furthermore, in step 101, the high-dimensional data matrix X h In the process of clustering rows in each column, K-Means clustering algorithm is used for clustering.

[0063] Beneficial effects:

[0064] 1. The present invention combines the theoretical method of dimensionality reduction of omics data with the actual biochemical reaction process. For example, in transcriptomics, this method uses PCR reaction to reduce the dimension to obtain gene data, and recovers high-dimensional gene information from it to characterize the state of the organism.

[0065] 2. The biochemical reaction used in the present invention is a traditional amplification method (such as PCR amplification), which does not require special instruments for operation. For example, in transcriptomics, only common PCR amplification reactions need to be used during the experiment, and only a small number of genes need to be measured once, which greatly reduces the difficulty of the experiment and ensures the accuracy of the measurement results.

[0066] 3. The present invention provides a method for acquiring omics data in a short time, accurately and at a low cost. For example, in transcriptomics, the present invention performs sparse sampling measurement of genes to obtain the expression of cell genes. The cost of a single experiment is about 4%-8% of the second-generation gene sequencing (NGS); the experimental time is 2 hours, which is about 1% of NGS. The measurement accuracy meets the accuracy requirements for practical application scenarios such as tumor typing and cell state observation. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 Schematic diagram of dividing each sample into blocks and downsampling.

[0068] Figure 2 This is a schematic diagram of the recovery process in the second implementation mode.

[0069] Figure 3 Schematic diagram of PCR 96-well plate.

[0070] Figure 4 Schematic diagram of the data reconstruction and recovery process of the embodiment. DETAILED DESCRIPTION

[0071] The key technology to solve the problem in the background technology is to reduce the dimension of omics data, obtain omics data by conventional amplification mode, and establish an experimental process that meets the needs of practical applications. The present invention takes transcriptomics as an example (according to the central dogma, the method applicable to transcriptomics is actually also applicable to other omics data processing), establishes a fast acquisition method of cell gene expression, based on the modular expression of genes and with compressible properties, first trains the high and low dictionary pairs for gene data, records the modular information of genes in low-dimensional dictionaries, and records the gene data structure in each low-dimensional module in the corresponding position in the high-dimensional dictionary. Afterwards, by a small amount of single PCR measurements for determining genes, dimension reduction perceives the current cell state. The cell state obtained (such as low-dimensional data Y in transcriptomics) is combined with high and low dictionary decoding to decode the state (high-dimensional gene information) of the tested organism. The data obtained uses single gene recovery accuracy, gene enrichment analysis, and information transfer entropy as indicators to evaluate the reply effect, and the verification results illustrate that the multi-omics data measured and detected by this method can meet the perception needs for the current state of the organism.

[0072] The present invention is oriented to transcriptomics data, uses the method of training gene high and low dictionary pairs, performs a small amount of separate measurements on high-dimensional gene sparse sampling through single PCR reaction, and decodes to obtain high-dimensional gene expression. Specifically, the present invention first uses open source data to establish high and low dictionary pairs for genes, then determines the measured genes through a downsampling matrix, uses single PCR reaction (real-time PCR, qPCR, RNA-seq, microarray, etc.) to obtain gene observations, and finally quickly obtains cell gene expression, establishes the current cell state, and provides data guidance at the gene level for major diseases such as cancer, chronic diseases, and treatment, prognosis, and pathological analysis of major infectious diseases. The present invention is further described below in conjunction with specific embodiments. Specific implementation method one:

[0074] This embodiment is a method for capturing and reconstructing omics data based on the JDLGR algorithm. Taking transcriptomics as an example, it can realize an accurate, efficient and low-cost method for measuring cell gene expression. We use a random sampling method to measure about 4% of all genes separately, use the PCR reaction process to realize the observation process of dimensionality reduction data, and use the JDLGR algorithm (Joint Dictionaries Learning for Genomic data Recovery) to train gene dictionaries through public gene sequencing data sets, perform dictionary training and reconstruct cell gene expression, and finally achieve the requirements of reducing experimental costs, shortening measurement time, and improving recovery accuracy. According to the central dogma, the method applicable to transcriptomics can also be applied to other omics data processing.

[0075] According to sparse coding theory, data can be represented as the product of an overcomplete dictionary and a sparse representation vector. The basic principle of data recovery based on the JDLGR algorithm (Joint Dictionaries Learning for Genomicdata Recovery) is to train a joint dictionary with a small amount of high- and low-dimensional data, and then use the joint dictionary to recover all high-dimensional data from all sequenced low-dimensional data. The following will explain the basic dictionary training method, basic data recovery method, algorithm implementation and improvement, and its beneficial results.

