Method for processing single cell level super enhancement subdata

Abstract

The invention relates to a method for processing single cell level super enhancer data, which belongs to the technical field of artificial intelligence processing of biological information, and comprises the following steps: step 1): collecting and sorting tumor single cell scATAC-seq and cell type specific marker gene data; step 2): selecting and analyzing an appropriate clustering algorithm of tumor single cell scATAC-seq data; step 3): obtaining a cluster gene region signal and a tumor single cell scATAC-seq data clustering cluster; 4) identifying the cancer cell type in the tumor and a non-coding region regulation unit of the cancer cell subtype of the cancer tissue; step 5): identifying a common enhancer of the cancer cell subtype of the tumor tissue through an scEnhancer algorithm; step 6), identifying super enhancers in the enhancers of the tumor tissue cancer cell subtypes authenticated in the step 5) through an ROSE algorithm; the method has the beneficial effects that the super enhancers in the enhancers of the tumor tissue cancer cell subtypes are identified and authenticated through the ROSE algorithm and the scEnhancer algorithm.

Description

Technical Field

[0001] The present invention belongs to the technical field of artificial intelligence processing of biological information, and in particular relates to a method for processing super enhancer data at the single-cell level. Background Art

[0002] Super enhancers are key driving factors that regulate abnormal high expression of tumor oncogenes in the remote non-coding regions of DNA and induce cancer. They play an important role in at least 18 common cancers, including liver cancer, lung cancer, gastric cancer, kidney cancer, intestinal cancer, breast cancer, blood cancer and lymphoma. For example, super enhancers can guide the abnormal transcription of oncogenes such as AJUBA and SIRT7 in liver cancer, LMO1 in neuroblastoma, CEBPA, IRF8 and ETV6 in acute myeloid leukemia, FOXC1, MET and ANLN in triple-negative breast cancer, and C19MC-LIN28A-MYCN in ETMR brain tumors, leading to the generation of cancer or maintaining its abnormal proliferation, migration and invasion characteristics.

[0003] At present, there are relatively few studies on super enhancers and their heterogeneity in tumor cells, and all of them are focused on the bulk level containing at least millions of cells and ignoring tumor heterogeneity. There are defects such as excessive background noise, high false positive rate and key cancer cell types being easily overlooked. It is impossible to comprehensively, accurately, systematically and efficiently reveal the regulation of super enhancers and potential oncogenes in key cancer cell types within the tumor (such as cancer cell populations containing trace cells in the early stage of the tumor). Summary of the invention

[0004] The technical problem to be solved by the present invention is how to identify super enhancers in tumor cells. The present invention provides a method for processing super enhancer data at the single-cell level. The ROSE algorithm is used to identify super enhancers in enhancers of authenticated tumor tissue cancer cell subtypes. The ROSE algorithm is used to draw a cumulative curve of the signal intensity in the enhancer region and find the inflection point of the cumulative curve. The enhancer after the inflection point is defined as a super enhancer, thereby realizing the ability to identify super enhancers using tumor single-cell ATAC-seq data, and further improving the analysis accuracy of gene regulatory elements.

[0005] In order to achieve the above object, the present invention is implemented by the following technical solutions:

[0006] A method for processing super enhancer data at the single cell level comprises the following steps:

[0007] Step 1): Collect and organize tumor single-cell scATAC-seq data and cell type-specific marker genes;

[0008] Step 2): Select a clustering algorithm suitable for tumor single-cell scATAC-seq data analysis;

[0009] Step 3): Obtain cluster gene region signals and tumor single cell scATAC-seq data clustering clusters;

[0010] Step 4): Identify the non-coding regulatory units of cancer cell types within the tumor and cancer cell subtypes of cancer tissues;

[0011] Step 5): Identify common enhancers of cancer cell subtypes of tumor tissues through the scEnhancer algorithm;

[0012] Step 6): Using the ROSE algorithm, identify super enhancers among the enhancers of the tumor tissue cancer cell subtype identified in step 5).

