A method, device, and medium for target screening to predict the occurrence of complex human disease phenotypes.
Through genome-wide association studies and colocalization analysis, targets for complex human diseases were screened, solving the problem of the inability to accurately predict disease phenotypes in existing technologies, and enabling rapid and accurate target screening and efficient targeted intervention.
Patent Information
- Application Number
- CN202411251941.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-06
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2044-09-06
AI Technical Summary
Current technologies cannot accurately predict the occurrence of complex human disease phenotypes, making it difficult for clinicians to intervene in and prevent the occurrence and development of age-related diseases in a timely manner.
By acquiring genotype-phenotype association data, we conducted genome-wide association studies and quantitative trait locus mapping, screened GWAS and QTL data, grouped and matched the data using preset windows and p-values, performed co-localization analysis and GWAS-QTL correspondence calculations, and screened target sites.
This technology enables precise localization of the genes and molecular mechanisms of complex human diseases in a short period of time, improves the accuracy and efficiency of target screening, discovers the regulatory mechanisms of multiple diseases/phenotypes, and increases the success rate of targeted intervention therapy.
Smart Images

Figure CN119007803B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of target screening, and in particular to a method, device and medium for target screening to predict the occurrence of complex human disease phenotypes. Background Technology
[0002] The inherent randomness of biochemical reactions leads to the widespread distribution of gene expression in cell populations, even in seemingly homogeneous cell populations; this phenomenon, known as "gene expression noise," is widely observed in both prokaryotic and eukaryotic systems. Numerous studies have shown that noise is involved in a variety of important biological and pathological processes. For example, changes in transcriptional variability in pluripotent cells have been observed at different stages of embryonic development. Furthermore, transcriptional variability increases with age, and disruption of noise control can lead to abnormal immune responses and cell recognition disorders, inducing a range of age-related diseases.
[0003] It remains unclear which specific targets transcriptional noise regulates to influence the development of complex disease phenotypes in humans. Furthermore, aging leads to a gradual increase in the incidence of chronic diseases and the co-occurrence of multiple diseases. Most chronic diseases are degenerative changes caused by a combination of internal and external factors during aging, exhibiting tissue or cellular differences. Accurately predicting the occurrence of these complex disease phenotypes, enabling clinicians to intervene and prevent the development and progression of age-related diseases, is a pressing challenge that needs to be addressed. Summary of the Invention
[0004] The purpose of this application is to provide a method, device, and medium for screening targets to predict the occurrence of complex human disease phenotypes, thereby rapidly and accurately screening target targets and improving the performance, efficiency, and accuracy of co-localization screening.
[0005] To achieve the above objectives, this application provides the following solution:
[0006] In a first aspect, this application provides a target screening method for predicting the occurrence of complex human disease phenotypes, including:
[0007] Obtain genotype-phenotype association data for a target human complex disease; wherein the target human complex disease includes at least one human complex disease;
[0008] Genome-wide association studies and quantitative trait locus mapping were performed on the genotype-phenotype association data to obtain GWAS data and QTL data.
[0009] Based on a preset window, the GWAS data is filtered and grouped to obtain a first set of leadsnp; the first leadnp in each subset of the first set of leadsnp is the SNP with the smallest p-value in the GWAS data within the corresponding preset window range; the size of the preset window corresponding to the first leadnp in different subsets is different; the p-value represents the significance of the genotype-phenotype association.
[0010] Based on the preset window and the first leadsnp set, the QTL data is matched, filtered, and grouped to obtain a second leadsnp set; the second leadsnp in each subset of the second leadsnp set is the SNP in the GWAS data and the QTL data with the smallest p-value within the corresponding preset window range;
[0011] Based on the second leadsnp set, the GWAS data, and the QTL data, colocalization analysis and GWAS-QTL correspondence calculation are performed to determine the target p-value and target correspondence score.
[0012] Based on the target p-value and the target correspondence score, target leads are selected from the second set of leads to serve as target points.
