Bacterial strain analysis method, analysis device and storage medium

Abstract

The present disclosure relates to an analysis method, an analysis device and a storage medium for bacterial strains. The method includes: obtaining pan-genome data of a target bacterial species, including reference gene sequences of multiple bacterial species; obtaining sequencing data of multiple samples (including multiple bacterial species); comparing the sequencing data of each sample with the pan-genome data to obtain comparison counts relative to each reference gene; merging the comparison counts of similar genes into comparison counts of gene families to determine a first matrix, wherein each element of the first matrix represents information related to comparison counts of each gene family in each sample; decomposing the first matrix into a second and a third matrix, wherein each element of the second matrix represents information related to the presence of each gene family in each strain, and each element of the third matrix represents information related to the relative abundance of each strain in each sample. The composition ratio of each strain of multiple bacterial species in each sample and the gene family inclusion of each strain can be accurately obtained.

Description

Technical Field

[0001] The present disclosure relates to an analysis method and an analysis device for microbial information, and more specifically, to an analysis method, an analysis device and a storage medium for bacterial strains implemented by using a computer. Background Art

[0002] The study of the physiological functions of bacterial communities has entered the strain era. More and more studies have found that different strains of the same species have huge differences in function and physiological effects, but the genomic research of these strains is seriously lagging behind. Traditional strain genome analysis requires the isolation and cultivation of bacteria from the sample, and then sequencing the colonies to understand the genome sequence of the strain. Among them, the isolation and cultivation of the strain of interest is a long trial-and-error process, which is also the bottleneck of traditional strain genome analysis.

[0003] With the popularization of high-throughput sequencing technology and the emergence of metagenomic analysis technology, researchers can directly extract bacterial community composition, gene family and functional information from sample sequencing data through bioinformatics analysis tools. This process does not rely on experimental cultivation, which greatly accelerates the progress of microbiome research. Earlier, the analysis of metagenomic strains mainly adopted non-reference (de novo) schemes. However, due to various errors in assembly, the accuracy of metagenomic strain analysis is poor.

[0004] Strain analysis methods based on genomes of reference databases (with references) are gradually emerging. One type of strain analysis method is based on single nucleotide polymorphisms (SNPs), which find mutation sites on conserved gene fragments from the data and use them to determine the strain composition of the species. This type of method has been increasingly used in metagenomic analysis and applied to scenarios such as environmental pathogen tracing and fecal microbiota transplantation (FMT) strain tracking. However, this type of strain analysis method cannot reflect the overall genetic information of the strain. For example, these conserved genes on conserved gene fragments sometimes do not reflect the specific functions of the strains in the species.

[0005] Another type of strain analysis method, represented by PanPhlAn, uses the pan-genome of the strain to determine the genetic composition of the strain based on the presence or absence of genes on the pan-genome of the strain. Compared with the first type of method, PanPhlAn provides the genetic composition of the strain, allowing researchers to conduct in-depth interpretation and analysis of the strain from the functional level of the gene. However, PanPhlAn only extracts the genetic composition of a single type of strain that is dominant. Since multiple strains of the same species often exist at the same time in the sample, and the strain composition changes according to time or sampling conditions, changes in the strain ratio may lead to dynamic changes in the dominant strain. Sometimes, the dominant strain only has a low-level advantage and will change under normal disturbances. PanPhlAn cannot reflect the composition of multiple strains, nor can it detect whether the change in the dominant strain is caused by normal disturbances or by drastic changes in the strain's microecological environment. It has considerable limitations in strain analysis. Summary of the invention

[0006] The present disclosure is provided to solve the above-mentioned problems existing in the prior art.

[0007] A bacterial strain analysis method, an analysis device and a storage medium are needed, which can analyze the strains from a pan-genome level, accurately obtain the composition ratio of each strain of multiple species in each sample and the gene family inclusion of each strain, so as to comprehensively analyze the functions of each strain of multiple species in each sample.

[0008] According to the first aspect of the present disclosure, a method for analyzing bacterial strains is provided. The analysis method may include obtaining pan-genome data of a target bacterial species (the pan-genome data may include reference gene sequences of multiple strains of the target bacterial species), and obtaining sequencing data of multiple samples, each sample including multiple bacterial species. The analysis method may include, via at least one processor, comparing the sequencing data of each sample with the pan-genome data of the target bacterial species to obtain a comparison count relative to each reference gene in the pan-genome, and merging the comparison counts of similar genes into the comparison counts of gene families to determine a first matrix. Each element of the first matrix may characterize the comparison count related information of each gene family in each sample. The analysis method may further include, via the at least one processor, decomposing the first matrix into a second matrix and a third matrix. Wherein, each element of the second matrix may characterize the presence related information of each gene family in each strain, and each element of the third matrix may characterize the relative abundance related information of each strain in each sample.