[0013] Optionally, in step 2), the clustering tool for tumor single-cell scATAC-seq data is clustered using Signac software.

[0014] Optionally, in step 6), super enhancers are identified in the tumor cell subpopulation ATAC-seq revealed in the tumor single cell scATAC-seq data using the ROSE algorithm, specifically using the following steps:

[0015] Step S1): Preliminary peak calling: Use tumor single-cell scATAC-seq data and standard peak calling tools to identify all potential enhancers in the peak signals of the entire DNA non-coding region. Enhancer peaks represent potential regulatory regions.

[0016] Step S2): Peak expansion: The ROSE algorithm is used to merge adjacent enhancer peaks whose distance is less than the threshold in order to capture large regulatory regions, and the parameters of the transcription start site exclusion zone are set by the user to exclude false positive signals near the transcription start site, so as to identify super enhancers composed of multiple adjacent enhancers;

[0017] Step S3): Signal intensity calculation: Calculate the signal intensity of the tumor single-cell scATAC-seq data for the merged enhancer regions, and use the standardized reading signal to measure the intensity of each enhancer region;

[0018] Step S4): Sorting: Sorting all the merged enhancers according to the signal intensity of each enhancer region in order to distinguish between common enhancers and super enhancers;

[0019] Step S5): Determine the threshold: Draw a cumulative curve of the signal intensity of the enhancer region using the ROSE algorithm and find the inflection point of the cumulative curve, and define the enhancer after the inflection point as a super enhancer.

[0020] Optionally, in step S1), after the peak tuning tool of the MASC2 software is used to identify the regulatory region of the enhancer peak, the ROSE algorithm is used to divide each genomic site identified by the MACS2 software into multiple bins, and the read density is mapped to each bin to improve the resolution of the signal intensity within each genomic site.

[0021] Furthermore, the size of each bin or the length of the genomic region covered by each bin is calculated using the following formula (1):

[0022] (1);

[0023] in, is the length of the genomic locus, Indicates the number of bins, The length of the genomic region covered by each bin.

[0024] Furthermore, in each bin at a genomic locus, the nth bin at the genomic locus is B n , and 1≤n≤N, the signal value S of each bin Bn It is expressed as formula (2):

[0025] (2);

[0026] Among them, i represents the starting genomic coordinate of each bin, d is the length of each bin, and R s,x Represents the signal value reading on the encoding chain at the x coordinate, R a,x Represents the signal value reading on the non-coding strand at the x-coordinate.

[0027] Furthermore, the signal value of each bin is normalized, and the ROSE algorithm is used to normalize the read density of each bin by the total million mapped reads MMR, specifically formula (3):

[0028] (3);

[0029] in, Normalization of the read density for the signal value for each bin.

[0030] Optionally, in step S4), V is represented as a vector arranged in ascending order, which contains the read densities of all the enhancers. V can be represented as: {v1,v2,v 3,…, v n}, where v represents the read density of each enhancer in the genome. For any enhancer with negative read density, its value will be set to zero. The elements in vector V are: 0≤v1≤v2≤v3≤ … ≤v n .

[0031] Furthermore, the slope of the straight line tangent to the read density curve of the enhancer is m, which is calculated as formula (4):

[0032] (4);

[0033] Among them, V max and V min Respectively represent the maximum and minimum values ​​in the vector V arranged in ascending order, L V Represents the total length of vector V in ascending order.