[0076] A method for capturing and reconstructing omics data based on the JDLGR algorithm, comprising:

[0077] S100, training joint dictionary:

[0078] The dictionary training in the present invention is based on the optimization method, using high-dimensional data X h , low-dimensional data X l Composed of joint data X c , for the high-dimensional dictionary D h , low-dimensional dictionary D l The combined dictionary D c Before explaining the training process, the meaning of the physical quantity is explained first:

[0079] X h , X l Represent high-dimensional and low-dimensional data matrices, respectively. Each column vector in the high-dimensional and low-dimensional data matrices represents the omics data contained in a sample, and each row vector represents the data of the same omics feature in different samples. For example, in transcriptomics, omics data is the gene expression intensity, which is generally represented by the Ct value of qPCR amplification. Each column represents the data in a sample, and each row represents the data of a gene in different samples.

[0080] X h , X l The data used for training is obtained based on the preprocessing of the original omics data X: h is obtained by normalizing and decentralizing X, where decentralizing X h Each sample in X is subtracted from the sample mean; l By randomly selecting X h About 4% of the total number of rows is obtained.

[0081] Joint DataX c , joint dictionary D c It is obtained by vertically splicing the previous quantities, as follows:

[0082]

[0083] Among them, N and M are respectively h , X l The number of rows in the matrix. Its physical meaning in transcriptomics is X h , X l The number of genes contained in .

[0084] The specific process of training the dictionary is as follows:

[0085] Ideal joint dictionary D c and the corresponding sparse representation vector α should satisfy X c =D c α, and according to sparse coding theory, α should be sparse enough to obtain better recovery effect. Applying the Lagrange dual problem, the above objective can be transformed into the following optimization problem:

[0086]

[0087] in, It is the Lagrange operator, which is transformed from the Lagrange dual problem. This hyperparameter is set artificially.

[0088] The solution to this optimization problem is the dictionary training process. The training adopts alternating training D c The strategies of and α are as follows:

[0089] S121, Initialization D c and α;

[0090] Among them, D c After randomly assigning any non-zero value to α, decentralization and normalization operations are required; α is all assigned to zero.

[0091] S122, the existing D c Substitute into formula (1), optimize the formula with α as the variable, and obtain the trained α.

[0092] S123, Substitute the existing α into formula (1), and use D c Optimize the formula for the variable and obtain the trained D c .

[0093] S124, repeat S121-S123 until the number of cycles reaches the preset value, and finally obtain the trained joint dictionary D c .

[0094] S200, using the trained joint dictionary to recover omics data:

[0095] Data recovery is based on the trained joint dictionary Dc , recover the omics information X of high-dimensional data from the low-dimensional data Y of full sequencing * The core of this process is to first pass D l Solve for α with Y * , and then use α * and D h Calculate X h * Before explaining the recovery process, the meaning of the physical quantity is explained first:

[0096] The low-dimensional data Y of the whole sequencing is the low-dimensional omics data measured by experiments (such as the Ct value obtained by qPCR, the mass-to-nuclear ratio obtained by mass spectrometry, etc.). The specific omics features contained in Y are combined with the training joint dictionary D c The low-dimensional data X used l The type and size of are the same. Similar to formula (1), using D l Solve for α with Y * Satisfies the relationship shown in formula (2):

[0097]

[0098] The specific process of recovering omics data is as follows:

[0099] S201, the known joint dictionary D c The low-dimensional dictionary component D in l And the low-dimensional data Y is substituted into formula (2), with α * Optimize the formula for the variable and obtain the sparse representation vector α after training * , α * Same size as α.

[0100] S202, according to the relationship in formula (3), use α * and D h Solve and restore the data texture matrix X h .

[0101] X h =D h α * (3)

[0102] S203, according to formula (4), X h Recovery centralization, obtain recovery data X h * :

[0103] X h * =X h +mean(Y) (4)

[0104] Among them, mean(Y) is the sample mean of the full sequencing low-dimensional data Y.