[0009] According to a second aspect of the present disclosure, a bacterial strain analysis device is provided. The analysis device may include an interface and at least one processor. The interface may be configured to receive sequencing data of multiple samples, each sample including multiple bacterial species. The at least one processor may be configured to execute the bacterial strain analysis method according to each embodiment of the present disclosure.

[0010] According to a third aspect of the present disclosure, a bacterial strain analysis device is provided. The analysis device may include a first acquisition unit, a second acquisition unit, a comparison unit, a merging unit, and a decomposition unit. The first acquisition unit may be configured to acquire pan-genome data of a target bacterial species (the pan-genome data may include reference gene sequences of multiple strains of the target bacterial species). The second acquisition unit may be configured to acquire sequencing data of multiple samples, each sample including multiple bacterial species. The comparison unit may be configured to compare the sequencing data of each sample with the pan-genome data of the target bacterial species to obtain the comparison counts of each sample relative to each reference gene in the pan-genome. The merging unit may be configured to merge the comparison counts of similar genes into the comparison counts of gene families to determine a first matrix, wherein each element of the first matrix represents the comparison count related information of each gene family in each sample. The decomposition unit may be configured to decompose the first matrix into a second matrix and a third matrix. Among them, each element of the second matrix may represent the existence related information of each gene family in each strain, and each element of the third matrix may represent the relative abundance related information of each strain in each sample.

[0011] According to a fourth aspect of the present disclosure, a non-volatile computer-readable storage medium is provided, on which executable instructions are stored. When the executable instructions are executed by at least one processor, the bacterial strain analysis method according to each embodiment of the present disclosure is implemented.

[0012] By utilizing the bacterial strain analysis method, analysis device and storage medium according to the various embodiments of the present disclosure, it is possible to analyze the strains at a pan-genome level, accurately obtain the overall composition of each strain of multiple species in each sample and the gene family inclusion of each strain, thereby enabling a comprehensive analysis of the functions of each strain of multiple species in each sample. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In the drawings, which are not necessarily drawn to scale, the same reference numerals may describe similar components in different views. The same reference numerals with letter suffixes or different letter suffixes may represent different instances of similar components. The accompanying drawings generally illustrate various embodiments by way of example and not limitation, and together with the description and claims, serve to illustrate the disclosed embodiments. Such embodiments are illustrative and are not intended to be exhaustive or exclusive embodiments of the present apparatus or method.

[0014] Figure 1 A flowchart showing the process of the method for analyzing bacterial strains according to an embodiment of the present disclosure;

[0015] Figure 2 A flowchart showing a sub-process of decomposing a first matrix into a second matrix and a third matrix in a method for analyzing bacterial strains according to an embodiment of the present disclosure;

[0016] Figure 3 A flowchart showing an initialization sub-process in a method for analyzing bacterial strains according to an embodiment of the present disclosure;

[0017] Figure 4 A diagram showing the configuration of an analysis device for bacterial strains according to an embodiment of the present disclosure;

[0018] Figure 5 A block diagram showing an analysis device for bacterial strains according to an embodiment of the present disclosure;

[0019] FIG6( a ) shows a comparison diagram of the reference strain composition of 5 types of strains of 10 preset samples and the strain composition obtained by analyzing the 10 preset samples using the bacterial strain analysis method according to an embodiment of the present disclosure;

[0020] FIG6( b ) is a diagram showing a precision-recall curve of the inclusion of five types of strains obtained by analyzing 10 preset samples using the bacterial strain analysis method according to an embodiment of the present disclosure;

[0021] Figure 7(a)-Figure 7(f) A diagram showing the application of the bacterial strain analysis method according to an embodiment of the present disclosure to a fecal microbiota transplantation (FMT) dataset to evaluate the composition changes of Faecalibacterium prausnitzii (also known as F. prausnitzii) strains of four donors and their respective recipients at various time points before and after FMT; and

[0022] Figure 8 Show Figure 7(a)-Figure 7(f) The genes of the three strains of Faecalibacterium prausnitzii shown in the figure correspond to the pathway functions. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the technical solution of the present disclosure, the present disclosure is described in detail in conjunction with the accompanying drawings and specific embodiments. The embodiments of the present disclosure are further described in detail in conjunction with the accompanying drawings and specific embodiments, but are not intended to be limiting of the present disclosure. The terms "first", "second" and "third" used in the present disclosure are merely intended to distinguish the corresponding features, and do not represent the need for such an order, nor do they necessarily represent only the singular form. The "pan-genome data" used in the present disclosure should be broadly interpreted as "including reference gene sequences of multiple types of strains of the target species", that is, it can be understood as including any reference gene sequences of any type of strains of any number of target species.