[0034] Furthermore, when the straight line is tangent to the read density curve, the number of points below the straight line is minimized, and the coordinate point of the intersection of the straight line and the read density curve on the x-coordinate is X cf , the number of points below the straight line is minimized by the following formula (5):

[0035] (5);

[0036] In formula (5), i represents the index of each enhancer in vector V, v i represents the read density of the i-th enhancer in the genome, b is the y-intercept of the line, and x i is the intersection of the ith straight line on the x coordinate and the reading density curve, b increases with x i The dynamic changes of are calculated as shown in formula (6):

[0037] (6);

[0038] In formula (6), y i represents the read density of the enhancer at the ith position in the vector V, and m is the slope of the line tangent to the read density curve of the enhancer;

[0039] The cutoff value Y between common enhancers and super enhancers cf It is expressed as formula (7):

[0040] (7);

[0041] In formula (7), since the intersection of the straight line and the reading density curve is at the x-coordinate point X cf ,but The coordinate point value on the x-coordinate of the intersection of the straight line on the vector V arranged in ascending order and the read density curve is used as the boundary value between common enhancers and super enhancers.

[0042] Beneficial effects of the present invention:

[0043] Based on the ROSE algorithm, the present invention uses the ROSE algorithm to identify super enhancers in enhancers of tumor tissue cancer cell subtypes identified in the identification, draws a cumulative curve of the signal intensity of the enhancer region through the ROSE algorithm and finds the inflection point of the cumulative curve, defines the enhancer after the inflection point as a super enhancer, and realizes the ability to identify super enhancers using tumor single-cell ATAC-seq data, further improving the analysis accuracy of gene regulatory elements. It not only helps to locate key regulatory regions more accurately, but also provides deeper insights into understanding cell heterogeneity and gene expression regulatory mechanisms, thereby promoting the development of precision medicine and targeted therapy. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0045] Figure 1 It is a work flow chart of the present invention;

[0046] Figure 2 It is a regulation diagram of the common enhancer of the present invention on the gene;

[0047] Figure 3 This is a diagram of abnormal regulation of genes by the super enhancer of the present invention;

[0048] Figure 4 This is an identification diagram of super enhancers based on signal strength arrangement of the present invention;

[0049] Figure 5 This is a signal strength comparison diagram of the super enhancer of the present invention and the common enhancer. DETAILED DESCRIPTION

[0050] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0051] Example 1

[0052] A method for processing super enhancer data at the single cell level comprises the following steps:

[0053] Step 1): Collect and organize tumor single-cell scATAC-seq and cell type-specific marker gene data;

[0054] Step 2): Select and analyze appropriate clustering algorithms for tumor single-cell scATAC-seq data;

[0055] Step 3): Obtain cluster gene region signals and tumor single cell scATAC-seq data clustering clusters;

[0056] Step 4): Identify the non-coding regulatory units of cancer cell types within the tumor and cancer cell subtypes of cancer tissues;

[0057] Step 5): Identify common enhancers of cancer cell subtypes of tumor tissues through the scEnhancer algorithm;

[0058] Step 6): Using the ROSE (Rank Ordering of Super-Enhancers) algorithm, super enhancers are identified among the enhancers of the tumor tissue cancer cell subtype identified in step 5).

[0059] Specifically, in step 2), the clustering tool for tumor single-cell scATAC-seq samples is clustered using Signac software.

[0060] In step 6), the ROSE (Rank Ordering of Super-Enhancers) algorithm has significant advantages in identifying super enhancers in tumor single-cell scATAC-seq data. The specific workflow can be divided into the following steps:

[0061] Step S1): Initial Peak Calling: Use tumor single-cell scATAC-seq data and standard peak calling tools (such as MACS2) to identify all enhancer peaks in the genome (all potential enhancers in the peak signals of the entire DNA non-coding region in the genome). Enhancer peaks represent potential regulatory regions.

[0062] Step S2): Peak Stitching: The ROSE algorithm is used to merge adjacent enhancer peaks whose distance is less than a threshold (usually 12.5 kb) to capture large regulatory regions. The parameters of the transcription start site exclusion zone (TSS exclusion zone) are set by the user to exclude false positive signals near the transcription start site, so as to identify super enhancers composed of multiple adjacent enhancers.