[0013] In a second aspect, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a target screening method for predicting the occurrence of complex human disease phenotypes.
[0014] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a target screening method for predicting the occurrence of complex human disease phenotypes.
[0015] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application provides a method, device, and medium for screening targets to predict the occurrence of complex human disease phenotypes. Genome-phenotype association data are subjected to genome-wide association studies (GWAS) and quantitative trait locus mapping (QTL) data, respectively, to obtain GWAS data and QTL data. Then, based on a preset window and using p-values as a benchmark, the GWAS data is screened and grouped, and the QTL data is matched, screened, and grouped to obtain a second set of leadsnp. Each leadnp in this set is the SNP with the smallest p-value within the corresponding preset window range in both GWAS and QTL data. Using the leadnp in this set for co-localization analysis can yield good co-localization properties. Based on this, colocalization analysis and GWAS-QTL correspondence calculations are performed to determine the target p-value and target correspondence score. That is, colocalization analysis and the calculation of target p-values and target correspondence scores are performed on leads with good colocalization attributes, ensuring the accuracy of the results. It should be noted that since the leads in the second set of leads are obtained after preliminary screening, their number is significantly reduced compared to the initial GWAS and QTL data, thus improving data processing efficiency. Then, based on the target p-value and target correspondence score, target leads are selected from the second set of leads as target sites, thereby enabling the acquisition of more accurate targets that can predict the occurrence of complex human disease phenotypes in a very short time. Based on the target sites finally obtained in this application, a massive amount of genetic evidence can be mined in a short time, accurately locating the genes and molecular mechanisms of a complex human disease, thereby improving the quality of target screening. This application improves the performance, efficiency, and accuracy of colocalization screening. Attached Figure Description
[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0017] Figure 1 This is a diagram illustrating the application environment of a target screening method for predicting the occurrence of complex human disease phenotypes, as described in an example of this application.
[0018] Figure 2A flowchart illustrating a target screening method for predicting the occurrence of complex human disease phenotypes, provided as an example of this application;
[0019] Figure 3 A flowchart illustrating a target screening method for predicting the occurrence of complex human disease phenotypes, as provided in another example of this application;
[0020] Figure 4 This is a schematic diagram of the structure of a computer device provided as an example of this application. Detailed Implementation
[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0022] Locus-based target prediction and screening offers a new approach to addressing the pressing challenge of accurately predicting the occurrence of complex disease phenotypes, enabling clinicians to intervene and prevent the development of age-related diseases. Building upon previously established locus analysis systems, further improvements and optimizations allow for the rapid extraction of vast amounts of genetic evidence, precisely locating susceptibility genes and molecular mechanisms within loci in specific peripheral blood immune cells, thus providing high-quality target predictions. Simultaneously, the introduction of a "single-cell + parallel mechanism" perspective identifies numerous previously overlooked gene-disease / phenotype regulatory mechanisms. These new regulatory mechanisms reveal that a single gene can simultaneously cause multiple diseases / phenotypes, thereby uncovering novel intervention and therapeutic targets for age-related comorbidities. For example, the rs1801133 variant of the tetrahydrofolate reductase MTHFR gene, which is associated with diseases such as cancer, coronary heart disease, and dementia, has been reported to affect blood characteristics MCV (mean corpuscular volume), PCT (platelet crit), PLT (platelet count), and RDW-CV (red cell distribution width-coefficient of variation) in CD8 cells by regulating the KIAA2013 gene. Furthermore, the APOE gene rs7412 (ApoEε2), associated with lipid metabolism in many diseases (cardiovascular disease, hypertension, dementia, etc.), affects reticulocyte count (RET) and the erythrocyte-reticulocyte percentage (RET%) in CD8 cells by regulating the lncRNA (CTB-129P6.11) rather than genes related to lipid metabolism. Further, it is possible to predict key regulatory targets, improve the success rate of targeted intervention therapies, shorten the research and development cycle, and reduce costs.
