Methods, devices, and models for obtaining spatial distributions of biomolecules in a target omics

Abstract

The application discloses a method, device and model for obtaining spatial distribution of biomolecules in target omics. The method of the application is based on experimental methods of microfluidic chips and computational methods of transfer learning, effectively reduces the number of samples to be detected, improves experimental efficiency, and can obtain spatial distribution and expression of target molecules on the whole detection tissue with high resolution. For example, in the example technical solution, one-dimensional target omics information at two angles can be used to predict two-dimensional distribution information of target omics molecules on the whole detection tissue through the transfer learning model, which significantly reduces the number of samples to be detected, and obtains spatial information of omics molecules in a high-throughput and high-spatial-resolution manner.

Description

TECHNICAL FIELD

[0001] The present application relates to omics methods, in particular to methods, devices and models for obtaining spatial distribution of biomolecules in target omics. BACKGROUND

[0002] Single-cell sequencing technology revolutionarily explains the heterogeneity among cells in tissues by detecting biological macromolecules in single cells, greatly promoting our understanding of biological processes and diseases. However, this technology first requires the dissociation of tissues to prepare single-cell suspensions, thereby losing the in situ information of cells and failing to give the spatial arrangement information of cells in solid heterogeneous tissues. The spatial transcriptomics technology developed in recent years solves this problem by detecting the in situ transcriptome of tissues while preserving the morphological structure of tissues.

[0003] However, the spatial transcriptomics technology is mainly limited to measuring the transcriptome of cells, and the transcriptome is only an indirect reflection of the state of cells, while most biological processes are controlled by proteins. In addition, the abundance correlation between proteins and transcripts is low, which leads to the inability to rely solely on transcriptomics for biological or medical research. In addition, the number and types of post-translational modifications (PTM) of proteins involved in a large number of biological processes far exceed the range of genomics measurements. Therefore, from the perspective of basic cell biology or clinically, spatial proteomics and downstream spatial metabolomics analysis are expected to completely change our understanding of biological processes.

[0004] Spatial proteomics is a science for obtaining spatial abundance information of different proteins in two-dimensional space of biological tissue samples. The currently reported spatial proteomics can be mainly divided into two categories: the first category is targeted antibody-based spatial proteomics, and the second category is non-targeted spatial proteomics based on laser microdissection technology and mass spectrometry detection technology.

[0005] The first category of targeted spatial proteomics technology is highly dependent on the types, prices, and stability of antibodies. Because the antibody preparation process is complex, the first problem of this technology is that it is very expensive. Secondly, the number of available antibodies is very small, usually in tens of antibodies, so most antibody-based spatial protein detection technologies can only detect tens of proteins, and cannot be applied to large-scale spatial distribution detection of proteins.

[0006] The second type is a non-targeted spatial proteomics detection technology based on laser microdissection. A laser microdissection instrument is used to cut the tissue into small tissue blocks with a specified resolution, and then a mass spectrometer is used to perform unbiased proteome detection on the small tissue blocks obtained by cutting. The main problems of this method are: 1) It is difficult to study large-area tissues. For example, a 5mmx 5mm tissue section needs to be cut 2500 times to obtain 0.1mm x 0.1mm small tissue blocks, and then 2500 mass spectrometry detections are performed. However, due to the limitations of mass spectrometry detection throughput and instrument stability, it is difficult to ensure the accuracy and effectiveness of quantitative detection on the above 2500 samples; 2) Low resolution. Due to the limitations of mass spectrometry detection sensitivity, it is difficult to effectively detect proteins in too small tissue blocks. The smallest tissue block that can be detected reported so far is 100μm, and for higher resolution such as 25μm tissue blocks, this scheme is difficult to implement. Although the recently developed "Deep Visual Proteomics(DVP)" technology classifies cell types in stained tissues using image recognition technology, and then uses laser microdissection technology to cut cells of the same type into a container for proteome analysis. However, this technology has the following problems: 1) The cell types that can be classified by images are limited; 2) Cannot know the specific spatial location information of the cells; 3) Cannot obtain spatial protein information at a large field of view (such as tissue level).

[0007] Currently, spatial metabolomics information is mainly obtained by mass spectrometry imaging (MSI). The mainstream mass spectrometry imaging techniques include: (1) Matrix-Assisted Laser Desorption Ionization-Mass Spectrometry Imaging (MALDI-MSI), (2) Desorption Electrospray Ionization-Mass Spectrometry Imaging (DESI-MSI), Secondary Ion Mass Spectrometer-Mass Spectrometry Imaging (SIMS-MSI), etc. The problems of the above methods include: 1) Cannot obtain accurate metabolite information. Because the abundance of metabolites on the tissue is very low, and the compounds are not separated by chromatography after ionization and directly sent into the mass spectrometer detector, it is difficult to distinguish compounds with the same molecular weight but different structures, and it is difficult to further detect the compounds after primary mass spectrometry detection; 2) Additional requirements for mass spectrometers, ordinary mass spectrometers cannot meet.

[0008] Currently, the spatial multi-omics detection technology only has the DBiT-seq technology. The DBiT-seq technology realizes the simultaneous detection of the in-situ transcriptome and a small number of target proteins (25) in the tissue by designing orthogonal chips for spatial combined labeling and antibody delivery labeled with nucleotides. The problems of this method are: 1) unable to detect proteins in a non-targeted manner; 2) unable to detect other omics data, such as metabolome, protein modification group, and transcription modification.

[0009] The information in the background art is merely intended to explain the general background of the application and should not be considered as admitting or in any form implying that these information constitutes prior art known to those skilled in the art. SUMMARY

[0010] To solve at least part of the technical problems in the prior art, the present application provides a method, device and model for obtaining the spatial distribution of biomolecules in target omics based on reference omics data. Specifically, the present application includes the following contents.

[0011] In a first aspect of the present application, a method for obtaining the spatial distribution of biomolecules in target omics is provided, which comprises the following steps:

[0012] (1) obtaining continuous sections of a target sample, and further dividing part of the sections into strip-shaped sub-tissues at different angles;

[0013] (2) simultaneously or separately obtaining the information of biomolecules in the target omics in each strip-shaped sub-tissue;

[0014] (3) obtaining the two-dimensional spatial distribution information of biomolecules in the reference omics in the sections; and

[0015] (4) training a learning model based on the two-dimensional spatial distribution information of biomolecules in the reference omics, and then using the trained learning model and the one-dimensional information of biomolecules in the target omics to predict the two-dimensional spatial information of biomolecules in the target omics.

[0016] In some embodiments, the method for obtaining the spatial distribution of biomolecules in target omics according to the present application, wherein the step (1) comprises:

[0017] obtaining x continuous sections of a target region of a target sample, respectively denoted as T1, T2, T3,... Tx, and placing the tissue sections in the regions of interest of the substrates, respectively, wherein x is a natural number greater than or equal to 3, the T1 section is placed along a first direction, the T3 section is placed along a first direction, and the first direction and the second direction are different, i.e. the first direction and the second direction can form an included angle;

[0018] imaging the continuous sections T1, T2, T3, Tx to generate sample images;

[0019] The sections T1 and T3 are further divided into strip sub-tissues, respectively, wherein the first and second strip sub-tissues of T1 are denoted as T1-1, T1-2 to the nth strip T1-n, and the first and second strip sub-tissues of T3 are denoted as T3-1, T3-2 to the nth strip T3-n, wherein n is a natural number from 2 to 500.

[0020] In some embodiments, the method for obtaining the spatial distribution of biomolecules in the target omics according to the present application, wherein the part of the sections in step (1) is further divided into strip sub-tissues by laser microdissection or microfluidic device.

[0021] In some embodiments, the method for obtaining the spatial distribution of biomolecules in the target omics according to the present application, wherein the microfluidic device comprises a section division area comprising 10-500 variable width microchannels, and there is a channel wall between each microchannel, and the width of the channel wall is between 500 nm-200 μm.

[0022] In some embodiments, the method for obtaining the spatial distribution of biomolecules in the target omics according to the present application, wherein the width of the strip sub-tissue obtained by laser microdissection is 5-200 μm, and the spacing between adjacent continuous strip sub-tissues is between 0-200 μm.

[0023] In some embodiments, the method for obtaining the spatial distribution of biomolecules in the target omics according to the present application, wherein in step (4), the target omics is proteomics, and the acquisition of one-dimensional information of biomolecules includes:

[0024] Deliver the tissue lysis reagent containing reference peptides to the regions of the strip sub-tissues of at least sections T1 and T3, respectively, so that the tissue lysis reagent lyses the strip sub-tissues T1-1…T1-n and T3-1…T3-n in the region of interest of the substrate;

[0025] Deliver the lysis solution after lysis of the strip sub-tissues T1-1…T1-n and T3-1…T3n from the region of interest of the substrate to the outlet of the microfluidic device, and transfer each strip sub-tissue lysate solution to an independent container, respectively;

[0026] Detect the biomolecule information in the lysates of T1-1…T1-n and T3-1…T3-n by liquid chromatography-mass spectrometry, respectively, to obtain the one-dimensional physicochemical information of the proteome of section T1 at angle α and section T3 at angle β.

[0027] In certain embodiments, according to the method for obtaining spatial distribution of biomolecules in a target omics of the present application, in step (4), the obtaining of one-dimensional information of biomolecules of the target omics comprises:

[0028] The solvent of the target compound is delivered to the regions of the strip sub-tissues T1-1...T1-n and T3-1...T3n, respectively, so that the solvent extracts the compounds in the strip sub-tissues T1-1...T1-n and T3-1...T3n in the region of interest of the substrate;

[0029] The solvent extracting the compounds of the strip sub-tissues T1-1...T1-n and T3-1...T3-n is delivered from the region of interest of the substrate to the outlet of the microfluidic device, and each strip sub-tissue solution is transferred to an independent container, respectively;

[0030] The compound information in the T1-1...T1-n and T3-1...T3-n solutions is detected by liquid chromatography-mass spectrometry, respectively, to obtain the one-dimensional physicochemical information of the metabolome of the slice T1 at the angle a and the slice T3 at the angle b.