[0063] Step S3): Signal Strength Calculation: Calculate the signal strength of the tumor single-cell scATAC-seq data for the merged enhancer regions, and use the standardized reading signal (such as RPKM or RPM) to measure the strength of each enhancer region;

[0064] Step S4): Rank Ordering: rank all the merged enhancers according to the signal strength of each enhancer region, so as to distinguish between common enhancers and super enhancers in step S5);

[0065] Step S5): Determine the cutoff determination: Draw the cumulative curve of the signal intensity of the enhancer region through the ROSE algorithm and find the inflection point of the cumulative curve (usually achieved by calculating the quantile of the enhancer signal intensity), and define the enhancer after the inflection point as a super enhancer. These super enhancers usually have higher signal intensity, indicating that they may play an important role in gene regulation.

[0066] In step S1), after the regulatory region of the enhancer peak is identified by the peak tuning tool of the MACS2 software, the ROSE algorithm is used to divide each genomic site identified by the MACS2 software into multiple bins, and the read density is mapped to each bin to improve the resolution of the signal intensity within each genomic site.

[0067] Furthermore, the size of each bin or the length of the genomic region covered by each bin is calculated using the following formula (1):

[0068] (1);

[0069] in, is the length of the genomic locus, Indicates the number of bins, The length of the genomic region covered by each bin.

[0070] Furthermore, in each bin at a genomic locus, the nth bin at the genomic locus is B n , and 1≤n≤N, the signal value S of each bin Bn It is expressed as formula (2):

[0071] (2);

[0072] Among them, i represents the starting genomic coordinate of each bin, d is the length of each bin, and R s,x Represents the signal value reading on the encoding chain at the x coordinate, Ra,x Represents the signal value reading on the non-coding strand at the x-coordinate.

[0073] Furthermore, the signal value of each bin is normalized, and the ROSE algorithm is used to normalize the read density of each bin by the total million mapped reads MMR, specifically formula (3):

[0074] (3);

[0075] in, Normalization of the read density for the signal value for each bin.

[0076] In step S4), V is represented as a vector in ascending order, which contains the read densities of all the enhancers. V can be represented as: {v1,v2,v 3,…, v n}, where v represents the read density of each enhancer in the genome. For any enhancer with negative read density, its value will be set to zero. The elements in vector V are: 0≤v1≤v2≤v3≤ … ≤v n .

[0077] Furthermore, the slope of the straight line tangent to the read density curve of the enhancer is m, which is calculated as formula (4):

[0078] (4);

[0079] Among them, V max and V min Respectively represent the maximum and minimum values ​​in the vector V arranged in ascending order, L V Represents the total length of vector V in ascending order.

[0080] Furthermore, when the straight line is tangent to the read density curve, the number of points under the straight line should be minimized, and the coordinate point of the intersection of the straight line and the read density curve on the x-coordinate is X cf , the number of points below the straight line is minimized by the following formula (5):

[0081] (5);

[0082] In formula (5), i represents the index of each enhancer in vector V, v i represents the read density of the i-th enhancer in the genome, b is the y-intercept of the line, and x i is the intersection of the ith straight line on the x coordinate and the reading density curve, b increases with x i The dynamic changes of are calculated as shown in formula (6):

[0083] (6);

[0084] In formula (6), y i represents the read density of the enhancer at the ith position in the vector V, and m is the slope of the line tangent to the read density curve of the enhancer;

[0085] The cutoff value Y between common enhancers and super enhancers cf It is expressed as formula (7):

[0086] (7);

[0087] In formula (7), since the intersection of the straight line and the reading density curve is at the x-coordinate point X cf ,but The coordinate point value on the x-coordinate of the intersection of the straight line on the vector V arranged in ascending order and the read density curve is used as the boundary value between common enhancers and super enhancers.