[0105] After the above steps, the recovered data X is obtained. h * . Specific implementation method 2:

[0107] This embodiment is a method for capturing and reconstructing omics data based on the JDLGR algorithm. Based on the first specific embodiment, this embodiment proposes an optimization and improvement scheme:

[0108] Omics data has the characteristics of high throughput, that is, the amount of data is large, and the amount of data that needs to be restored at one time can reach millions. The iterative optimization algorithm used in the algorithm has high requirements for computing power and memory. If it is directly calculated without processing, it often runs slowly, and may even fail to complete due to insufficient memory.

[0109] In addition, if the data is not normalized or decentralized when training the dictionary, the high- and low-dimensional data matrix X may be used. h , X l There is no close connection between each row of data, that is, X l Each row as a feature may represent X h In the adjacent rows, X h The corresponding features of truly adjacent rows in the , may be very different, which will cause the trained sparse representation vector α to be not sparse enough and not meet the requirements of sparse coding theory. Therefore, the correlation between the restored data obtained by using the dictionary trained by the original algorithm and the original data is low. Sometimes the correlation of the experimental results is only about 33%, which is far from meeting the requirements of sequencing accuracy.

[0110] In order to solve the above problems, an improved JDLGR algorithm based on consensus clustering and block training is proposed on the basis of the first embodiment. This algorithm is mainly improved in the following aspects:

[0111] 1. Perform consensus clustering on the training data and arrange the results from high to low according to the correlation (evaluated by Pearson coefficient or Spearman coefficient). This ensures that the correlation between adjacent rows is high and the features are fully extracted;

[0112] 2. Perform block training on the joint dictionary, group amplify Y, and restore the data in blocks. This not only reduces the memory usage, but also avoids interference with the results caused by crossing too large a dimension.

[0113] The improved JDLGR algorithm involves two major processes: dictionary training and data recovery. The specific process is as follows:

[0114] 1. The specific process of training the dictionary is as follows:

[0115] Step 101: for the high-dimensional data matrix X h For each column, the rows are clustered by consensus using the K-Means clustering algorithm, and the results are arranged from high to low according to the correlation (evaluated by Pearson coefficient or Spearman coefficient);

[0116] Step 102: X h Each sample in each column is divided into n m×1 blocks of the same size (blocks are Figure 1 Group1, Group2, etc.), for each block, downsample at the same position to obtain X l1 , X l2 ,......,X ln , the same position sampling means that the relative position of the sampling in each block is the same. For example, in the first block, the 2nd, 4th, 9th, 10th, 11th, 13th, and 14th rows are selected for sampling, and for other blocks, the 2nd, 4th, 9th, 10th, 11th, 13th, and 14th rows are also sampled. The blocks formed after these samplings are vertically spliced ​​as low-dimensional data blocks X l A column in, that is, a sample, is operated as follows Figure 1 As shown;

[0117] Step 103: train the joint dictionary D according to the dictionary training method in S100 c , the specific steps are as follows: training data preprocessing

[0118] X h , X l Normalize X h Decentralize the process. h , X l Use the following formula to splice into joint data X c At the same time, according to the training data size and the set dictionary feature number, generate the corresponding size D h , D l , and then spliced ​​into D according to the following formula c

[0119]

[0120] Optimizing training of joint dictionary

[0121] According to formula (5), D is trained alternately. c and α, the specific steps are as follows:

[0122]

[0123] Step 121: Initialize D c and α;

[0124] Among them, D c After randomly assigning any non-zero value to α, decentralization and normalization operations are required; α is all assigned to zero.

[0125] Step 122: Change the existing D c Substitute into formula (5), optimize the formula with α as the variable, and obtain the trained α.

[0126] Step 123: Substitute the existing α into equation (5) and use D c Optimize the formula for the variable and obtain the trained D c .

[0127] Step 124: Repeat steps 121 to 123 until the number of cycles reaches a preset value, and finally obtain the trained joint dictionary D c .

[0128] It should be noted that the training process in this scheme is a whole training, and the recovery is done in blocks. The recovery process is as follows Figure 2 shown.

[0129] 2. Use the trained joint dictionary to recover omics data:

[0130] For Y, start from the upper left corner and take m×1 blocks without overlapping, record them as y, and perform the following operations for each y:

[0131] Step 201: Based on the low-dimensional dictionary D l and y, solve the optimization equation where α * is a sparse representation vector, is the Lagrange operator;

[0132] Step 202: Generate a high-dimensional data texture block x=D h α * , where α * is the result of the sparse representation vector obtained by optimizing the equation in step 201, and x is the recovered decentralized high-dimensional data block;

[0133] Step 203: Calculate the high-dimensional gene data block x * =x+mean(y), where the mean method is used to find the mean;

[0134] Step 204: generate each data block x * As a high-dimensional data matrix X * A column vector of X is finally returned * .