[0031] In certain embodiments, according to the method for obtaining spatial distribution of biomolecules in a target omics of the present application, in step (4), the target omics is metabolomics, and the obtaining of one-dimensional information of biomolecules comprises:

[0032] The reverse transcription reagent and the barcoded polynucleotide are delivered to the regions of the strip sub-tissues T1-1...T1-n and T3-1...T3n to generate cDNA in the strip sub-tissues T1-1...T1-n and T3-1...T3-n in the region of interest of the substrate;

[0033] The lysis buffer or denaturation reagent is delivered to the strip sub-tissues to generate a lysed or denatured tissue sample;

[0034] The cDNA is extracted from the lysed or denatured tissue sample; and

[0035] The cDNA is sequenced to generate cDNA reads.

[0036] In certain embodiments, according to the method for obtaining spatial distribution of biomolecules in a target omics of the present application, wherein the barcoded polynucleotide comprises a PCR handle end sequence, a strip sub-tissue barcode sequence, a unique molecular identifier sequence, and a polyT sequence, optionally, wherein the PCR handle end sequence is end-functionalized with biotin.

[0037] In certain embodiments, according to the method for obtaining spatial distribution of biomolecules in a target omics of the present application, in step (4), the target omics is transcriptomics, and the obtaining of one-dimensional information of biomolecules comprises:

[0038] delivering histone removal reagent to the regions of the long strip sub-tissues T1-1…T1-n and T3-1…T3n, removing histones in the genomes in the strip sub-tissues T1-1…T1-n and T3-1…T3-n in the substrate’s region of interest;

[0039] delivering fragmentation reagent to the strip sub-tissues to generate fragmented genomes;

[0040] lysis of the strip sub-tissues T1-1…T1-n and T3-1…T3-n and delivering the lysed liquid from the substrate’s region of interest to the microfluidic device outlet and transferring each strip sub-tissue solution to an independent container, respectively;

[0041] amplifying the genomic fragments in the containers; and

[0042] sequencing the genomic amplification products to obtain sequence information.

[0043] In some embodiments, according to the method for obtaining the spatial distribution of biomolecules in target omics of the present application, in step (4), the target omics is genomics, and the acquisition of one-dimensional information of biomolecules includes:

[0044] delivering a solvent of a target compound to the regions of the long strip sub-tissues T1-1…T1-n and T3-1…T3n, respectively, to extract the compound in the strip sub-tissues T1-1…T1-n and T3-1…T3-n in the substrate’s region of interest with the compound solvent;

[0045] delivering the compound-extracted strip sub-tissue T1-1…T1-n and T3-1…T3-n solvent from the substrate’s region of interest to the microfluidic device outlet and transferring each strip sub-tissue solution to an independent container, respectively;

[0046] detecting the compound information in the T1-1…T1-n and T3-1…T3-n solutions with liquid chromatography-mass spectrometry, respectively, to obtain the one-dimensional physicochemical information of the metabolome of the tissue slice T1 at the angle a and the tissue slice T3 at the angle b;

[0047] delivering rehydration reagent and cleaning reagent to the separated long strip sub-tissue regions of the tissue slices T1 and T3, respectively, and discarding the outflow solutions of each strip sub-tissue;

[0048] delivering reverse transcription reagent and barcoded polynucleotides to the regions of the long strip sub-tissues T1-1…T1-n and T3-1…T3n, to generate cDNA in the strip sub-tissues T1-1…T1-n and T3-1…T3-n in the substrate’s region of interest;

[0049] delivering a lysis buffer or denaturing reagent to the strip sub-tissue to generate a lysed or denatured tissue sample;

[0050] delivering the lysate solution of the lysed strip sub-tissue from the region of interest of the substrate to the microfluidic device outlet, and separately transferring each strip sub-tissue lysate solution to an independent container, and isolating cDNA and protein lysate from the solution using magnetic beads;

[0051] detecting biological information including sequence information, protein modification information and abundance information of proteins in the lysates of T1-1…T1-n and T3-1…T3-n respectively using liquid chromatography-mass spectrometry, and further obtaining proteomic one-dimensional physicochemical information of slice T1 at angle a and slice T3 at angle b; and

[0052] performing library construction and sequencing on the cDNA to generate cDNA reads.

[0053] In some embodiments, the method for obtaining spatial distribution of biological molecules in a target omics according to the present application, wherein the step (3) obtaining two-dimensional spatial distribution information of biological molecules of the reference omics in the slice comprises at least one selected from the group consisting of metabolomics, transcriptomics, proteomics and histological information.

[0054] In some embodiments, the method for obtaining spatial distribution of biological molecules in a target omics according to the present application, wherein the two-dimensional information of the metabolomics is obtained by mass spectrometry imaging technology, which comprises at least one selected from the group consisting of matrix-assisted laser desorption mass spectrometry imaging technology, electrospray desorption ionization imaging technology and secondary ion mass spectrometry imaging technology.

[0055] In some embodiments, the method for obtaining spatial distribution of biological molecules in a target omics according to the present application, wherein the two-dimensional information of the transcriptomics is obtained by spatial transcriptomics technology, which comprises microimaging-based spatial transcriptomics technology and / or spatial nucleic acid tag array-based technology.

[0056] In some embodiments, the method for obtaining spatial distribution of biological molecules in a target omics according to the present application, wherein the obtaining of the two-dimensional information of the proteomics comprises at least one selected from the group consisting of antibody tag-based proteomics technology, nucleic acid sequence-labeled antibody tag and / or metal tag-labeled antibody.

[0057] In some embodiments, the method for obtaining spatial distribution of biological molecules in a target omics according to the present application, wherein the obtaining of the two-dimensional information of the histology comprises at least one selected from the group consisting of hematoxylin-eosin staining, acetylcholinesterase staining, Nissl staining and Masson staining.

[0058] In some embodiments, the method for obtaining spatial distribution of biomolecules in target omics according to the present application, wherein step (4) comprises: taking the reference omics as a training set, the sampling process simulation as an encoder, and the sampling value reconstruction as a decoder, wherein the trained decoder is used as a reconstruction model for subsequent use.

[0059] In some embodiments, the method for obtaining spatial distribution of biomolecules in target omics according to the present application, wherein the encoder comprises at least one of parallel slicing simulation, random point sampling simulation, and laser cutting simulation; and the decoder comprises at least one of a machine learning model, a deep learning model, and a probabilistic inference model.

[0060] In some embodiments, the method for obtaining spatial distribution of biomolecules in target omics according to the present application, wherein step (4) comprises: training a learning model with the two-dimensional spatial distribution information of biomolecules in the reference omics, and then using the trained learning model and the one-dimensional information of biomolecules in the target omics to predict the two-dimensional spatial information of biomolecules in the target omics.

[0061] In a second aspect, the present application provides a microfluidic device for obtaining spatial distribution of biomolecules in target omics, comprising a slicing partition region arranged on a substrate, which comprises 10-500 variable-width microchannels, each microchannel having an inlet end and an outlet end, and a channel wall between each microchannel, the width of the channel wall being between 500 nm and 200 μm.

[0062] In some embodiments, the microfluidic device for obtaining spatial distribution of biomolecules in target omics according to the present application, wherein the microchannels are arranged in parallel, and the lengths of the microchannels are equal.

[0063] In some embodiments, the microfluidic device for obtaining spatial distribution of biomolecules in target omics according to the present application, further comprising a cover sheet, which is detachably pressed together with the slicing partition region by a fixing structure.

[0064] In some embodiments, the microfluidic device for obtaining spatial distribution of biomolecules in target omics according to the present application, further comprising a reagent region, wherein the reagents in the reagent region enter each microchannel through the inlet end of each microchannel.

[0065] In some embodiments, the microfluidic device for obtaining spatial distribution of biomolecules in target omics according to the present application, wherein the reagent region is provided with a plurality of reagent receiving cavities, one end of each reagent receiving cavity being in communication with the inlet end of a corresponding microchannel.

[0066] In some embodiments, the microfluidic device for obtaining spatial distribution of biomolecules in target omics according to the present application, wherein the negative pressure zone is further configured to drive the reagents from the reagent zone into the microchannel to reach the cutting tissue zone in the microchannel.

[0067] In some embodiments, the microfluidic device for obtaining spatial distribution of biomolecules in target omics according to the present application, wherein the negative pressure zone is further configured to drive the reagents from the reagent zone into the microchannel to reach the cutting tissue zone in the microchannel.

[0068] In some embodiments, the microfluidic device for obtaining spatial distribution of biomolecules in target omics according to the present application, wherein the negative pressure zone is further configured to drive the reagents from the reagent zone into the microchannel to reach the cutting tissue zone in the microchannel.

[0069] In some embodiments, the microfluidic device for obtaining spatial distribution of biomolecules in target omics according to the present application, wherein the negative pressure zone is further configured to drive the reagents from the reagent zone into the microchannel to reach the cutting tissue zone in the microchannel.

[0070] In some embodiments, the microfluidic device for obtaining spatial distribution of biomolecules in target omics according to the present application, wherein the negative pressure zone is further configured to drive the reagents from the reagent zone into the microchannel to reach the cutting tissue zone in the microchannel.

[0071] a. obtaining information of biomolecules in target omics in a plurality of strip-like sub-tissues obtained from a tissue section;

[0072] b. obtaining two-dimensional spatial distribution information of biomolecules in reference omics in the tissue section; and

[0073] c. training a learning model based on the two-dimensional spatial distribution information of biomolecules in the reference omics, and then using the trained learning model and one-dimensional information of biomolecules in the target omics to predict two-dimensional spatial information of biomolecules in the target omics.

[0074] In some embodiments, the microfluidic device for obtaining spatial distribution of biomolecules in target omics according to the present application, wherein the negative pressure zone is further configured to drive the reagents from the reagent zone into the microchannel to reach the cutting tissue zone in the microchannel.

[0075] at least one processor; and

[0076] a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the construction method according to the present application.