[0088] Example 2

[0089] Based on Example 1, the selection of clustering algorithm is as follows:

[0090] An excellent clustering algorithm is essential for identifying super enhancers using tumor single-cell ATAC-seq data. Super enhancers are usually expressed as highly active regulatory regions in the genome, and their identification relies on the accurate classification of different cell types in single-cell data. Accurate clustering can better capture the activity patterns of super enhancers in different cell populations, thereby revealing their roles in regulating gene expression. Incorrect clustering may dilute or lose signals in key regulatory regions, thus affecting downstream functional analysis and biological conclusions. Therefore, choosing a clustering algorithm that can provide high accuracy and stability is essential for accurately identifying super enhancers and understanding their biological functions. When choosing a suitable clustering tool for tumor single-cell scATAC-seq data, we compared the performance of multiple tools and finally decided to use Signac. This choice was based on many considerations. First, although SnapATAC has outstanding clustering accuracy, its installation process is complicated, has high requirements for data format, and is prone to errors during operation. In contrast, Signac is not only easier to install and run, but also has significantly improved operating efficiency and reduced memory consumption after multiple updates. In addition, Signac improves clustering accuracy through optimized quality control indicators and improved dimensionality reduction algorithms, such as latent semantic indexing (LSI) and uniform manifold approximation and projection (UMAP). In particular, Signac can also calculate tumor single-cell ATAC-seq peak signals within gene coding regions to define gene expression signal values, which further enhances its practicality in biological analysis. Therefore, based on the tool's stability, accuracy, and ease of use, we chose Signac as the single-cell scATAC-seq data clustering analysis tool for this project.

[0091] We used Signac to remove the batch effect of samples in our analysis. Specifically, we used the "by common feature" method in Signac to integrate different batches of tumor single-cell scATAC-seq datasets. This method successfully reduced the bias caused by batch differences by matching and normalizing common features, ensuring the accuracy and consistency of clustering results.

[0092] Example 3

[0093] Based on Example 1, Figure 2-Figure 5 It is a comparative analysis of super enhancers and ordinary enhancers. Figure 2 It is the regulation of genes by common enhancers. Figure 3 It is the abnormal regulation of genes by super enhancers. Figure 4 It is the identification of super enhancers based on the arrangement of signal strength. Figure 5 It is a comparison of the signal strength between super enhancers and ordinary enhancers.

[0094] Throughout the specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in conjunction with the embodiment or example is included in at least one embodiment of the present invention. Therefore, the phrases "one embodiment," "an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily all refer to the same embodiment or example. In addition, particular features, structures, or characteristics may be combined in one or more embodiments or examples in any suitable combination and / or sub-combination. In addition, it will be appreciated by those of ordinary skill in the art that the figures provided herein are for illustrative purposes and that the figures are not necessarily drawn to scale. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0095] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope of the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for processing super enhancer data at the single cell level, characterized in that: The steps include: Step 1): Collect and organize tumor single-cell scATAC-seq data and cell type-specific marker gene data; Step 2): Select and analyze the clustering algorithm of tumor single cell scATAC-seq data; Step 3): Obtain cluster gene region signals and tumor single cell scATAC-seq data clustering clusters; Step 4): Identify the non-coding regulatory units of cancer cell types within the tumor and cancer cell subtypes of cancer tissues; Step 5): Identify common enhancers of cancer cell subtypes of tumor tissues through the scEnhancer algorithm; Step 6): Using the ROSE algorithm, identify super enhancers among the enhancers of the tumor tissue cancer cell subtype identified in step 5).

2. A method for processing super enhancer data at the single cell level according to claim 1, characterized in that: In the step 2), the clustering tool for the tumor single-cell scATAC-seq data is clustered using Signac software.