[0135] Example

[0136] The omics data is captured and reconstructed in the manner of the second specific implementation method. The specific process is as follows:

[0137] Take transcriptomics data qPCR as an example to design the reaction experiment process:

[0138] Design PCR primers for the genes corresponding to the designed observation matrix, and customize the designed upstream and downstream PCR primers to be assembled into the same 96-well plate, such as Figure 3 As shown, the primer concentration in a single well is 10 nmol / ul. At this point, the primer preparation process is completed.

[0139] Through qPCR (PCR) reaction, a qPCR instrument (PCR instrument) is used for amplification process. The number of amplification rounds is generally selected as 30 or 40 rounds, and the entry type data of the genetic data can be obtained, which is recorded as the reduced dimension observation value of the genetic data and recorded in the observation result matrix y (m, 1).

[0140] Then, the omics data is captured and reconstructed based on the JDLGR algorithm, such as Figure 4 The specific process is shown as follows:

[0141] (1) Dictionary training:

[0142] Use the JDLGR algorithm to perform dictionary training. During the training process, for a set of genetic data X(M*N), M is the number of genes and N is the number of samples. First, according to step 101 in the JDLGR dictionary training algorithm, the genes are arranged according to the strength of the correlation through the consensus clustering method based on K-Means to obtain the genetic data set X_sort(M*N). Then, according to step 102 in the JDLGR dictionary training algorithm, the data X_sort is divided into x1, x2..., and x1, x2... are downsampled in the same way (same relative position) to obtain the observation values ​​y1, y2..., and then the dictionary module is trained to obtain the joint dictionary D c The effect of dictionary training is evaluated by Pearson correlation and Spearman correlation, and the number of blocks is adjusted according to the evaluation results.

[0143] (2) Data recovery:

[0144] According to step 201 of the JDLGR data recovery algorithm, the gene Y to be measured is obtained, single-plex PCR primers are designed and the observed value Y is obtained through single-plex PCR experiment, and then according to step 202 of the JDLGR data recovery algorithm, the recovery algorithm is used to solve the recovered gene set X.

[0145] The main process of data recovery is as follows: Divide each sample y in Y into k blocks of equal size 1 ,y 2 ,......,yi ,......,y k , for each block y i The corresponding sparse representation vector can be obtained according to step 201 Through step 202, the restored decentralized high-dimensional gene data block x can be obtained. i , and then obtain the high-dimensional gene data block through step 203 x 1 , x 2 ,......,x i ,......,x k Vertical splicing to restore high-dimensional omics data x * . Each recovered omics data x * By performing horizontal splicing according to step 204 , the restored omics feature set X can be obtained.

[0146] Taking transcriptomics and proteomics as examples, the following specific embodiments and their effects are described.

[0147] Embodiment 1:

[0148] The gene amplification method of the present invention uses a PCR instrument or a qPCR instrument produced by ThermoFisher for gene amplification, which includes a matching 96-well plate, a 384-well plate and a fluorescence value reading function. The traditional cell RNA extraction method is used for extraction, and then a 20ul reverse transcription system and a PCR instrument are configured to reverse transcribe the gene to obtain the corresponding cDNA system, and then a 20ulqPCR reaction system is configured. After the configuration is completed, the sampled genes are added to the reaction system at the corresponding position of the 96-well plate or the 384-well plate, and the qPCR amplification process is started.

[0149] Using the public data from websites such as TCGA and microarray, the high and low dictionary pairs of genes are established in advance. The present invention can establish dictionaries for all cells with public data sets. During the establishment process, the JDLGR algorithm (Joint Dictionaries Learning for Genomic data Recovery) is used for dictionary training. The trained dictionary can be classified according to different types of cells for subsequent use.

[0150] After the qPCR reaction is completed, the Ct value in the well is obtained and the observed value is recorded according to the recording method. Then the JDLGR algorithm (Joint Dictionaries Learning for Genomicdata Recovery) is used to reproduce the gene expression profile.

[0151] In this embodiment, the qPCR amplification process can give a relatively accurate Ct value. Under the premise of a compressed sensing sampling rate of about 6%-10%, the present invention theoretically has a correlation of 75%-95% between the reproducing gene expression and the original data. For different samples, the Pearson correlation coefficient is above 70%, and the Spearman correlation coefficient is above 65%. The effect of reproducing gene expression profile is good.