[0077] The advantages of the technical solutions of the present application include the following:

[0078] The method of the present application is a method based on transfer learning, which effectively reduces the number of samples to be detected, improves experimental efficiency, and obtains the spatial distribution and expression amount of target molecules on the entire detection tissue with high resolution. For example, in the exemplary technical solution, only one-dimensional target omics information at two angles can be used to predict the two-dimensional distribution information of target omics molecules on the entire detection tissue through the transfer learning model, which significantly reduces the number of samples to be detected, and obtains the spatial information of omics molecules in a high-throughput and high-spatial-resolution manner.

[0079] The method of the present application is widely applicable. For example, the method of the present application can be used for the analysis of omics molecules such as tissue cell proteins, mRNA, metabolites, DNA, chromatin, and carbohydrates, thereby realizing multi-omics detection of spatial proteomics, transcriptomics, metabolomics, and genomics.

[0080] Information that is difficult to obtain by traditional methods can be obtained. For omics such as proteomics and metabolomics, the method of the present application can obtain information about target molecules that is difficult to obtain by traditional methods, including secondary structure of target molecules, differentiation of isomers, and more molecular species, in combination with HPLC chromatographic separation and mass spectrometry secondary detection.

[0081] The sample preparation cost in the method of the present application is low and does not require complex equipment assistance. For example, in the exemplary embodiment, a specific microfluidic device is used as a means to obtain one-dimensional information of target omics molecules, and the overall cost of the chip and device is low and does not require complex external equipment, effectively reducing the cost of the experiment.

[0082] The data generated by the method of the present application has the characteristic of high spatial resolution. For example, the method of the present application uses a microfluidic device to achieve high-resolution one-dimensional information sampling, thereby realizing high-resolution two-dimensional biomolecule analysis, and the highest spatial omics analysis can achieve 10 μm resolution.

[0083] The method of the present application has the following advantages over traditional spatial proteomics: 1) we use a microfluidic device to obtain one-dimensional information of the target omics, and the microfluidic device has the advantages of high spatial resolution, high throughput, low cost, easy preparation and operation, and does not require complex laser microdissection equipment; 2) the present application first proposes a strategy for spatial biomolecular reconstruction using transfer learning algorithm in deep learning. The number of samples to be detected can be significantly reduced. For example, for a 0.1mm x 0.1mm spatial resolution of a 5mm x 5mm tissue block, only 50 protein samples are needed for reconstruction and reconstruction in the present application, while 2500 protein samples are needed for the traditional laser microdissection-based scheme; 3) the present application can achieve higher spatial resolution, and the highest spatial resolution obtained by the present application is 25μm, which is significantly higher than the reported technical solutions.

[0084] The advantages of the method of the present application over traditional metabolomics include: 1) the one-dimensional sample compound concentration of the target omics is high, and after primary mass spectrometry detection, secondary detection of the compound can be performed, compared with mass spectrometry imaging which relies on primary mass spectrometry data and database comparison to obtain molecular structure information, the method of the present application can obtain more accurate compound molecular structure based on secondary mass spectrometry information; 2) one-dimensional samples will be separated by chromatography, and isomers with different polarities can be separated before entering the mass spectrometer, thereby effectively distinguishing isomer molecules; 3) one-dimensional sample compounds can be derivatized according to research needs to obtain target metabolite information.

[0085] The advantage of the method of the present application over traditional spatial multi-omics is that it can simultaneously detect multiple omics, including genome, transcriptome, proteome and metabolome, and the proteome is obtained in a non-targeted manner, with a number much larger than that obtained by targeted antibody capture. The method of the present application can simultaneously detect 20,000 genes, more than 4,000 proteins and 3,500 metabolites, which is not achievable by the currently reported technologies. BRIEF DESCRIPTION OF DRAWINGS

[0086] Figure 1 Schematic diagram of an exemplary microfluidic device.

[0087] Figure 2 Design and optimization of microfluidic chips. Among them, Figure 2 A is a microfluidic chip design drawing with equal channel length designed and used in the present patent. Figure 2 B is a physical diagram of a microfluidic device. Figure 2 C is a common microfluidic chip design drawing with unequal channel length. Figure 2 D is a comparison of the differences in fluid resistance of microchannels with equal length and unequal length.

[0088] Figure 3 Flow chart of the method for obtaining the spatial distribution of biomolecules in the target omics.

[0089] Figure 4 Computer simulation verification of the reliability of Flow2Spatial based on published spatial transcriptome data. Among them, Figure 4 A is the flow chart of Flow2Spatial algorithm. Figure 4 B is the flow chart of Flow2Spatial algorithm verified by slide-seq spatial transcriptome data. Figure 4 C is the comparison of the reconstruction results of Flow2Spatial and the true value.

[0090] Figure 5 For different embedding agent samples, no polymer peak is observed for Cryo-gel, and obvious polymer peak is observed for OCT sample.

[0091] Figure 6 For the tissue embedded without embedding agent, the edge of the tissue is curled during the sectioning of the tissue, which affects the sealing effect of the microfluidic chip.

[0092] Figure 7 For the red fluorescent protein labeled mouse cerebellum tissue section, the picture after in situ digestion in the chip.

[0093] Figure 8 For different lysis solutions, i.e. the number of proteins obtained by different protein sample preparation methods in the chip.

[0094] Figure 9 For different channel pretreatment methods, the effect of reducing non-specific adsorption of sample proteins.

[0095] Figure 10 For the reliability evaluation of mass spectrometric quantitative detection of protein groups in samples. Figure 10 A schematic diagram of sample gradient dilution and corresponding mass spectrometric detection results of protein groups. Figure 10 B Protein abundance heat map and protein distribution density map between samples in gradient dilution samples. Figure 10 C Protein abundance coefficient of variation distribution of intra-group repeats in gradient dilution samples. Figure 10 D Deviation coefficient of actual protein abundance proportion and theoretical proportion in gradient dilution samples. Figure 10 E Randomly select 6 proteins with a deviation coefficient and a coefficient of variation less than 30%. Figure 10 F Correlation between different gradient samples. Figure 10 G Protein sampling and mass spectrometric detection schematic diagram of cerebellum sample. Figure 10 H Coefficient of variation of protein abundance between QC samples. Figure 11I Correlation between the abundance of housekeeping protein Gapdh and the amount of tissue.

[0096] Figure 12 To validate the cross-contamination of protein samples between microchannels using E. coli proteins.

[0097] Figure 12 To validate the reliability of Flow2Spatial method using mouse cerebellum tissue sections. Figure 12 A Schematic diagram of using reconstruction algorithm Flow2Spatial and Tomographer for mouse cerebellum spatial proteome analysis, respectively. Figure 12 B Left: brain region distribution map of mouse cerebellum from Allen Brain database, middle and right: brain region distribution maps reconstructed by Flow2Spatial at different clustering resolution. Figure 12 C Venn diagram of brain region specific expressed genes found by three methods: Slide-seq, Tomographer, Flow2Spatial. Figure 12 D UMAP plot of brain region clustering, asterisks represent marker genes of different brain regions. Figure 13 E Immunofluorescence staining to verify the accuracy of reconstructed results.

[0098] Figure 13 To analyze the spatial proteome of rat colon villi. Among them, Figure 13 A Schematic diagram of the tissue structure of colon villi. Figure 13 B Left: microscopic picture of the chip pressing on the villus tissue, right: the number of proteins detected by each microfluidic channel. Figure 13 C Spatial proteome data identified at different clustering resolution. Figure 14 D Immunofluorescence staining results of colon villi. Figure 15 E Spatial distribution of differential transporters on villi.

[0099] Figure 15 Mass spectrometry peaks of metabolites extracted from microchannels using 80% ethanol as solvent.

[0100] Figure 15 To verify the application of this spatial multi-omics method to spatial transcriptome. Among them, Figure 16 A Flowchart of in situ reverse transcription and cDNA amplification of mRNA in microchannels. Figure 16 B Capillary electrophoresis detection results of cDNA amplification products. Figure 16 C The number of genes detected by each microchannel tissue after sequencing and the correlation between samples between channels.

[0101] Figure 16The results of the spatial multi-omics method applied to spatial multi-omics. Among them, Step (1) A is the result of metabolite detection in the tissue in the microchannel. Step (2) B is the result of the number of proteins detected in different microchannels. Step (3) C is the result of capillary electrophoresis detection of cDNA amplification products. DETAILED DESCRIPTION

[0102] Various exemplary embodiments of the present application will now be described in detail, which should be considered to be illustrative of certain aspects, features and embodiments of the present application, but not a limitation of the present application.

[0103] It should be understood that the terms used in the present application are merely for the purpose of describing particular embodiments and are not intended to limit the present application. In addition, for numerical ranges in the present application, it is understood that the upper limit and the lower limit of the range and every intermediate value between them are specifically disclosed. Each smaller range within the range of any stated value or stated range, as well as any other stated value or intermediate value within the stated range, is also included in the present application. The upper limit and the lower limit of these smaller ranges can be included or excluded independently from the range.

[0104] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. Although methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present application, preferred methods and materials are described. All documents mentioned in this specification are incorporated by reference to disclose and describe the methods and / or materials in connection with which the documents are cited. In case of conflict between the content of the specification and that of any document incorporated herein by reference, the content of the specification prevails. Unless otherwise stated, "%" is a percentage by weight.

[0105] Herein, the term "omics" refers to a systematic collection of a class of biological molecules, examples of which include, but are not limited to, genomics, proteomics, metabolomics, transcriptomics, protein modification group, transcription modification group, lipidomics, immunomics, glycomics, and RNAomics, etc.

[0106] Herein, the term "target omics" refers to the omics that needs to be analyzed or to be analyzed. The term "reference omics" refers to the omics that belongs to the same class as the target omics, but the information of the target biological molecules in the reference omics is known. The information of the target biological molecules in the reference omics can be known in the art, or can be obtained by any method. These methods include methods known before the present application, such as imaging methods, database collection, etc., or new methods after the present application, methods of the present application, etc.