[0024] Figure 1 FIG. 1 is a flow chart showing the process of the method for analyzing bacterial strains according to an embodiment of the present disclosure. Figure 1 As shown, the process can start from step 101 and step 102. In step 101, the pan-genome data of the target bacterial species can be obtained, and the pan-genome data can include reference gene sequences of multiple strains of the target bacterial species. In some embodiments, it is possible to connect to a pan-genome database via a network so as to obtain the pan-genome data of the target bacterial species therefrom and download it locally for invocation. In some embodiments, the genome sequences of various strains of the target bacterial species can also be obtained, and based on this, the pan-genome data of the target bacterial species can be generated for use. In step 102, sequencing data of multiple samples (such as but not limited to DNA sequencing data in fastq format, etc.) can be obtained, each sample may include multiple bacterial species, and each bacterial species may have multiple strains.

[0025] In step 103, the sequencing data of each sample is compared with the pan-genome data of the target bacterial species via at least one processor to obtain the comparison counts of each sample relative to each reference gene in the pan-genome. By using the pan-genome data of the target bacterial species (especially the gene sequence of each reference gene in the pan-genome) as a comparison reference, the comprehensive gene information of each sample can be accurately reflected, and sample-level analysis can be performed at the pan-genome level.

[0026] In step 104, the alignment counts of similar genes can be merged into the alignment counts of gene families to determine the first matrix. Each element of the first matrix represents the alignment count related information of each gene family in each sample. For the convenience of description, the first matrix is ​​also referred to as D matrix in the present disclosure. A gene family is a group of related genes with structural and functional similarity sequences generated by a common ancestral gene through duplication and mutation; similar genes are considered to belong to the same gene family in the present disclosure. Similar genes can be selected according to at least one of gene name, gene annotation, sequence similarity, etc. It can be considered that similar genes belong to the same gene family. By merging the alignment counts of each group of similar genes, the alignment count related information of the gene family level in each sample can be obtained, which significantly reduces the dimension of the calculation matrix and enhances the robustness of the distribution law of the alignment count related information compared to the alignment count related information of the gene level before the merger, thereby laying a foundation for the decomposition stability of step 105.

[0027] In step 105, the D matrix (i.e., the first matrix) can be decomposed into a second matrix and a third matrix via the at least one processor. Each element of the second matrix represents the presence-related information of each gene family in each type of strain, and each element of the third matrix represents the relative abundance-related information of each type of strain in each sample. For ease of description, the second matrix is ​​also referred to as the P matrix and the third matrix is ​​referred to as the S matrix in the present disclosure. The P matrix can also be referred to as the strain gene family spectrum, and the S matrix can also be referred to as the strain abundance spectrum. The inventor creatively found that by determining in step 104 that each element represents the comparison count-related information of each gene family in each sample, various methods can be used to iterate to obtain stable P matrices and S matrices. By decomposing the obtained P matrix and S matrix, the presence or absence of each gene family in each type of strain can be accurately obtained, and the relative abundance of each type of strain in each sample can be accurately obtained, so that the functional conditions of various strains under multiple species in each sample can be comprehensively analyzed.

[0028] In some embodiments, the rows of the D matrix may correspond to each gene family and the columns may correspond to each sample, and accordingly, the rows of the P matrix may correspond to each gene family and the columns may correspond to each strain, and the rows of the S matrix may correspond to each strain and the columns may correspond to each sample. Of course, the arrangement of the rows and columns of the D matrix, the P matrix, and the S matrix is ​​not limited thereto, for example, each of the matrices may be transposed, etc.

[0029] In some embodiments, each element of the P matrix may be binary processed so that the existence-related information is binary information indicating whether or not, for example, but not limited to, using 1 to indicate existence and 0 to indicate absence. In this way, the P matrix may be sparse, that is, the decomposition of the D matrix may be converted into a sparse decomposition, thereby reducing the computational load and speeding up the analysis.

[0030] Figure 2 FIG. 1 is a flowchart showing a sub-process of decomposing a D matrix into a P matrix and an S matrix in a method for analyzing bacterial strains according to an embodiment of the present disclosure. Figure 2 As shown, after obtaining the D matrix, the S matrix can be initialized (step 202) to generate an initial S matrix. The initial S matrix can be generated based on information of the D matrix and the initial P matrix. For example, the initial S matrix can be obtained by multiplying the D matrix with the inverse matrix of the initial P matrix, or by regression (for example, but not limited to linear regression). Specifically, the initial S matrix can be obtained by solving the following optimization equation (1) through linear regression.

[0031]

[0032] Among them, D ij represents the element in the i-th row and j-th column of the D matrix, P ik represents the element in the i-th row and k-th column of the P matrix, S kj Represents the element in the kth row and jth column of the S matrix. The inventors have found that the D matrix has good linear decomposability relative to the P matrix and the S matrix defined above, and the initial S matrix can be stably decomposed using the optimization equation defined by formula (1) through linear regression with less computational burden.

[0033] The initial S matrix (hereinafter referred to as S ini ) can be substituted to implement alternating iterative solutions of the P matrix and the S matrix (see steps 203 and 204) until convergence (see step 209) or the preset number of iterations is reached. Figure 2 Step 203 ("estimate the P matrix") and step 204 ("estimate the S matrix") are shown in sequence, but it should be noted that this is not limited to this. The order of estimation can also be swapped as long as the P matrix and the S matrix are estimated alternately and iteratively.