[0152] For commonly used PCR and qPCR instruments, 96-well plates and 384-well plates can be used to perform amplification experiments on the corresponding number of reaction wells at the same time. The number of genes in the reproducible gene expression profile is more than 10,000, and the overall experimental time is within 2-3 hours, which can meet the requirements of gene expression profile reproducibility and real-time performance in most practical applications.

[0153] Embodiment 2:

[0154] In this embodiment, the amplification process of the PCR and qPCR instruments can be changed to the amplification process of the magnetic beads connected to the primers. It is only necessary to connect the gene primers in the reaction wells to the magnetic beads, and the rest can be carried out according to the above experimental scheme.

[0155] Embodiment 3:

[0156] In the present invention, the amplification process of PCR and qPCR instruments can be changed to an amplification process using microarray droplets. It is only necessary to drop the reaction system containing gene primers in the reaction wells on the glass sheet in the form of a microarray, and the rest can be carried out according to the above experimental scheme.

[0157] Embodiment 4:

[0158] In the present invention, the amplification process of the PCR and qPCR instruments can be changed to an amplification process using a T7 promoter. It is only necessary to connect the front end or the rear end of the gene primer in the reaction well to the T7 promoter, and then amplify by linear amplification. The rest can be carried out according to the above experimental scheme.

[0159] Embodiment 5:

[0160] In the present invention, the amplification process of PCR and qPCR instruments can be replaced by an amplification process using NGS or microarray. The measured gene only needs to be sent to a gene sequencing company, and then the gene expression amount can be measured using the corresponding instrument. The rest can be carried out according to the above experimental scheme.

[0161] Embodiment 6:

[0162] According to the central dogma, there is a correspondence between various RNAs in transcriptomics and various proteins in proteomics. Therefore, the methods applicable to transcriptomics can also be applied to proteomics data analysis.

[0163] The above calculation examples of the present invention are only used to explain the calculation model and calculation process of the present invention in detail, and are not intended to limit the implementation methods of the present invention. For ordinary technicians in the relevant field, other forms of changes or modifications can be made based on the above description. It is impossible to list all the implementation methods here. All obvious changes or modifications derived from the technical solution of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for capturing and reconstructing omics data based on the JDLGR algorithm, characterized in that: Using the trained joint dictionary D c Perform omics data recovery; The joint dictionary D c The training process includes the following steps: Using high-dimensional data X h , low-dimensional data X l Composition of joint data X c , high-dimensional data X h is a matrix, each column vector in the matrix represents the omics data contained in a sample, and each row vector represents the data of the same omics feature in different samples; low-dimensional data X l X is selected by random sampling h The total number of rows is extracted; Using a high-dimensional dictionary D h , low-dimensional dictionary D l Composition of joint dictionary D c , for the joint dictionary D c Conduct training. The goals of the training process are as follows: in, is the Lagrange operator; α is the sparse representation vector; The trained joint dictionary D c The process of recovering omics data is based on the trained joint dictionary D c , recover the omics information X of high-dimensional data from the low-dimensional data Y of full sequencing * In this process, firstly based on formula (2), using D l Solve for α with Y * ; get α * Then use α * and D h Calculate X h * ; Among them, α * is a sparse representation vector; The low-dimensional data Y of the whole sequencing is the low-dimensional omics data measured by experiments. The specific omics features contained in Y are combined with the training joint dictionary D c The low-dimensional data X used l of the same type and size.

2. The method for capturing and reconstructing omics data based on the JDLGR algorithm according to claim 1, characterized in that: High dimensional dataX h Based on the original omics data X, that is, the original omics data X is normalized and then decentralized, where the decentralized operation refers to X h The sample mean is subtracted from each sample.

3. The method for capturing and reconstructing omics data based on the JDLGR algorithm according to claim 1 or 2, characterized in that: Joint DataX c , joint dictionary D c as follows: Among them, N and M are respectively h , X l The number of rows in the matrix.