[0107] In the present context, the term "target sample" refers to a biological sample to be analyzed or detected, non-limiting examples of which include tissues or cells of mammalian origin such as mouse, rat, dog, pig, avian origin such as chicken, human or other animals. The morphology of the target sample is not limited, for example, in some embodiments, the target sample is a fixed biological sample such as a paraformaldehyde fixed sample. In some embodiments, the target sample is a frozen non-fixed sample. In other embodiments, the target sample is a biological tissue which is embedded with a mass spectrometry compatible tissue embedding agent such as Cryo-gel (Leica Microsystems), and optional embedding agents such as OCT can also be chosen.

[0108] In the present context, the term "serial sections" refers to a combination of sections obtained by successively cutting a target region of a target sample. The thickness of each section in the serial sections is preferably such that the spatial distribution of biomolecules in each section of the serial section combination is the same. For this purpose, the thickness of each section in the serial sections is generally controlled to be 25 μm or less, such as 20 μm or less, 15 μm or less, for example 1-15 μm, 2-12 μm, 5-10 μm, etc. The thickness of each section can be freely selected within the above range depending on, for example, the biomolecules to be analyzed, the number of sections in the serial section combination, etc. The thickness of different sections in the serial section combination can be the same or different. The number x of sections in the serial sections is not particularly limited and is generally a natural number of 2 or more, such as 2, 3, 5, 7, 9, 11, 13, 15, etc. On the other hand, x is generally a natural number of 100 or less, such as 50 or less, 30 or less, 20 or less.

[0109] In the present context, the term "reference section" refers to a part of the serial sections, such as a certain section, or several sections, selected or designated from the serial sections. In the case of several sections, the sections can be consecutive or spaced apart. In general, the reference section is not actually further divided into strip-like sub-tissues. Alternatively, the reference section can be subjected to a simulated cutting or sampling process.

[0110] [Method for obtaining spatial distribution of biomolecules in target omics]

[0111] In a first aspect, the present application provides a method for obtaining spatial distribution of biomolecules in target omics, which is sometimes referred to herein as "the method of the present application", which comprises at least the following steps:

[0112] (1) obtaining serial sections of a target sample, and further dividing part of the sections in the serial sections into strip-like sub-tissues at different angles;

[0113] (2) simultaneously or separately obtaining information of biomolecules in the target omics in each strip-like sub-tissue;

[0114] (3) obtaining two-dimensional spatial distribution information of biomolecules of the reference omics in the section; and

[0115] (4) training a learning model based on the two-dimensional spatial distribution information of biomolecules of the reference omics, and then using the trained learning model and one-dimensional information of biomolecules of the target omics to predict two-dimensional spatial information of biomolecules of the target omics.

[0116] Those skilled in the art should understand that steps (1), (2), (3), (4), etc. are only for the purpose of distinguishing different steps and do not mean the order of the steps. The order of the above steps is not particularly limited as long as the purpose of the present application can be achieved. In addition, two or more of the above steps can be combined and performed simultaneously, for example, two or more steps can be performed simultaneously, or two or more steps can be combined into one step. In addition, those skilled in the art should also understand that other steps or operations can be included before and after a particular step or between any of these steps, for example, further optimizing and / or improving the method described in the present application.

[0117] Step (4)

[0118] According to the method of the present application, step (1) is a step of cutting the target sample, which includes two different cuts of the target sample, the first cut is to obtain continuous sections of the target sample, and the second cut is to further cut at least part of the continuous sections into strip-shaped sub-tissues. The cutting method of the two cuts is not particularly limited and can be any known method, such as blade cutting, laser microdissection, etc. The methods of the first cut and the second cut can be the same or different, which is not particularly limited.

[0119] According to the method of the present application, x continuous sections are obtained by the first cut in step (1), which are respectively denoted as T1, T2, T3, …, Tx. The angle of each section placed on the slide is different. For example, when the T1 section is placed along a first direction and the T3 section is placed along a second direction, the first direction and the second direction are different, i.e. an angle can be formed between them. The angle is not limited in size, preferably, the angle between the first direction and the second direction is greater than 0 degrees to less than 180 degrees, preferably the angle is 30-160 degrees, more preferably 45-135 degrees, such as 50 degrees, 60 degrees, 70 degrees, 80 degrees, 90 degrees, 100 degrees, 120 degrees, etc. Similarly, when the number of sections is larger, there are first direction, second direction, third direction, …, and at this time these directions are all different, and an angle can be formed between any two directions. These angles can be the same or different.

[0120] According to the method of the present application, the second cutting in step (1) comprises further cutting of the at least two sections, wherein the angle of the further cutting of each section is different. For example, the direction for the first section and the direction for the second section form an angle of 90 degrees with each other. The number n of the strip-like sub-tissues obtained after the second cutting is not particularly limited, and n can be a natural number, for example, 2-500, preferably 5-400, 10-300, etc., more preferably 20-200, 30-100, such as 30, 40, 50, 60, 70, 80, 90, etc. For the sake of distinction of different strip-like sub-tissues, the first and second strip-like sub-tissues of T1 are exemplarily denoted as T1-1, T1-2, and similarly, the nth strip-like sub-tissue is denoted as T1-n. Similarly, the first and second strip-like sub-tissues of T3 are denoted as T3-1, T3-2, and the nth strip-like sub-tissue is denoted as T3-n. The width of the strip-like sub-tissue after the second cutting is not limited, and is generally 5-200 μm, preferably 10-150 μm, such as 15-100 μm, 20-50 μm, etc. The distance between the adjacent continuous strip-like sub-tissues is between 0-200 μm, preferably 0.5-100 μm, more preferably 1-80 μm, 1-50 μm, etc., such as 2 μm, 4 μm, 5 μm, 6 μm, 8 μm, 10 μm, 15 μm, etc.

[0121] In certain embodiments, the second cutting of step (1) in the method of the present application is performed by a microfluidic device. An exemplary structure of the microfluidic device is described below.

[0122] According to the method of the present application, step (1) further comprises a step of imaging each section in the continuous section to generate a sample image. The imaging method is not limited, and can be performed using optical or fluorescent microscopy.

[0123] Figure 1

[0124] According to the method of the present application, step (2) is to obtain the information of the biomolecules of the target omics in each strip-like sub-tissue, particularly one-dimensional information, simultaneously or separately. Generally, the acquisition of the biomolecule information comprises a step of extracting or labeling the biomolecules in each strip-like sub-tissue and a step of detecting the biomolecules in each strip-like sub-tissue. These steps can be performed by known methods, and the acquisition of the information of different biomolecules is different.

[0125] In certain embodiments, the acquisition of the biomolecule information of the target omics is performed by a specific microfluidic device. See below for the microfluidic device.

[0126] Figure 2

[0127] According to the method of the present application, step (2) is to obtain the two-dimensional spatial distribution information of the biomolecules of the reference omics in the section. The two-dimensional spatial distribution information of the biomolecules of the reference omics needs to be obtained by detection on adjacent tissue sections, and the detection method is not particularly limited.

[0128] In certain embodiments, obtaining the two-dimensional distribution information of the biomolecules of the reference omics includes metabolomics information. Among them, the two-dimensional information of the transcriptome is obtained by spatial transcriptome technology, examples of which include but are not limited to the following methods: imaging-based spatial transcriptomics methods, such as Multiplexed error-robust FISH (MERFISH), in situ sequencing (ISS), Xenium of 10x Genomics, etc.; spatial indexing strategies, such as 10x Visium, Stereo-seq, DBiT-seq, etc.

[0129] In certain embodiments, obtaining the two-dimensional distribution information of the biomolecules of the reference omics includes proteomics information. Among them, the two-dimensional information of the proteomics is obtained by methods including but not limited to the following methods: antibody tag-based proteomics technology, such as multiplexed immunofluorescence (MxIF), nucleic acid sequence-labeled antibody tag (spatial cite-seq), metal tag-labeled antibody (CyTOF mass cytometry); spatial proteomics information obtained by microdissection technology.

[0130] In certain embodiments, obtaining the two-dimensional distribution information of the biomolecules of the reference omics includes histology information. Among them, the two-dimensional information of the histology is obtained by methods including but not limited to the following methods: hemotoxylineosin (HE), Acetylcholinesterase (AchE), Nissl staining, masson staining, etc.

[0131] Figure 2

[0132] According to the method of the present application, step (4) is to train a learning model based on the two-dimensional spatial distribution information of the biomolecules of the reference omics, and then use the trained learning model and the one-dimensional information of the biomolecules of the target omics to predict the two-dimensional spatial information of the biomolecules of the target omics.

[0133] According to the method of the present application, the learning model in step (4) is not limited, examples of which include but are not limited to machine learning models, deep learning models, probabilistic inference models, etc. The present application can use one of the models or a combination of two different models. In an exemplary embodiment, the present application utilizes a transfer learning algorithm in deep learning to perform spatial biomolecular reconstruction.

[0134] [Microfluidic device]

[0135] According to a second aspect of the present application, there is provided a microfluidic device for obtaining spatial distribution of biomolecules in a target omics, sometimes referred to herein simply as "the device of the present application", comprising a sectioning zone provided on a substrate. The sectioning zone is used for performing a second cutting on a section obtained by a first cutting, thereby obtaining a plurality of strip-like sub-tissues.

[0136] According to the device of the present application, the sectioning zone generally comprises a plurality of microchannels. Optionally, the width of the microchannels is variable, and the number of the microchannels is not particularly limited, and is generally 10-500, such as 15, 20, 25, 30, 40, 50, 60, 80, 100, 150, 200, 250, 300, 400, etc. There is a channel wall between adjacent microchannels, and the width of the channel wall is between 500 nm and 200 μm. Preferably, the width of the channel wall is between 800 nm and 100 μm, more preferably 1-80 μm, 2-50 μm. Each microchannel of the sectioning zone has an inlet end and an outlet end, respectively. The diameter of the inlet and outlet is between 1 mm and 4 mm, such as 2 mm, 3 mm, etc. The positional relationship of each microchannel in the sectioning zone is not limited, and preferably each microchannel is arranged in a parallel manner.