[0034] In some embodiments, at each iteration step, the P matrix may be solved in the following manner: For ease of description, the P matrix and S matrix obtained by solving each iteration step are represented as P and S, respectively, and the intermediate matrices related to P and S used in each iteration step are distinguished by superscripts.

[0035] The fourth matrix P′ can be determined according to formula (2):

[0036] P′=DS′ -1 Formula (2)

[0037] Among them, S' represents the current S matrix. Figure 2 Taking the alternating iterative solution order of the P matrix and the S matrix shown in as an example, the current S matrix S′ used in the nth (n is a natural number greater than 1) iteration step is the S matrix obtained at the end of the n-1th iteration step, and the current S matrix S′ used in the first iteration step is the initial S matrix S ini .

[0038] The fourth matrix P' can be used as an intermediate matrix for P calculation. The difference is that each element of P' is continuous data, and the elements of each column can characterize the quantitative distribution information of each gene family relative to the corresponding class of strains. The probability density function associated with each gene family in each column can be determined. The obtained probability density function has a low-order distribution peak (i.e., a peak close to 0) and a non-low-order distribution peak (usually corresponding to the gene family actually existing in the strain). The inventor found that the low-order distribution peak reflects noise information, and the elements corresponding to the low-order distribution peak of the corresponding probability density function can be removed from each column of the fourth matrix P', thereby effectively filtering out the noise. The gene family corresponding to the so-called noise information should not actually exist in the strain, but is mistakenly introduced due to errors or disturbances; by filtering out the noise in a targeted and efficient manner, the failure rate of D matrix decomposition can be significantly reduced. Then, each element in the fourth matrix P' after filtering out the noise can be compared with the first threshold, for example, a non-zero value (such as 1) is assigned to the elements exceeding the first threshold and a zero value is assigned to other elements, thereby obtaining a binary P matrix. The binary processing based on the first threshold not only facilitates improving the efficiency of the subsequent iterative solution of S, but also helps to improve the robustness of P and further improve the stability of the D matrix decomposition.

[0039] For the nth iteration step, after obtaining P, the S matrix can be solved by a regression method based on the D matrix and the currently obtained P matrix (ie, P). The regression method includes but is not limited to linear regression.

[0040] Combine the following Figure 3 The example of step 202, i.e. the initialization processing step, is further described.

[0041] In some embodiments, in step 301, an initial P matrix P can be generated based on the pan-genomic data of the target bacterial species. ini , and based on the D matrix and the initial P matrix P ini , the initial S matrix S is obtained by regression ini, for example, according to formula (1) to optimize the regression solution. In some embodiments, the initial P matrix P can be generated based on the gene-related information of multiple strains of the target bacteria (which can constitute at least a part of the pan-genome data). ini In some embodiments, the gene-related information of multiple strains of the target bacterial species can be clustered (for example, but not limited to hierarchical clustering), a representative strain can be selected from each cluster (for example, the strain in the middle can be selected as the representative strain), and the gene-related information of each representative strain can be converted into a list of gene family existence information, thereby obtaining an initial P matrix P ini .

[0042] In some embodiments, based on the D matrix and the initial P matrix P ini The S matrix obtained by regression solution (also called the fifth matrix for convenience of description to distinguish it from the S matrix in the iterative process and the initial S matrix itself) can be directly used as S ini , each element of the fifth matrix represents information related to the relative abundance of each strain in each sample (see step 302).

[0043] In other embodiments, the fifth matrix obtained in step 302 may be further processed. For example, the representative value of each element of the fifth matrix may be compared with the second threshold value in step 303 to select the strain class whose representative value of the corresponding element exceeds the second threshold value. Note that in the present disclosure, "the representative value of each element" may represent each element itself, or may represent the value obtained by representative processing of each element, such as the average value, median value, minimum value, etc. The representative value of the corresponding element means the representative relative abundance of the strain of the corresponding class in all samples. By selecting the strain class whose former exceeds the second threshold value, the strain class whose representative relative abundance exceeds a certain threshold value in all samples may be screened out. In this way, not only the interference information (also referred to as "false positive information") of the strain class that does not really exist due to interference or noise can be filtered out, but also the uncertain information of the strain class that exists but has low abundance and little biological significance can be filtered out. The inventors creatively discovered that by filtering out both interference information and uncertain information, retaining the columns corresponding to the selected strain classes in the initial P matrix to obtain the processed initial P matrix (step 304), based on the D matrix and the processed initial P matrix, S is obtained by regression. ini (Step 305), which can significantly improve the regression solution S ini stability and success rate.

[0044] In various embodiments of the present disclosure, for solving the initial S matrix S ini Various methods can be used to iteratively solve the S matrix, which will not be described here.