[0023] This research is expected to revolutionize the traditional target discovery model and, by targeting the single-cell level, enable precise prediction and targeted intervention, thereby overcoming the global challenges of late detection, comorbidity, severe disease progression, and untimely treatment of complex human diseases. The traditional target discovery model is incremental: first, understanding the biological mechanisms of the target, and then, based on hypotheses, verifying the relationship between the target and disease step by step from "molecular-cellular-animal-human." This application is diametrically opposed to the locus-guided target prediction model: it does not rely on any pre-existing assumptions, but first conducts genome-wide association studies on real-world populations to discover statistical relationships between numerous loci and various diseases; then, combined with single-cell transcriptomics, it locates the truly pathogenic genes and corresponding mechanisms within specific cell types at specific loci. Therefore, the target recommendation system provided in this application, based on "single-cell + parallel mechanism," will offer a new research paradigm for the crucial link of precise prediction of complex human diseases.
[0024] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0025] This application provides a target screening method for predicting the occurrence of complex human disease phenotypes, which can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be set up independently, integrated into server 104, or placed in the cloud or on another server. Terminal 102 can send genotype-phenotype association data of a target complex human disease to server 104. After receiving the genotype-phenotype association data, server 104 performs a series of processes to obtain target points that can be used to predict the occurrence of complex human disease phenotypes. Server 104 can feed back the obtained target points to terminal 102. Furthermore, in some embodiments, the target point screening method for predicting the occurrence of complex human disease phenotypes can also be implemented independently by server 104 or terminal 102.
[0026] The terminal 102 can be, but is not limited to, various desktop computers, laptops, etc. The server 104 can be implemented using a standalone server or a server cluster composed of multiple servers, or it can be a cloud server.
[0027] like Figure 2 and Figure 3 As shown, a target screening method for predicting the occurrence of complex human disease phenotypes is provided. This method is executed by a computer device, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method is applied to... Figure 1Taking server 104 as an example, the explanation includes the following steps 201 to 206.
[0028] Step 201: Obtain genotype-phenotype association data for the target human complex disease; wherein, the target human complex disease includes at least one human complex disease, and can be set according to actual research needs. In practical applications, relevant data can be directly obtained from publicly available databases on the internet.
[0029] Step 202 involves performing genome-wide association studies (GWAS) and quantitative trait locus mapping (QTL) on the genotype-phenotype association data to obtain GWAS and QTL data. In practical applications, some publicly available online databases have already performed GWAS and QTL on the genotype-phenotype association data, providing p-values for each SNP; others do not provide p-values for each SNP, requiring technical personnel to use appropriate software or programs for calculation. This is common knowledge in the field and will not be elaborated upon here. Similarly, the r-value between two adjacent loci in a genome-wide association study can be obtained. 2 value.
[0030] Step 203: Based on a preset window, the GWAS data is filtered and grouped to obtain a first set of leadsnp; the first leadnp in each subset of the first set of leadsnp is the SNP with the smallest p-value in the GWAS data within the corresponding preset window range; the size of the preset window corresponding to the first leadnp in different subsets is different; the p-value represents the significance of the genotype-phenotype association.
[0031] In one application example, step 203 includes:
[0032] 31) Based on a first preset p-value threshold, the GWAS data is filtered; in one example, the first preset p-value threshold is set to 5e-08 (i.e., 5 × 10⁻⁸). -8 If p < 5e-08, then the data in the GWAS data will be filtered out.
[0033] 32) Based on multiple preset r 2 A threshold value is used to group the filtered GWAS data to obtain multiple SNP groups to be processed; where r 2 The value represents the degree of association between alleles at different loci; in one instance, a preset value r can be set. 2 If the threshold is 0.2, then based on r 2 The criterion of >0.2 removes linkage imbalances from the filtered GWAS data, and then groups them.
[0034] 33) For any of the SNP groups to be processed, determine the SNP with the smallest p-value from the SNP group and use it as the benchmark SNP. In an example algorithm, this benchmark SNP can be defined as the lead SNP. eLD .