[0137] According to the device of the present application, further comprising a cover sheet, which is detachably pressed together with the sectioning zone by a fixing structure. The fixing structure is not limited, and can be a clamp, a buckle structure, etc. The cover sheet is not limited, and can be, for example, a glass slide.

[0138] According to the device of the present application, further comprising a reagent zone, which generally comprises a plurality of reagent receiving cavities. The number of the reagent receiving cavities is not limited, and is generally equal to the number of the microchannels. One end of each reagent receiving cavity is in communication with the inlet end of the corresponding microchannel. The communication between each reagent receiving cavity and the corresponding microchannel is closed communication. The arrangement of each reagent receiving cavity is not particularly limited, and can be in any manner. For example, arranged in a square array.

[0139] The device according to the present application further comprises a negative pressure zone through which the reagent is driven from the reagent zone into the microchannel to reach the cut tissue zone in the microchannel. Preferably, a reaction or extraction of biomolecules is performed in the cut tissue zone. In some embodiments, the negative pressure zone is provided with a plurality of collection chambers, the number of which is not limited and is generally equal to the number of the microchannels, and one end of each of the collection chambers is in communication with the outlet end of a corresponding microchannel. Similar to the reagent storage chamber, the communication of each collection chamber with the corresponding microchannel is closed communication. The arrangement of each collection chamber is not particularly limited and can be in any manner. For example, it can be arranged in a square array.

[0140] In some embodiments, the lengths of the plurality of microchannels in the device according to the present application are equal. Here, the length refers to the distance between, for example, the reagent storage chamber and the collection chamber.

[0141] The device according to the present application can be integrally formed with the cut tissue division zone, the reagent zone and the negative pressure zone, or can be provided separately as three independent structures. As long as the closed communication between the reagent storage chamber, the microchannel and the collection chamber can be achieved. The cut tissue division zone, the reagent zone and the negative pressure zone can be designed in the form of a chip. The material forming the chip is not limited and can be, for example, polydimethylsiloxane (PDMS), silicon dioxide, silicon, polyvinyl chloride, etc.

[0142] [Model construction method]

[0143] In a third aspect of the present application, a method for constructing a model for obtaining the spatial distribution of biomolecules in a target omics is provided, which is sometimes referred to herein as "the construction method of the present application", which generally comprises:

[0144] a. obtaining information of biomolecules of a target omics in a plurality of strip-like sub-tissues obtained from a tissue section;

[0145] b. obtaining two-dimensional spatial distribution information of biomolecules of a reference omics in the tissue section; and

[0146] c. training a learning model based on the two-dimensional spatial distribution information of biomolecules of the reference omics, and then using the trained learning model and the one-dimensional information of biomolecules of the target omics to predict the two-dimensional spatial information of the biomolecules of the target omics.

[0147] The skilled in the art should understand that steps a, b, c, etc. are only for the purpose of distinguishing different steps and do not mean the sequence of the steps. The sequence of the above steps is not particularly limited as long as the purpose of the present application can be achieved. In addition, two or more of the above steps can be combined and performed simultaneously, for example, two or more steps can be performed simultaneously, or two or more steps can be combined into one step. In addition, the skilled in the art should also understand that other steps or operations can be included before and after a particular step or between any of these steps, for example, further optimizing and / or improving the method described in the present application.

[0148] According to the construction method of the present application, a training strategy using a self-encoder is adopted, that is, the reference omics is taken as a training set, the sampling process simulation is taken as an encoder, and the sampling value reconstruction is taken as a decoder, wherein the trained decoder is the reconstruction model used subsequently. The encoder includes parallel strip cutting simulation, random point sampling simulation, and laser cutting simulation. Preferably, parallel strip cutting simulation is used. The decoder of the present application includes a machine learning model, a deep learning model, and a probabilistic inference model. Preferably, a deep learning model is used as the decoder.

[0149] [Device]

[0150] In a fourth aspect, the present application provides a device for obtaining the spatial distribution of biomolecules in a target omics, sometimes referred to herein simply as "the device of the present application", comprising at least one processor; and a memory communicatively connected to the at least one processor. Wherein,

[0151] According to the arrangement of the present application, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the construction method of the third aspect of the present application.

[0152] The device of the present application is not limited in form and can be, for example, a computer, a processor, etc.

[0153] Example 1

[0154] This example is an exemplary microfluidic device for obtaining the spatial distribution of biomolecules in a target omics.

[0155] 1. Microfluidic device structure

[0156] As shown in Figure 2 The microfluidic device of this example is designed in the form of a chip, which includes a reagent area 100, a slice partition area 200, and a negative pressure area 300. The reagent area 100 includes an array of 6X8, i.e. 48 reagent storage cavities 110, each reagent storage cavity 110 being an independent container that can store the required reagent inside. Each reagent storage cavity 110 is in communication with a corresponding microchannel.

[0157] In this embodiment, the slice partition area 200 includes 48 microchannels 210 of equal length and arranged substantially in parallel. In order to make the lengths of the microchannels equal, the arrangement of the microchannels can be adjusted as appropriate. For example, the microchannels are not designed in a straight line form, but are arranged in a zigzag manner.

[0158] In this embodiment, the negative pressure area 300 includes an array of 6X 8, i.e. 48, collection chambers 310, each reagent storage chamber 310 being an independent container.

[0159] In this embodiment, a filter device 120 is arranged between the reagent storage chamber 110 and the microchannel 210, and a filter device 320 is arranged between the microchannel 210 and the collection chamber 310.

[0160] 2. Preparation of the microfluidic device

[0161] The microfluidic chip of this embodiment can be made of polydimethylsiloxane (PDMS). The mold of the chip is made of SU-8 photoresist, which is subjected to steps such as exposure, development and baking to produce a silicon wafer mold for PDMS replication. Then the microchannel pattern on the silicon wafer mold is replicated onto the PDMS chip through a reverse molding process. The PDMS is prepared by mixing the A and B reagents of Dow Corning 184 PDMS at a ratio of 9:1 to prepare a PDMS precursor, stirring and mixing uniformly, degassing for half an hour, then pouring the mixture into the above-mentioned mold, degassing again for 30 minutes, and curing at 60°C for about 4-5 hours. The cured PDMS plate is cut, peeled off, and punched with inlet and outlet holes to complete the manufacturing. The diameter of the inlet and outlet holes is 2-4 mm, which can accommodate a solution of 13-50 μL at most.

[0162] The microfluidic device design of this embodiment includes:

[0163] 1) The chip design includes 48 parallel microchannels in the center, the width of the parallel microchannels is between 10-200 μm, and these microchannels are connected to the same number of inlets and outlets on both sides of the PDMS plate.

[0164] 2) The 48 channels of the microfluidic chip are equal in length, which can effectively reduce the fluid difference between different channels, so that the liquid flows more uniformly over the tissue surface Figure 4 B). Figure 4 C) Comparison of the fluid resistance of the equal-channel-length and unequal-channel-length chips Figure 4 A) shows that the microfluidic flow is more uniform in the equal-channel-length chip, which is more conducive to experimental control.

[0165] 3) The chip and the glass slide are not sealed by permanent bonding, which makes the device difficult to disassemble and use. To solve this problem, the chip is pressed together with the PDMS chip and the glass slide by an acrylic clamp, and the liquid in the inlet of the chip is driven into the channel by negative pressure to reach the tissue area. For each inlet and outlet of the chip, a corresponding round hole is designed on the acrylic plate, which can effectively seal the outlet and inlet of the chip and ensure the sealing effect.

[0166] Example 2

[0167] This example is to verify the reliability of the simulation data verification algorithm.

[0168] The goal of the Flow2Spatial algorithm is to predict the two-dimensional spatial distribution of the target omics on the tissue section according to the one-dimensional information of the target omics. Figure 5 A). The algorithm uses the two-dimensional spatial distribution information of the reference omics molecules to train a deep learning model, and uses the trained deep learning model and the one-dimensional information of the target omics molecules to predict the two-dimensional spatial information of the target omics molecules. Specifically, it is a method of transferring the information of the reference omics, using artificial intelligence to reconstruct the original two-dimensional distribution from the sampled values. The specific steps include: calculating the simulation sampling process to obtain the sampling information of the reference omics; training an artificial intelligence model (autoencoder), taking the sampling information of the reference omics as input and the original information of the reference omics as output; inputting the real sampling value into the trained model to obtain the two-dimensional distribution with fine spatial structure.

[0169] In order to verify the reliability of the Flow2Spatial algorithm, a computer simulation experiment was performed using the published mouse cerebellum spatial transcriptome data set. The spatial distribution of the transcriptome was used as the standard control of the algorithm. Figure 6B). In brief, to mimic the sampling strategy based on microfluidic chip, the tissue was similarly sectioned along two orthogonal axes. For each angle, the tissue sections were cut into consecutive 25 pm strips across the whole tissue, and in each strip, gene expression was obtained by accumulating the expression data of each gene. The original spatial transcriptome profile was compared with the predicted results obtained using Flow2Spatial, and the Tomographer algorithm was also compared with the method of the present embodiment. To evaluate the two-dimensional information predicted by the two algorithms, the reconstruction accuracy of genes characterized by different sparsity and distribution patterns was determined. The Spearman correlation coefficient of all genes, the relative error of the predicted results and the true distribution were calculated. The distribution of these indicators was evaluated, and it was concluded that the present embodiment was significantly better than the Tomographer algorithm in reconstructing the spatial pattern of a given gene. In addition, unlike the Tomographer algorithm with lower reconstruction resolution, Flow2Spatial provided more detailed spatial distribution information and showed higher correlation with the true spatial distribution Figure 7 C), this high-resolution reconstruction obviously helps subsequent clustering analysis. These results demonstrate the high superiority of the algorithm of the present embodiment.

[0170] Embodiment 3

[0171] The present embodiment performs spatial reconstruction on tissue sections of mouse cerebellum, and obtains the spatial distribution of about 4200 proteins in mouse cerebellum, proving the reliability of the method of the present application.