[0045] The D matrix fed to the initialization step 202 may be pre-standardized (see step 201). Specifically, the sequencing data of multiple samples may be compared with the pan-genome data of the target bacterial species to determine the comparison count of each sample relative to each gene in the pan-genome data of the target bacterial species to obtain a sixth matrix. Each element of the sixth matrix may represent the comparison count of each gene in each sample.

[0046] Based on the gene length and the sample sequencing depth, each element of the sixth matrix is ​​standardized. As an example but not a limitation, each element of the sixth matrix can be standardized according to formula (3).

[0047]

[0048] Wherein, read count represents each element of the sixth matrix, i.e., the comparison count of each sample relative to each reference gene in the pan-genome of the target bacterial species, gene length represents gene length, library size represents sample sequencing depth, and RPKM represents each element after standardization. Through such standardization, the influence of gene length and sample sequencing depth on the comparison count can be eliminated. Note that formula (3) is only used as an example, and other methods can also be used. For example, other expressions (such as but not limited to polynomials, logarithmic processing, etc.) using gene length and sample sequencing depth as parameters are used in the denominator of formula (3) to perform standardization so as to eliminate the influence of gene length and sample sequencing depth.

[0049] According to formula (4), the D matrix can be obtained by summing the elements of each gene belonging to the same gene family in the sixth matrix after the standardization process.

[0050]

[0051] Among them, GeneFamily i represents the gene family corresponding to the i-th row of the D matrix, and g represents each gene belonging to the gene family.

[0052] Figure 4 FIG. 4 is a diagram showing a configuration of a bacterial strain analysis device 400 according to an embodiment of the present disclosure. Figure 4As shown, the analysis device 400 may include a first acquisition unit 401, a second acquisition unit 402, a comparison unit 403, a merging unit 404, and a decomposition unit 405. The first acquisition unit 401 may be configured to acquire pan-genome data of a target bacterial species, for example, the pan-genome data of a local target bacterial species may be acquired, or the pan-genome data of a target bacterial species may be downloaded and acquired via a network (for example, from a pan-genome database of the bacterial species), wherein the pan-genome data includes reference gene sequences of multiple strains of the target bacterial species.

[0053] The second acquisition unit 402 can be configured to acquire sequencing data of multiple samples, each sample including multiple bacterial species, wherein each bacterial species can have several types of strains. The comparison unit 403 can be configured to compare the sequencing data of multiple samples with the pan-genome data of the target bacterial species to obtain the comparison counts of each sample relative to each reference gene in the pan-genome. The merging unit 404 can be configured to merge the comparison counts of similar genes into the comparison counts of gene families to determine a D matrix, wherein each element of the D matrix represents the information related to the comparison counts of each gene family in each sample. The decomposition unit 405 can be configured to: decompose the D matrix into a P matrix and an S matrix, wherein each element of the P matrix can represent the information related to the presence of each gene family in each type of strain, and each element of the S matrix can represent the information related to the relative abundance of each type of strain in each sample.

[0054] In some embodiments, the decomposition unit 404 may further include an initialization unit 406 and an iterative solution unit 407 to respectively perform initialization processing and iterative solution processing according to various embodiments of the present disclosure. Variations of the initialization processing and iterative solution processing according to various embodiments of the present disclosure may be combined here and will not be elaborated herein.

[0055] Specifically, the initialization unit 406 can be configured to: generate an initial P matrix based on the pan-genome data of the target bacterial species; and obtain an initial S matrix through regression (including but not limited to one or more regressions) based on the information of the D matrix and the initial P matrix. The iterative solution unit 407 can be configured to iteratively solve the P matrix and the S matrix. Specifically, the iterative solution unit 407 can be further configured to: determine the fourth matrix by multiplying the D matrix with the inverse matrix of the currently obtained S matrix in each iteration step; determine the probability density function associated with each gene family in each column of the fourth matrix; remove the elements corresponding to the low-order distribution peaks of the corresponding probability density function from each column of the fourth matrix to filter out noise; compare each element in the fourth matrix after filtering out noise with the first threshold to obtain the current P matrix; solve the S matrix based on the D matrix and the currently obtained P matrix through a regression method so as to be used to determine the fourth matrix in the next iteration step.

[0056] In some embodiments, Figure 4 Each unit in the software module can be implemented as a software module, and the computer executable instructions of the software module can be executed by the processor to implement the corresponding computing process. As an example, the software module can be written and implemented in the framework of nextflow using R language, but this is only an example, and can also be written in languages ​​such as C++, python, etc., and can also be implemented using other frameworks.

[0057] Figure 5 FIG. 5 is a block diagram of a bacterial strain analysis device 500 according to an embodiment of the present disclosure. Figure 5 As shown, the analysis device 500 may include a communication interface 503 and at least one processor 504. The communication interface 503 may be configured to receive sequencing data of multiple samples, each sample including multiple strains. In some embodiments, the communication interface 503 may also be configured to download pan-genome data of the target strain via a network for local use. The pan-genome data may include reference gene sequences of multiple strains of the target strain. Alternatively, in some embodiments, the communication interface 503 may also be configured to receive a strain genome sequence of the target strain, so as to generate pan-genome data of the target strain based on the received strain genome sequence of the target strain using the at least one processor 504. The at least one processor 504 may be further configured to perform the analysis method of the bacterial strain according to various embodiments of the present disclosure.