[0035] 34) Obtain preset windows of different sizes; in one example, the size of the preset window is 10K, 30K, 50K, 100K, etc., and the size and number of preset windows can be adjusted by relevant technical personnel.
[0036] 35) For any size preset window, starting from the position of the reference SNP, expand in two directions respectively. When the reference SNP is the SNP with the smallest p value within the range of the preset window, mark the reference SNP as the first lead SNP corresponding to the preset window.
[0037] 36) Based on the size of the preset window, all the first leadSNPs are grouped to obtain multiple first leadSNP subsets; each first leadSNP subset corresponds to a preset window of a certain size; multiple first leadSNP subsets constitute a first leadSNP set. In other words, in this application, a certain window is defined as needed for leadSNPs. eLD We need to consider whether a value is presented as a lead within a certain window, and divide it into different groups accordingly. For example, the set of leadSNPs that present a lead within 100K is denoted as the first subset of leadSNPs corresponding to 100K, which can be represented as leadSNPs for easier mathematical derivation later. gwas 100K .
[0038] To facilitate understanding, the following example is provided for illustrative purposes only: Assume there are 100 SNPs in the GWAS data, transmitted via r 2 After grouping the values, ten SNP groups are obtained, and each SNP group contains one lead SNP. eLD If the current window size is 100K, then use leadSNP. eLD Starting with 50K in each of the two directions, then determining if the leadSNP is within a combined range of 100K. eLD Is the p-value still the smallest? If so, then change the leadsnp value. eLDThe first lead SNP is marked as the one corresponding to 100K. Similarly, the same checks can be performed for windows of 200K, 300K, etc. After all ten groups of SNPs to be processed have completed the checks for whether they are considered leads under windows of 100K, 200K, and 300K, a subset is formed from all the SNPs that are considered leads under window 100K. Similarly, subsets corresponding to windows of 200K and 300K are obtained. Finally, the first set of lead SNPs is formed from multiple subsets.
[0039] More specifically, each of the ten pending SNP groups corresponds to one of the ten lead SNPs. eLD If we label them as SNP1, SNP2, ..., SNP9, and SNP10, and the p-value of SNP4 is less than the p-value of SNP2; and further assume that within a 100K window, SNP2, SNP4, and SNP8 still represent the leading SNPs, then the first leading SNP set within the 100K window includes SNP2, SNP4, and SNP8. The reason SNP4's p-value is less than SNP2's p-value, yet SNP2 and SNP4 are still retained as the first leading SNPs, is because the position difference between SNP2 and SNP4 might be 200K, but the current window is 100K. SNP4 is not within the window starting with SNP2, so SNP2 has the smallest p-value within the 100K window and is therefore labeled as the first leading SNP.
[0040] Step 203 above can filter out the leading SNPs in the GWAS data. The leading SNP within a certain window is the SNP with the smallest p-value among all SNPs within that window, which is the most significant SNP.
[0041] Step 204: Based on the preset window and the first leadsnp set, the QTL data is matched and filtered to obtain a second leadsnp set; the second leadsnp in each subset of the second leadsnp set is the SNP in the GWAS data and the QTL data with the smallest p value within the corresponding preset window range.
[0042] In one application example, step 204 includes:
[0043] 41) Based on a second preset p-value threshold, the QTL data is filtered; in one example, the second preset p-value threshold is set to 0.05, then data with a p-value < 0.05 in the QTL data are filtered out. The second preset p-value threshold is not limited to 0.05, and relevant technicians can adjust it as needed.
[0044] 42) Based on the first set of leadSNPs, the filtered QTL data is matched to obtain significant SNP sites corresponding to different gene types in the QTL data. Specifically, the leadSNPs in the GWAS data obtained in step 203 are matched with the SNPs in the QTL data obtained in step 41). During this process, the gene types appearing in the QTL data are denoted as gene. qtl The i-th gene is denoted as gene. i The results were then categorized according to different gene types, and denoted as follows:
[0045] 43) For each gene type, starting from the significant SNP site corresponding to the QTL data, expand in two directions respectively. When the significant SNP site is the SNP with the smallest p value within a preset window of any size, mark the significant SNP site as the second leadSNP corresponding to the preset window; multiple second leadSNPs with the same preset window size constitute a second leadSNP subset; multiple second leadSNP subsets constitute a second leadSNP set.