[0172] 1. Sample preparation:

[0173] The sample preparation process in the present embodiment is as follows:

[0174] 1) The target sample is taken out from the animal body, such as the brain, cerebellum, colon, etc. of a mouse;

[0175] 2) The tissue is embedded with a mass spectrometry compatible embedding agent and the tissue is quickly frozen with liquid nitrogen;

[0176] 3) The sample is frozen in a -80°C refrigerator, and is taken out when sectioning is needed.

[0177] It is worth noting that the embedding agent is reasonably selected according to the target omics or no embedding agent is used. For example, when the target omics is proteome, in order to ensure the consistency of the tissue sections and the compatibility with downstream mass spectrometry detection, a mass spectrometry compatible embedding agent Cryo-gel is usually used in the upstream sample preparation. Figure 8 The effects of different embedding agents on downstream mass spectrometry detection are shown as follows: the non-mass spectrometry compatible embedding agent OCT will obviously inhibit the mass spectrometry signal in downstream mass spectrometry detection, while Cryo-gel does not see obvious inhibition of the mass spectrometry signal. In addition, fromFigure 8 It can be seen that when no embedding agent is used, the tissue is prone to edge curling and poor consistency between adjacent sections, while the Cryo-gel embedded tissue has good morphology and high consistency between adjacent sections.

[0178] 2. Tissue sectioning:

[0179] After taking the sample out of the -80 °C refrigerator, it is balanced in the -18 °C microtome for half an hour. After trimming the tissue to the target area, 3 or 4 tissue sections with a thickness of 8-10 μm are cut in succession. The tissue sections are attached to the glass slide as follows: the first tissue section is horizontally attached to the center area of the chip glass slide (the geometric long axis of the tissue section is horizontal), and the third tissue section is vertically attached to the center area of the chip glass slide (the geometric long axis of the tissue section is vertical). The second tissue section is attached to the glass slide according to the reference omics, for example, if spatial transcriptomics is selected as the reference omics, the second tissue section should be attached to the target area of the spatial expression chip (for example, the Visium expression chip glass slide of the 10X Genomics company), if spatial metabolomics is selected as the reference omics, the tissue is attached to the ITO (indium tin oxide) glass slide, and if histological staining is selected as the reference omics, the tissue is attached to the ordinary adhesive glass slide. When multiple omics are selected as the reference omics, the fourth tissue section is needed, and the type of glass slide is selected according to the reference omics. For details, please refer to the second tissue section above. After obtaining the sections, the tissue sections are imaged as soon as possible to obtain the first imaging picture of the tissue.

[0180] 3. Protein sample preparation:

[0181] The two PDMS chips are respectively attached to the first and third tissues, and it is confirmed that the tissues are within the parallel channels, and the PDMS chip and the glass slide are pressed tightly with an acrylic clamp. The first and third tissue sections after pressing the chip are imaged to obtain the second imaging picture. Prepare the tissue lysis solution, which includes trypsin, iRT standard peptide segments, ammonium bicarbonate solution, etc. Intracellular proteases (Lys-C) can also be included in the lysis solution to improve the lysis efficiency. Add 10-20 μl of lysis solution per well to the holes at the inlet of the first and third tissue chips. The negative pressure drives the lysis solution in the inlet to enter the microchannel and flow through the tissue area. After 1-10 min of negative pressure, the chip is placed in a humidifying box for humidification, and then the humidifying box is placed in a 37 °C incubator for 60 min. After incubation, further negative pressure is applied to drive the lysis solution in the inlet to completely flow to the outlet of the chip. The lysis solution at the outlet is sucked out and transferred to a PCR tube, and the PCR is further incubated at 37 °C for 3 h. The tissue lysis solution flows through the chip microchannel to lyse the tissue, and the protein group solution of the strip-shaped tissue is obtained. Figure 9 The red fluorescent protein-labeled mouse cerebellum section before and after lysis in the chip is shown.

[0182] 4. Mass spectrometry detection:

[0183] The lysate sample was first separated by liquid chromatography (LC) and then injected into a mass spectrometer for analysis. The LC model was EASY-nLC 1200 (Thermo Scientific), using a C18 reversed-phase column (75μm ID×20cm, 1.9μm). Dr. Maisch GmbH), the maximum chromatographic pressure was 400 bar. The chromatographic gradient was set as follows: 0–2 min, 6%–12% mobile phase B (80% acetonitrile, 0.1% formic acid); 2–18 min, 12%–30% phase B; 18–22 min, 30%–42% phase B; finally, 95% phase B for 4 min. Molecular mass analysis after chromatographic separation was performed using a Q Exactive HF mass spectrometer (ThermoScientific). Parameters were: data-independent acquisition (DIA) detection mode, scan range 398–1202 m / z, full-spectrum scan resolution 120,000; DIA scan resolution 30,000, NCE: 28%; AGC target: 3e6; maximum injection time: 100 ms. Protein annotation and quantification were performed using Spectronaut software (15.2.210819, Biognosys, Schlieren, Switzerland).

[0184] 5. Optimization of protein sample preparation conditions:

[0185] Because the type of lysis buffer can affect the downstream proteomics sample preparation process, Figure 10 This demonstrates the number of proteins detected under different lysis buffers, i.e., different protein sample preparation methods, such as... Figure 11 As shown, the highest number of proteins were obtained by directly digesting tissues with trypsin, while the number of proteins obtained by using mass spectrometry compatible surfactants such as AZO or DDM to lyse tissues followed by denaturation, reductive alkylation, and trypsin digestion was relatively low.

[0186] On the other hand, the advantage of using trypsin as a lysis buffer for direct in-situ tissue digestion within microchannels is that it reduces the non-specific adsorption of sample proteins onto the chip surface. PDMS, a polydimethylsiloxane, has hydrophobic properties, thus readily adsorbing large molecules such as proteins non-specifically. However, when trypsin is used to digest tissue within the channels, the trypsin is pre-adsorbed onto the microchannel surface, thereby reducing the non-specific adsorption of proteins from the tissue onto the channels. This is crucial for the micro-proteomics within the chip. Figure 12The protein sample was shown to flow through the BSA pretreated microchannel, the untreated microchannel, and the trypsin treated microchannel, and the protein loss was compared. The results showed that trypsin can effectively reduce the non-specific adsorption of sample protein.

[0187] In addition, accurate quantification of protein abundance information in the sample is important for subsequent accurate prediction of the two-dimensional spatial distribution of proteins. Therefore, iRT standard peptides were added as internal standards to assist proteomics quantification in the lysate, i.e. the lysate in each microchannel contains an equal amount of iRT. According to the relative comparison of protein abundance and iRT in the sample, the protein sample can be more accurately quantified by mass spectrometry. Figure 12 The iRT protein sample was shown to exhibit a good linear relationship with the gradient dilution of the sample.

[0188] 6. Mass spectrometry imaging:

[0189] The second piece of sliced tissue was taken out of the -80°C refrigerator and quickly placed in a vacuum dryer for drying for more than 30 min. The matrix solution required for mass spectrometry imaging was prepared, which usually includes organic solvents (such as methanol, acetonitrile, etc.), matrix, and trifluoroacetic acid. The selection of the matrix is mainly based on the characteristics of the compound to be imaged. The optional matrix compounds include but are not limited to: SA (4-hydroxy-3, 5-dimethoxycinnamic acid, mustard acid), CHCA (α-cyano-4-hydroxycinnamic acid), DHB (2, 5-dihydroxybenzoic acid), 1, 5-DAN (1, 5-diaminonaphthalene), etc. The matrix solution is further uniformly covered on the tissue, and the method of covering the matrix on the tissue includes but is not limited to: pneumatic spray, large droplet, sublimation. The prepared glass slide was placed in a mass spectrometer for mass spectrometry imaging, and a suitable resolution (such as 10 μm, 25 μm, 50 μm, 100 μm) was selected for matrix assisted laser desorption / ionization (MALDI-MSI) mass spectrometry imaging.

[0190] Alternatively, other mass spectrometry imaging methods can also be used to obtain two-dimensional information of the reference omics, including but not limited to: desorption electrospray ionization imaging technology (Desorption Electrospray Ionization-Mass Spectrometry Imaging, DESI-MSI), secondary ion mass spectrometry imaging technology (Secondary Ion Mass Spectrometer-Mass Spectrometry Imaging, SIMS-MSI), etc.

[0191] 7. Histological staining:

[0192] In addition to using mass spectrometry imaging to obtain two-dimensional information about tissue compounds for reference omics, the second or third tissue sample can also be used to obtain two-dimensional information about the tissue itself for reference omics. Histological staining includes, but is not limited to, hematoxylin-eosin staining (HE) and acetylcholinesterase (AChE) staining. HE staining can be performed using commercially available staining kits (such as G1120, SolarBio), and the basic steps include: tissue fixation, hematoxylin staining of cell nuclei, eosin staining of cytoplasm, dehydration and mounting. Similarly, AChE staining can also be performed using commercially available staining kits (such as G2111, SolarBio).

[0193] 8. Cross-contamination assessment between passageways:

[0194] To verify whether cross-contamination, or cross-channeling, occurs between samples from parallel channels during in-situ lysis and digestion of tissue in microchannels. Figure 12 The cross-contamination between channels was verified using E. coli. The cross-contamination of tissue after lysis in the chip was tested using three adjacent microchannels. The mass spectrometry detected E. coli protein only in the middle channel and not in the two side channels, indicating that there is almost no cross-contamination of protein samples between channels when tissue sections are lysed in situ in the microchannels.