[0058] The analysis device 500 may be a special purpose computer or a general purpose computer. For example, the analysis device 500 may be a computer customized to perform the processing task. Figure 5 As shown in , the analysis device 500 may include a communication interface 503, at least one processor 504, a memory 505, a storage 506 and a display 507. In some embodiments, the analysis device 500 may be implemented as a user terminal or a server (for example but not limited to a cloud server, etc.).

[0059] In some embodiments, the communication interface 503 may include a network adapter, a cable connector, a serial connector, a USB connector, a parallel connector, a high-speed data transmission adapter (such as optical fiber, USB3.0, Thunderbolt interface, etc.), a wireless network adapter (such as a WiFi adapter), a telecommunication (3G, 4G / LTE, 5G, etc.) adapter, etc. The analysis device 500 may be connected to other components through the communication interface 503, such as but not limited to a gene sequencing device, a pan-genome database, etc. Accordingly, the communication interface 503 may be configured to obtain sequencing data from a gene sequencing device, or to obtain pan-genome data of a target bacterial species from a pan-genome database, etc.

[0060] The processor 504 may be a processing device including one or more general-purpose processing devices, such as a microprocessor, a central processing unit (CPU), a graphics processing unit (GPU), etc. More specifically, the processor may be a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, a processor running other instruction sets, or a processor running a combination of instruction sets. The processor may also be one or more special-purpose processing devices, such as an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), a system on a chip (SoC), etc. The processor 504 may be communicatively coupled to the memory 505 and configured to execute computer executable instructions stored thereon to perform processing of the analysis method of the bacterial strains such as described in the various embodiments of the present disclosure.

[0061] Memory 505 / storage 506 may be a non-transitory computer-readable medium, such as read-only memory (ROM), random access memory (RAM), phase-change random access memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), electrically erasable programmable read-only memory (EEPROM), other types of random access memory (RAM), flash disks or other forms of flash memory, cache, registers, static memory, compact disk read-only memory (CD-ROM), digital versatile disks (DVD) or other optical storage, cassettes or other magnetic storage devices, or any other possible non-transitory medium used to store information or instructions that can be accessed by a computer device.

[0062] In some embodiments, the computer executable instructions may be stored in the storage 506 so as to be loaded into the memory 505 and executed by the processor 504 , thereby implementing the processing of the bacterial strain analysis method according to various embodiments of the present disclosure.

[0063] In some embodiments, the display 507 may include a liquid crystal display (LCD), a light emitting diode display (LED), a plasma display, or any other type of display, and provide a graphical user interface (GUI) presented on the display for user input and image / data / chart display. The display may include many different types of materials (such as plastic or glass) and may be touch-sensitive to receive commands from the user. For example, the display may include a substantially rigid touch-sensitive material (such as Gorilla Glass™) or a substantially flexible (such as Willow Glass™) touch-sensitive material.

[0064] Ten samples were preset through simulation generation. Specifically, five types of sequenced Faecalibacterium prausnitzii strains were selected, whose IDs in the NCBI database were GCF_000162015, GCF_002549755, GCF_002549975, GCF_003287415, and GCF_003433905 (respectively referred to as strain 1, strain 2, strain 3, strain 4, and strain 5), and data were randomly extracted from the whole genome sequencing data of each type of strain and mixed into 10 samples according to a preset ratio, namely sample 1-sample 10.

[0065] The reference strain composition of each preset sample is known and can be used as the ground truth (shown as "reference strain composition" in FIG. 6(a)) to compare with the strain composition obtained by analyzing the preset 10 samples using the analysis method of the bacterial strain according to the embodiment of the present disclosure (i.e., the S matrix, shown as "predicted strain composition" in FIG. 6(a)). As shown in FIG. 6(a), the strain composition obtained by analyzing the preset 10 samples using the analysis method of the bacterial strain according to the embodiment of the present disclosure is highly consistent with the ground truth.

[0066] Figure 6(b) shows a graphical representation of the precision-recall curve of the inclusion of five types of strains (for example, but not limited to the scoring order of the P matrix) obtained by analyzing 10 preset samples using the bacterial strain analysis method according to an embodiment of the present disclosure. It can be seen that the area under the precision-recall curve exceeds 0.99, confirming that it is highly consistent with the inclusion of the five types of strains.