3. A method for processing super enhancer data at the single cell level according to claim 1, characterized in that: In step 6), the super enhancer is identified in the tumor single cell scATAC-seq data using the ROSE algorithm, specifically using the following steps: Step S1): Preliminary peak calling: Use tumor single-cell scATAC-seq data and standard peak calling tools to identify all enhancer peaks in the genome, which represent potential regulatory regions; Step S2): Peak expansion: The ROSE algorithm is used to merge adjacent enhancer peaks whose distance is less than the threshold in order to capture large regulatory regions, and the parameters of the transcription start site exclusion zone are set by the user to exclude false positive signals near the transcription start site, so as to identify super enhancers composed of multiple adjacent enhancers; Step S3): Signal intensity calculation: Calculate the signal intensity of the tumor single-cell scATAC-seq data for the merged enhancer regions, and use the standardized reading signal to measure the intensity of each enhancer region; Step S4): Sorting: Sorting all the merged enhancers according to the signal intensity of each enhancer region in order to distinguish between common enhancers and super enhancers; Step S5): Determine the threshold: Draw a cumulative curve of the signal intensity of the enhancer region using the ROSE algorithm and find the inflection point of the cumulative curve, and define the enhancer after the inflection point as a super enhancer.

4. A method for processing super enhancer data at the single cell level according to claim 3, characterized in that: In the step S1), after the peak adjustment tool of the MASC2 software is used to identify the regulatory region of the enhancer peak, the ROSE algorithm is used to divide each genomic site identified by the MACS2 software into multiple bins, and the read density is mapped to each bin to improve the resolution of the signal intensity within each genomic site.

5. A method for processing super enhancer data at the single cell level according to claim 4, characterized in that: The size of each bin or the length of the genomic region covered by each bin is calculated using the following formula (1): (1)； in, is the length of the genomic locus, Indicates the number of bins, The length of the genomic region covered by each bin.

6. A method for processing super enhancer data at the single cell level according to claim 4, characterized in that: In each bin at the genomic locus, the nth bin at the genomic locus is B n , and 1≤n≤N, the signal value S of each bin Bn It is expressed as formula (2): (2)； Among them, i represents the starting genomic coordinate of each bin, d is the length of each bin, and R s,x Represents the signal value reading on the encoding chain at the x coordinate, R a,x Represents the signal value reading on the non-coding strand at the x-coordinate.

7. A method for processing super enhancer data at the single cell level according to claim 6, characterized in that: The signal value of each bin is normalized, and the read density of each bin is normalized by the total million mapped reads MMR using the ROSE algorithm, specifically formula (3): (3)； in, Normalization of the read density for the signal value for each bin.

8. A method for processing super enhancer data at the single cell level according to claim 3, characterized in that: In step S4), V is represented as a vector arranged in ascending order, which contains the read densities of all enhancers. V can be represented as: {v1,v2,v 3,…, v n }, where v represents the read density of each enhancer in the genome. For any enhancer with negative read density, its value will be set to zero. The elements in vector V are: 0≤v1≤v2≤v3≤ … ≤v n .

9. A method for processing super enhancer data at the single cell level according to claim 8, characterized in that: The slope of the straight line tangent to the read density curve of the enhancer is m, which is calculated by formula (4): (4)； Among them, V max and V min Respectively represent the maximum and minimum values ​​in the vector V arranged in ascending order, L V Represents the total length of vector V in ascending order.

10. A method for processing super enhancer data at the single cell level according to claim 9, characterized in that: When the straight line is tangent to the read density curve, the number of points below the straight line is minimized, and the coordinate point of the intersection of the straight line and the read density curve on the x-coordinate is X cf , the number of points below the straight line is minimized by the following formula (5): (5)； In formula (5), i represents the index of each enhancer in vector V, v i represents the read density of the i-th enhancer in the genome, b is the y-intercept of the line, and x i is the intersection of the ith straight line on the x coordinate and the reading density curve, b increases with x i The dynamic changes of are calculated as shown in formula (6): (6)； In formula (6), y i represents the read density of the enhancer at the ith position in the vector V, and m is the slope of the line tangent to the read density curve of the enhancer; The cutoff value Y between common enhancers and super enhancers cf It is expressed as formula (7): (7)； In formula (7), since the intersection of the straight line and the reading density curve is at the x-coordinate point X cf ,but The coordinate point value on the x-coordinate of the intersection of the straight line on the vector V arranged in ascending order and the read density curve is used as the boundary value between common enhancers and super enhancers.

Images