[0195] 9. Two-dimensional proteome of mouse cerebellum:

[0196] To further illustrate the potential of Flow2Spatial in analyzing complex tissues, the reliability of the method was demonstrated by obtaining spatial proteomics data from mouse cerebellar tissue slices. Figure 12 A). Specifically, five consecutive tissue sections were obtained from the mouse cerebellum. The second section was used for H&E staining, and the other sections were digested in situ on a microarray to obtain one-dimensional protein information (microchannel width = 100 μm). Mass spectrometry was then used to identify the samples in each microchannel. Subsequently, the data from the first three sections were used to reconstruct the two-dimensional information of the proteome using Flow2Spatial, while all data from the first five sections were used to reconstruct the spatial information of the proteome using Tomographer. In this embodiment, approximately 4000 proteins were detected per microchannel, which is at least an order of magnitude higher than the number of proteins obtained from antibody-based spatial proteomics (maximum 300). This significant increase in the number of detected proteins provides a foundation for in-depth research on the spatial proteins of tissues. During the spatial protein distribution reconstruction, the Leiden community detection algorithm was used to cluster each pixel. Figure 13 B). Clustering results showed that protein partitioning was superior to the regions obtained by Tomographer.Figure 13 E), and the brain region distribution identified was basically consistent with that identified by Allen brain. Furthermore, immunofluorescence staining was used to verify the authenticity of the protein spatial distribution obtained by the method in this embodiment. Figure 13 E)(e.g., proteins Mbp and Rbfox3). These results suggest that Flow2Spatial can be used to reveal the heterogeneity of the proteome in tissues.

[0197] Example 4

[0198] This embodiment reconstructs the spatial proteins in tissue sections of rat colonic villi, obtaining the spatial distribution of approximately 2400 proteins from rat colonic villi, further demonstrating the reliability of the method in this application for analyzing the spatial proteome of different tissue types. In this embodiment, the microfluidic chip channel width is 25 micrometers, with a total of 72 microfluidic channels; other operational steps, such as microfluidic device design, sample preparation, tissue slicing, protein sample preparation, mass spectrometry imaging, and mass spectrometry detection, are the same as in Embodiments 1 and 3.

[0199] To verify that our Spring method can perform high-resolution spatial proteomics analysis and is applicable to different tissue types, we used a microfluidic chip with a 25-micron channel width. Figure 13 B) High-resolution spatial proteomic analysis of rat colonic villi ( Figure 13 A) This method can detect over 2400 proteins in a single channel sample, far exceeding previously reported methods. Cluster analysis of the spatial proteomic data of the villi identified four basic tissue types, including intestinal epithelial cells, lamina propria of the villi, and muscle layer. Figure 14 C), which is basically consistent with the tissue structure of the villi. To verify the reliability of the above results, we further stained the colonic villi with antibodies, and the staining results were consistent with the Spring reconstruction results (C). Figure 15 D). Furthermore, we identified numerous proteins related to substance transport in subgroups of epithelial cells, further demonstrating the reliability of our method. Figure 15 E).

[0200] Example 5

[0201] This embodiment divides tissue sections into strip-shaped sub-tissues and further obtains one-dimensional information of tissue metabolomics molecules, thereby obtaining spatial metabolomics information of the tissue. In this embodiment, the sample is not embedded with an embedding agent during sample preparation, and other operation steps such as microfluidic device structure, microfluidic device design, sample preparation, and tissue sectioning are the same as in Embodiment 3.

[0202] Metabolite sample preparation: Two pieces of PDMS chips were respectively adhered to the first and third pieces of tissue, and it was confirmed that the tissues were within the parallel channels. The PDMS chips and glass slides were compressed by using acrylic clamps. The first and third pieces of tissue after compression were imaged to obtain the second imaging picture. Metabolite solvent liquid was added to the holes of the inlet of the first and third pieces of tissue chip one by one, 10-20 μl / hole of solvent. The types of solvent liquid include but are not limited to pure methanol, 80% methanol, acetonitrile, etc. Optionally, an appropriate amount of formic acid molecules can also be added to the solvent liquid. The solvent liquid at the inlet was driven into the microchannel and flowed through the tissue area by negative pressure. After 1-10 min of negative pressure, the solvent liquid at the inlet was completely flowed to the outlet of the chip. The lysate at the outlet was sucked out and transferred to the corresponding PCR tube one by one. The structure, abundance, etc. of the compound molecules in each PCR tube were detected by liquid chromatography-mass spectrometry (LC-MS) or other metabolite detection methods. Figure 16

[0203] The subsequent steps are the same as those in Example 3.

[0204] Example 6

[0205] In this example, the tissue slices are divided into strip sub-tissues, and further one-dimensional information of the tissue transcriptomics molecules is obtained, and then the spatial transcriptome information of the tissue is obtained. In this example, the pretreatment steps such as microfluidic device, microfluidic device design, sample preparation, tissue slice, etc. are the same as those in Example 3.

[0206] Transcriptome sample preparation: Two pieces of PDMS chips were respectively adhered to the first and third pieces of tissue, and it was confirmed that the tissues were within the parallel channels. The PDMS chips and glass slides were compressed by using acrylic clamps. The first and third pieces of tissue after compression were imaged to obtain the second imaging picture. The reverse transcription solution mixture containing channel-specific barcode nucleic acids, reverse transcriptase, dNTP, etc. was added to the holes of the inlet of the first and third pieces of tissue chip one by one. The reverse transcription solution at the inlet was driven into the microchannel and flowed through the tissue area by negative pressure. After 1-10 min of negative pressure, the reverse transcription liquid filled the entire microchannel. The microfluidic device was placed in a wet box, and the wet box was placed in a constant temperature box at 65°C, so that the mRNA molecules were reversely transcribed into cDNA molecules in situ in the tissue. After reverse transcription, the microchannel was washed with PBS, etc. to clean the unreacted reverse transcription solution. Next, the whole piece of tissue was digested with lysate, and the subsequent steps were carried out: cDNA purification, strand displacement, cDNA amplification, NGS library construction, sequencer sequencing, etc. ​ B and C are the capillary electrophoresis results of the cDNA amplification products, and the number of genes detected in each channel.

[0207] The subsequent steps are the same as those in Example 3. ​

[0208] Channel-specific barcoding nucleic acids include a 22-mer PCR handle functionalized with biotin at the end, a 10-mer unique molecular identifier (UMI), an 8-mer unique spatial barcode, and a 16-mer poly-T sequence ​ A).

[0209] Example 7

[0210] This example separates tissue sections into strip sub-tissues and further obtains one-dimensional information of tissue transcriptomics molecules, and then obtains spatial transcriptome information of the tissue. In this example, the pretreatment steps such as microfluidic device, microfluidic device design, sample preparation, tissue section, etc. are the same as in Example 3.

[0211] Metabolite sample preparation is the same as in Example 5. The rehydration reagent and dilute solution are respectively added to the holes of the first and third tissue chip inlet. After negative pressure for 1-10 min, the solvent liquid at the inlet is completely flowed to the chip outlet, and the lysate liquid at the outlet is sucked out and discarded one by one. Prepare reverse transcription, and the transcriptome sample preparation is the same as in Example 6. The transcriptome and proteome are separated by magnetic beads, and then the proteome is detected by mass spectrometer and the transcriptome is detected by sequencing ​ ).

Claims

1. A method for obtaining a spatial distribution of biomolecules in a target omics, characterized in that, The method comprises the following steps: (1) obtaining continuous sections of a target sample, and further dividing part of the sections into strip sub-tissues at different angles, specifically comprising: obtaining x continuous sections of a target region of a target sample, denoted as T1, T2, T3, …, Tx respectively, and placing the tissue sections in the regions of interest of substrates, wherein x is a natural number greater than or equal to 3, the T1 section is placed along a first direction, the T3 section is placed along a second direction, and the first direction and the second direction are different; imaging the continuous sections T1, T2, T3, …, Tx to generate sample images; and further dividing the sections T1 and T3 into strip sub-tissues, respectively, wherein the first and second strip sub-tissues of T1 are denoted as T1-1, T1-2, …, and the nth strip sub-tissue of T1 is denoted as T1-n, and the first and second strip sub-tissues of T3 are denoted as T3-1, T3-2, …, and the nth strip sub-tissue of T3 is denoted as T3-n, wherein n is a natural number ranging from 2 to 500; and imaging the divided continuous sections T1, T3, …, Tx to generate sample images; (2) simultaneously or separately obtaining one-dimensional information of biomolecules of a target omics in each strip sub-tissue; (3) taking part of the continuous sections as reference sections, obtaining two-dimensional spatial distribution information of biomolecules of a reference omics in the reference sections; and (4) training a deep learning model based on the two-dimensional spatial distribution information of biomolecules of the reference omics, and then predicting two-dimensional spatial information of biomolecules of the target omics based on the trained model and the one-dimensional information of biomolecules of the target omics by using a transfer learning strategy, specifically comprising: taking the reference omics as a training set to construct a training data generator; then constructing a deep learning model based on an autoencoder, which generates strip sub-tissue data by electronically cutting each spatial data simulated by the data generator, and further constructing a connection between the spatial data and the strip sub-tissue data; and finally using the model trained by the reference omics data to reconstruct the real target omics spatial information by using a transfer learning strategy.

2. The method for obtaining spatial distribution of biomolecules in a target omics according to claim 1, characterized in that, In step (1), part of the sections is further divided into strip sub-tissues by laser microdissection or a microfluidic device.

3. The method for obtaining spatial distribution of biomolecules in a target omics according to claim 2, characterized in that, The microfluidic device comprises a section division area comprising 10-500 variable-width microchannels, and a channel wall exists between each microchannel, and the width of the channel wall is between 500 nm and 200 μm.

4. The method for obtaining spatial distribution of biomolecules in target omics according to claim 2, characterized in that, The width of the strip sub-tissue obtained by laser microdissection is 5-200 μm, and the distance between adjacent continuous strip sub-tissues is between 0 and 200 μm.

5. The method for obtaining spatial distribution of biomolecules in target omics according to claim 1, wherein, In step (2), the target omics is proteomics, and the acquisition of one-dimensional information of biomolecules comprises: respectively delivering tissue lysis reagents containing reference peptides to the regions of at least the strip sub-tissues of the sections T1 and T3, so that the tissue lysis reagents lyse the strip sub-tissues T1-1…T1-n and T3-1…T3-n in the regions of interest of the substrates; delivering the lysate solution of the strip sub-tissue T1-1…T1-n and T3-1…T3-n from the region of interest of the substrate to the microfluidic device outlet, and transferring each strip sub-tissue solution to a separate container, respectively; detecting the biomolecule information in the lysate of T1-1…T1-n and T3-1…T3-n, respectively, using liquid chromatography-mass spectrometry to obtain the proteomic one-dimensional physicochemical information of slice T1 at angle a and slice T3 at angle b.