[0067] The inventors applied the analysis method of bacterial strains according to the embodiments of the present disclosure to the fecal microbiota transplantation (FMT) dataset to evaluate the changes in the composition of Faecalibacterium prausnitzii strains at various time points before and after FMT in healthy donors and recipients. Specifically, the FMT dataset includes samples from multiple time points (e.g., identified by FMT5, FMT6, FMT39, FMT41, etc. on the horizontal axis) from 4 healthy donors and 19 recipients. Figure 7(a) shows the strain composition of the first donor (shown as "donor MGH01D" in the figure) and its recipient, Figure 7(b) shows the strain composition of the second donor (shown as "donor MGH06D" in the figure) and its recipient, and Figure 7(c) shows the strain composition of the third donor (shown as "donor MGH07D" in the figure) and its recipient. Figure 7(d)-Figure 7(f) The strain composition of the fourth donor (shown as "donor MGH03D" in the figure) and its recipients is shown.

[0068] from Figure 7(a)-Figure 7(f)It can be seen that the samples of the four donors mainly contain three types of strains (i.e., strain 1, strain 2, and strain 3, represented by black, dark gray, and light gray, respectively); some recipients inherited part or all of the strains from the donors, but the composition ratio of the strains of the recipients is different from that of the donors. The bacterial strain analysis method according to the embodiment of the present disclosure can conveniently and accurately track the composition changes of the strains of Faecalibacterium prausnitzii at various time points before and after FMT in healthy donors and recipients.

[0069] Figure 8 The pathway functions corresponding to the genes of these three strains are shown. Dark colors represent strains with this pathway function, and light colors represent the lack of this pathway function. It can be seen that these three strains have differences in material metabolism, which may also affect their adaptability and survival in the host intestinal environment. By conveniently and accurately tracking the composition changes of Faecalibacterium prausnitzii strains at various time points before and after FMT in healthy donors and recipients, combined with the pathway functions corresponding to the genes of each strain, it is conducive to the comprehensive analysis of the inheritance effect of the recipient on the strain composition ratio of the donor at the metabolic and pathway function levels.

[0070] In addition, although exemplary embodiments have been described herein, the scope includes any and all embodiments based on the present disclosure with equivalent elements, modifications, omissions, combinations (e.g., various embodiments intersecting schemes), adaptations or changes. The elements in the claims will be interpreted broadly based on the language adopted in the claims, and are not limited to the examples described in this specification or during the implementation of this application, and the examples will be interpreted as non-exclusive. Therefore, this specification and examples are intended to be considered as examples only, and the true scope and spirit are indicated by the following claims and the full scope of their equivalents.

[0071] The order of the various steps in the present disclosure is merely exemplary and not restrictive. The execution order of the steps may be adjusted without affecting the implementation of the present disclosure (without destroying the required logical relationship between the steps), and the various embodiments obtained after the adjustment still fall within the scope of the present disclosure.

[0072] The above description is intended to be illustrative rather than restrictive. For example, the above examples (or one or more of them) can be used in combination with each other. For example, a person of ordinary skill in the art can use other embodiments when reading the above description. In addition, in the above-mentioned specific embodiments, various features can be grouped together to simplify the present disclosure. This should not be interpreted as an intention that a disclosed feature that is not required to be protected is necessary for any claim. On the contrary, the subject matter of the present invention may be less than all the features of a specific disclosed embodiment. Thus, the following claims are incorporated into the specific embodiments as examples or embodiments, wherein each claim is independently used as a separate embodiment, and it is considered that these embodiments can be combined with each other in various combinations or arrangements. The scope of the present invention should be determined with reference to the attached claims and the full scope of equivalent forms granted by these claims.

Claims

1. A method for analyzing bacterial strains, characterized in that: The analysis method comprises: Obtaining pan-genome data of the target bacterial species, wherein the pan-genome data includes reference gene sequences of multiple strains of the target bacterial species; Obtain sequencing data of multiple samples, each sample including multiple bacterial species; Comparing the sequencing data of each sample with the pan-genome data of the target bacterial species via at least one processor to obtain alignment counts relative to each reference gene in the pan-genome, and merging alignment counts of similar genes into alignment counts of gene families to determine a first matrix, wherein each element of the first matrix represents alignment count-related information of each gene family in each sample; Decomposing the first matrix into a second matrix and a third matrix by the at least one processor, wherein each element of the second matrix represents information related to the presence of each gene family in each strain, and each element of the third matrix represents information related to the relative abundance of each strain in each sample; Decomposing the first matrix into a second matrix and a third matrix further comprises: Performing initialization processing to generate an initial second matrix; Generate an initial third matrix according to information of the first matrix and the initial second matrix; Based on the initial third matrix, the second matrix and the third matrix are iteratively solved.

2. The analysis method according to claim 1, characterized in that The rows of the first matrix correspond to each gene family and the columns correspond to each sample, the rows of the second matrix correspond to each gene family and the columns correspond to each strain, the rows of the third matrix correspond to each strain and the columns correspond to each sample, and the existence-related information is binary information indicating whether or not; Wherein, solving the third matrix includes solving the third matrix by a regression method based on the first matrix and the currently obtained second matrix, Solving the second matrix further includes: Determine a fourth matrix by multiplying the first matrix by the inverse matrix of the third matrix currently obtained; Determine a probability density function of each column of the fourth matrix associated with each gene family; Removing elements corresponding to low-order distribution peaks of the corresponding probability density function from each column of the fourth matrix to filter out noise; Each element in the fourth matrix after the noise is filtered out is compared with the first threshold to obtain the second matrix.