4. The method for capturing and reconstructing omics data based on the JDLGR algorithm according to claim 3, characterized in that: For the joint dictionary D c The training process includes the following steps: S121, Initialization D c and α; Among them, D c After randomly assigning any non-zero value, decentralization and normalization operations are required; α is assigned to zero; S122, the existing Dc 代 Enter formula (1), optimize the formula with α as the variable, and obtain the trained α; S123, Substitute the existing α into formula (1), and use D c Optimize the formula for the variable and obtain the trained D c ; S124, repeat S121-S123 until the number of cycles reaches the preset value, and finally obtain the trained joint dictionary D c .

5. The method for capturing and reconstructing omics data based on the JDLGR algorithm according to claim 4, characterized in that: Recovering high-dimensional omics information X from low-dimensional data Y of full sequencing * The process includes the following steps: S201, the low-dimensional dictionary component D l Substitute the low-dimensional data Y into formula (2), and use α * For the variable optimization formula (2), we can obtain the sparse representation vector α after training. * ; S202, according to the relationship in formula (3), use α * and D h Solve and restore the data texture matrix X h ; X h =D h a * (3) S203, according to formula (4), X h Recovery centralization, obtain recovery data X h * : X h * =X h +mean(Y) (4) Among them, mean(Y) is the sample mean of the full sequencing low-dimensional data Y.

6. A method for capturing and reconstructing omics data based on the JDLGR algorithm, characterized in that: Using the trained joint dictionary D c Perform omics data recovery; The joint dictionary D c The training process includes the following steps: Step 101: Obtain high-dimensional data X h , X h is a matrix, each column vector in the matrix represents the omics data contained in a sample, and each row vector represents the data of the same omics feature in different samples; For a high-dimensional data matrix X h For each column, cluster them by rows and arrange the results from high to low in terms of relevance; Step 102: X h Each sample in the matrix, i.e. each column, is divided into n m×1 blocks of the same size, and each block is downsampled at the same position to obtain X l1 , X l2 , ..., X ln , the same position sampling means that the relative position of the sampling in each block is the same; these blocks formed after sampling are vertically spliced ​​as low-dimensional data blocks X l One column in , so that X l ; Step 103: Use high-dimensional dictionary D h , low-dimensional dictionary D l Composition of joint dictionary D c , for the joint dictionary D c Conduct training. The goals of the training process are as follows: in, is the Lagrange operator; α is the sparse representation vector; The trained joint dictionary D c The process of recovering omics data includes the following steps: For Y, we start from the upper left corner and take m×1 blocks without overlap, which are recorded as y. The low-dimensional data Y of the whole sequencing is the low-dimensional omics data measured by experiments. The specific omics features contained in Y are consistent with the training joint dictionary D. c The low-dimensional data X used l The type and size are the same; for each y, the following processing is performed: Step 201: Base D l and y, solve the optimization equation where α * is a sparse representation vector, is the Lagrange operator; Step 202: Generate a high-dimensional data texture block x=D h α * , where α * is the result corresponding to the sparse representation vector obtained by the optimization equation in step 201, and x is the restored decentralized high-dimensional data block; Step 203: Calculate the high-dimensional data block x * =x+mean(y), where mean means the mean; Step 204: generate each data block x * As a high-dimensional data matrix X * A column vector of X * .

7. The method for capturing and reconstructing omics data based on the JDLGR algorithm according to claim 6, characterized in that: High dimensional dataX h Based on the original omics data X, that is, the original omics data X is normalized and then decentralized, where the decentralized operation refers to X h The sample mean is subtracted from each sample.

8. The method for capturing and reconstructing omics data based on the JDLGR algorithm according to claim 6 or 7, characterized in that: Joint DataX c , joint dictionary D c as follows: Among them, N and M are respectively h , X l The number of rows in the matrix.

9. The method for capturing and reconstructing omics data based on the JDLGR algorithm according to claim 8, characterized in that: For the joint dictionary D c The training process includes the following steps: Step 121: Initialize D c and α; Among them, D c After randomly assigning any non-zero value in, decentralization and normalization operations are required; α is all assigned zero; Step 122: Change the existing D c Substitute into formula (5), optimize the formula with α as the variable, and obtain the trained α; Step 123: Substitute the existing α into equation (5) and use D c Optimize the formula for the variable and obtain the trained D c ; Step 124: Repeat steps 121 to 123 until the number of cycles reaches a preset value, and finally obtain the trained joint dictionary D c .

10. The method for capturing and reconstructing omics data based on the JDLGR algorithm according to claim 9, characterized in that: In step 101, the high-dimensional data matrix X h In the process of clustering rows in each column, K-Means clustering algorithm is used for clustering.

Images