6. The method for obtaining the spatial distribution of biomolecules in the target omics according to claim 1, wherein in step (2), the target omics is metabolomics, and the obtaining of the one-dimensional information of the biomolecules comprises: delivering the solvent of the target compound to the regions of the strip sub-tissues T1-1…T1-n and T3-1…T3-n, respectively, to extract the compounds in the strip sub-tissues T1-1…T1-n and T3-1…T3-n in the region of interest of the substrate; delivering the solvent in which the compounds of the strip sub-tissues T1-1…T1-n and T3-1…T3-n are extracted from the region of interest of the substrate to the microfluidic device outlet, and transferring each strip sub-tissue solution to a separate container, respectively; detecting the compound information in the solutions of T1-1…T1-n and T3-1…T3-n, respectively, using liquid chromatography-mass spectrometry to obtain the metabolomic one-dimensional physicochemical information of slice T1 at angle a and slice T3 at angle b.

7. The method for obtaining the spatial distribution of biomolecules in the target omics according to claim 1, wherein in step (2), the target omics is transcriptomics, and the obtaining of the one-dimensional information of the biomolecules comprises: delivering the reverse transcription reagent and the barcoded polynucleotide to the regions of the strip sub-tissues T1-1…T1-n and T3-1…T3-n to generate cDNA in the strip sub-tissues T1-1…T1-n and T3-1…T3-n in the region of interest of the substrate; delivering the lysis buffer or denaturation reagent to the strip sub-tissues to generate lysed or denatured tissue samples; extracting the cDNA from the lysed or denatured tissue samples; and sequencing the sequencing library constructed from the cDNA to generate cDNA reads.

8. The method for obtaining spatial distribution of biomolecules in a target omics according to claim 7, characterized in that, The barcoded polynucleotide comprises a PCR handle end sequence, a strip sub-tissue barcode sequence, a unique molecular identifier sequence, and a polyT sequence, and optionally, wherein the PCR handle end sequence is end-functionalized with biotin.

9. The method for obtaining spatial distribution of biomolecules in target omics of claim 1, wherein, In step (2), the target omics is genomics, and the obtaining of the one-dimensional information of the biomolecules comprises: delivering the histone removal reagent to the regions of the strip sub-tissues T1-1…T1-n and T3-1…T3-n to remove the histones in the genome in the strip sub-tissues T1-1…T1-n and T3-1…T3-n in the region of interest of the substrate; delivering the fragmentation reagent to the strip sub-tissues to generate fragmented genomes; lysing the strip sub-tissues T1-1…T1-n and T3-1…T3-n and delivering the lysate from the region of interest of the substrate to the microfluidic device outlet, and transferring each strip sub-tissue solution to a separate container, respectively; amplifying the genomic fragments in the container; and sequencing the genomic amplification products to obtain sequence information.

10. The method for obtaining spatial distribution of biomolecules in target omics of claim 1, wherein, In step (2), the target omics is multi-omics, and the acquisition of one-dimensional information of biomolecules includes: respectively delivering a solvent of the target compound to the regions of the long strip sub-tissues T1-1…T1-n and T3-1…T3n, so that the compound solvent extracts the compounds in the strip sub-tissues T1-1…T1-n and T3-1…T3n in the region of interest of the substrate; delivering the compound solvent extracted from the strip sub-tissues T1-1…T1-n and T3-1…T3-n from the region of interest of the substrate to the outlet of the microfluidic device, and respectively transferring each strip sub-tissue solution to an independent container; respectively detecting the compound information in the T1-1…T1-n and T3-1…T3-n solutions by liquid chromatography-mass spectrometry, and further obtaining the one-dimensional physicochemical information of the metabolome of the tissue piece T1 at the angle α and the tissue piece T3 at the angle β; respectively delivering a rehydration reagent and a cleaning reagent to the separated long strip sub-tissue regions of the tissue sections T1 and T3, and discarding each strip sub-tissue outflow solution; delivering a reverse transcription reagent and a barcoded polynucleotide to the regions of the strip sub-tissues T1-1…T1-n and T3-1…T3n to generate cDNA in the strip sub-tissues T1-1…T1-n and T3-1…T3-n in the region of interest of the substrate; delivering a lysis buffer or a denaturing reagent to the strip sub-tissues to generate a lysed or denatured tissue sample; delivering the lysate solution of the strip sub-tissues after lysis from the region of interest of the substrate to the outlet of the microfluidic device, and respectively transferring each strip sub-tissue lysate solution to an independent container, and using magnetic beads to separate the cDNA and protein lysates in the solution; respectively detecting the biological information including the sequence information, protein modification information and abundance information of the proteins in the lysates of T1-1…T1-n and T3-1…T3-n by liquid chromatography-mass spectrometry, and further obtaining the two-dimensional physicochemical information of the proteome of the section T1 at the angle α and the section T3 at the angle β; and performing library construction and sequencing on the cDNA to generate cDNA reads.

11. The method for obtaining spatial distribution of biomolecules in target omics of claim 1, wherein, The two-dimensional spatial distribution information of the biomolecules of the reference omics in the section obtained in step (3) includes at least one selected from the group consisting of metabolome, transcriptome, proteome and histology information.

12. The method for obtaining spatial distribution of biomolecules in a target omics according to claim 11, characterized in that, The two-dimensional information of the metabolome is obtained by mass spectrometry imaging technology, including at least one of matrix-assisted laser desorption mass spectrometry imaging technology, electrospray desorption ionization imaging technology and secondary ion mass spectrometry imaging technology.

13. The method for obtaining spatial distribution of biomolecules in a target omics according to claim 11, wherein, The two-dimensional information of the transcriptome is obtained by spatial transcriptome technology, including microimaging-based spatial transcriptome technology and / or spatial nucleic acid tag array-based technology.

14. The method for obtaining spatial distribution of biomolecules in target omics of claim 11, wherein, Wherein, The two-dimensional information of the proteome is obtained by at least one of fluorescent antibody tag-based proteome technology, nucleic acid sequence-labeled antibody tag and / or metal tag-labeled antibody.

15. The method for obtaining spatial distribution of biomolecules in target omics of claim 11, wherein, The obtaining of the two-dimensional information of the histology includes at least one of hematoxylin-eosin staining, acetylcholinesterase staining, Nissl staining, and Masson staining.

16. The method for obtaining spatial distribution of biomolecules in target omics according to claim 1, wherein the encoder comprises at least one of parallel strip simulation, random point simulation, and laser cutting simulation; and the decoder comprises at least one of a machine learning model, a deep learning model, and a probabilistic inference model.

17. The method for obtaining spatial distribution of biomolecules in target omics of claim 1, wherein, Step (4) comprises training a learning model with reference to the two-dimensional spatial distribution information of the biomolecules of the reference omics, and then predicting the two-dimensional spatial information of the biomolecules of the target omics by using the trained learning model and the one-dimensional information of the biomolecules of the target omics.

18. A microfluidic device, characterized in that, It is an apparatus for implementing the method according to any one of claims 1-17, which comprises a slice segmentation area arranged on a substrate, the slice segmentation area comprising 10-500 microchannels, each microchannel being separated from another by a channel wall, and the width of the channel wall being between 500 nm and 200 μm, and each microchannel having an inlet end and an outlet end.

19. The microfluidic device for obtaining spatial distribution of biomolecules in a target omics according to claim 18, wherein, The microchannels are arranged in parallel, and the lengths of the microchannels are equal.

20. The microfluidic device for obtaining spatial distribution of biomolecules in a target omics of claim 18, wherein, Further comprising a cover sheet, and the cover sheet and the slice segmentation area are detachably pressed together by a fixing structure.

21. The microfluidic device for obtaining spatial distribution of biomolecules in a target omics of claim 18, wherein, Further comprising a reagent area, and the reagent in the reagent area enters each microchannel through the inlet end of the microchannel.

22. The microfluidic device for obtaining spatial distribution of biomolecules in a target omics according to claim 21, wherein, The reagent area is provided with a plurality of reagent storage cavities, and one end of each reagent storage cavity is in communication with the inlet end of a corresponding microchannel.

23. The microfluidic device for obtaining spatial distribution of biomolecules in a target omics of claim 21, wherein, Further comprising a negative pressure area, and the reagent is driven from the reagent area to the cutting tissue area in the microchannel by the negative pressure area.

24. The microfluidic device for obtaining spatial distribution of biomolecules in a target omics according to claim 23, wherein, The negative pressure area is provided with a plurality of collection cavities, and one end of each collection cavity is in communication with the outlet end of a corresponding microchannel.

25. The microfluidic device for obtaining spatial distribution of biomolecules in a target omics of claim 23, wherein, The path of each reagent storage cavity to the corresponding collection cavity through the microchannel can form a closed connection.

26. The microfluidic device for obtaining spatial distribution of biomolecules in a target omics of claim 23, wherein, The path of each reagent storage cavity to the corresponding collection cavity through the microchannel has an equal length. 27.A method for constructing a transfer learning model, characterized in that, It comprises: a. obtaining information of biomolecules of target omics in a plurality of strip sub-tissues obtained from a tissue slice; b. obtaining two-dimensional spatial distribution information of biomolecules of reference omics in adjacent tissue slices; and c. training a learning model based on the two-dimensional spatial distribution information of biomolecules of the reference omics, and then predicting the two-dimensional spatial information of the biomolecules of the target omics by using the trained learning model and the one-dimensional information of the biomolecules of the target omics; The transfer learning model is a deep learning model in the method according to any one of claims 1-17.

28. An apparatus for obtaining spatial distribution of biomolecules in target omics, comprising: at least one processor; and a memory in communication connection with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the construction method according to claim 27.

Images