3. The analysis method according to claim 2, characterized in that The initialization process further includes: Generate an initial second matrix based on gene-related information of representative strains of the target bacterial species; The initial third matrix is ​​obtained by solving the first matrix and the initial second matrix through regression.

4. The analysis method according to claim 3, characterized in that Based on the first matrix and the initial second matrix, solving the initial third matrix through regression further includes: Based on the first matrix and the initial second matrix, a fifth matrix is ​​obtained by regression, wherein each element of the fifth matrix represents information related to the relative abundance of each strain in each sample; By comparing the representative value of each element of the fifth matrix with the second threshold, the strain class whose representative value of the corresponding element exceeds the second threshold is selected; Obtaining a processed initial second matrix by retaining the columns corresponding to the selected strain classes in the initial second matrix; and Based on the first matrix and the processed initial second matrix, the initial third matrix is ​​obtained by regression.

5. The analysis method according to any one of claims 2 to 4, characterized in that: The regression comprises a linear regression towards non-negative values.

6. The analysis method according to any one of claims 1 to 4, characterized in that Comparing the sequencing data of the plurality of samples with the pan-genome data of the target bacterial species to determine that the first matrix further comprises: By the comparison, the comparison count of each sample relative to each gene in the pan-genome data of the target bacterial species is determined to obtain a sixth matrix, wherein each element of the sixth matrix represents the comparison count of each gene in each sample; Based on gene length and sample sequencing depth, each element of the sixth matrix is ​​standardized; The first matrix is ​​obtained by summing the elements of each gene belonging to the same gene family in the sixth matrix after the standardization process.

7. A bacterial strain analysis device, characterized in that: The analysis device comprises: An interface configured to: receive sequencing data of a plurality of samples, each sample including a plurality of bacterial species; and At least one processor configured to execute the bacterial strain analysis method according to any one of claims 1-6.

8. The analysis device according to claim 7, characterized in that The interface is further configured to download pan-genome data of the target bacterial species, wherein the pan-genome data includes reference gene sequences of multiple strains of the target bacterial species; or The interface is also configured to receive a genome sequence of a strain of the target bacterial species, and the at least one processor is further configured to generate pan-genome data of the target bacterial species based on the received genome sequence of the strain of the target bacterial species.

9. A bacterial strain analysis device, characterized in that: The analysis device comprises: A first acquisition unit is configured to acquire pan-genome data of a target bacterial species, wherein the pan-genome data includes reference gene sequences of multiple strains of the target bacterial species; A second acquisition unit is configured to acquire sequencing data of a plurality of samples, each sample including a plurality of bacterial species; A comparison unit configured to compare the sequencing data of each sample with the pan-genome data of the target bacterial species to obtain a comparison count of each sample relative to each reference gene in the pan-genome; a merging unit configured to merge the alignment counts of similar genes into the alignment counts of a gene family to determine a first matrix, wherein each element of the first matrix represents information related to the alignment counts of each gene family in each sample; and A decomposition unit, configured to: decompose the first matrix into a second matrix and a third matrix, wherein each element of the second matrix represents information related to the presence of each gene family in each strain, and each element of the third matrix represents information related to the relative abundance of each strain in each sample; Decomposing the first matrix into a second matrix and a third matrix further comprises: Performing initialization processing to generate an initial second matrix; Generate an initial third matrix according to information of the first matrix and the initial second matrix; Based on the initial third matrix, the second matrix and the third matrix are iteratively solved.

10. The analysis device according to claim 9, characterized in that The decomposition unit further comprises: Initialize the unit, its configuration is: Generate an initial second matrix based on the pan-genomic data of the target bacterial species; and According to the information of the first matrix and the initial second matrix, an initial third matrix is ​​obtained through regression; The iterative solution unit is configured to iteratively solve the second matrix and the third matrix, specifically comprising, in each iteration step: Determine a fourth matrix by multiplying the first matrix by the inverse matrix of the third matrix currently obtained; Determine a probability density function of each column of the fourth matrix associated with each gene family; Removing elements corresponding to low-order distribution peaks of the corresponding probability density function from each column of the fourth matrix to filter out noise; Compare each element in the fourth matrix after the noise is filtered out with the first threshold value to obtain the current second matrix; A third matrix is ​​solved by a regression method based on the first matrix and the currently obtained second matrix, so as to be used for determining the fourth matrix in the next iteration step.

11. A non-volatile computer-readable storage medium having executable instructions stored thereon, wherein when the executable instructions are executed by at least one processor, the bacterial strain analysis method according to any one of claims 1 to 6 is implemented.

Images