[0046] The size of the preset window used in this step corresponds to the size of the preset window used in step 203. That is, in this step, based on the preset window, the size of the window obtained via step 42) is... Consider whether a given window of data will display as a lead, and group the data into different categories accordingly. For example, let's denot the set of leadSNPs that display as leads within a 100K window. The set obtained in this step is different from the set obtained in step 35). In the former, the leadsnp are most significant in both the GWAS and QTL data, while in the latter, the leadsnp are only significant in the GWAS data.
[0047] For simplicity, the second leadsnp set is recorded as φ n φ n This is a set of n leadSNPs. In another application example, the snps corresponding to the leadSNPs in the QTL data of each window are sorted by the p value, and this sorting is recorded as sortφ. n MK Where M is the value of window. For example, when window is 100K, the sorting is sortφ. n 100KIt is important to note that the order set in this step forms the basis for the colocalization analysis sequence in subsequent steps. By setting the order in advance, the stable and rapid progress of subsequent colocalization analysis can be ensured.
[0048] Since the lead snp in QTL data and the lead snp in GWAS data are not necessarily the same site, that is, a site that is most significant in GWAS data may not be the most significant in QTL data, step 204 above is set up so that after processing the GWAS data, it is also necessary to determine whether it is also a lead snp (i.e., the most significant) in QTL data.
[0049] Step 205: Based on the second leadsnp set, the GWAS data, and the QTL data, perform colocalization analysis and GWAS-QTL correspondence calculation to determine the target p-value and target correspondence score.
[0050] In one application example, step 205 includes:
[0051] 51) Extract a test lead NP set from the second lead NP set; the test lead NP set includes multiple test lead NP subsets, and each test lead NP subset includes multiple lead NPs. For example, in one instance, 1% is randomly selected from the second lead NP set as the test set, divided into three subsets, each containing one thousand lead NPs. Co-localization analysis is performed on all three lead NP subsets, and the results of the co-localization analysis are sorted according to the p-value and different correspondence scores are selected for measurement. The specific steps are as follows.
[0052] 52) Perform colocalization analysis on each of the test leadsnp subsets to obtain colocalization analysis results; the colocalization analysis results include the test p value corresponding to each leadsnp in the test leadsnp subset.
[0053] 53) Calculate the target p-value based on the colocalization analysis results corresponding to multiple test leadnp subsets.
[0054] In a specific practical application, step 53) includes:
[0055] 531) For any of the test leadnp subsets, based on the colocalization analysis results, multiple leadnp sites with good colocalization attributes are determined from all leadnp sites in the test leadnp subset according to the third preset p-value threshold; that is, leadnp sites with p-values less than the third preset p-value threshold are considered as leadnp sites with good colocalization attributes. Corresponding to the three test leadnp subsets mentioned above, the number of leadnp sites with good colocalization attributes obtained are 20, 22, and 21, respectively.
[0056] 532) Determine the site with the largest test p value from multiple leadsnp sites with good colocalization attributes; specifically, since the smaller the p value, the better the colocalization attribute, select the site with the largest p value from multiple leadsnp sites with good colocalization attributes in the three subsets respectively, and mark the corresponding p value as P1, P2, P3.
[0057] 533) Based on the sites with the largest test p-values corresponding to multiple subsets of the test leadsnp, the mean of the test p-values is calculated to obtain the target p-value. The calculation formula is as follows:
[0058]
[0059] The target p-value is an effective threshold boundary; only those with a p-value greater than the target p-value have a small probability of having good co-localization properties.
[0060] 54) For each subset of test leadsnp, calculate the GWAS-QTL correspondence of each leadnp by combining the GWAS data and the QTL data.
[0061] In a specific practical application, step 54) includes:
[0062] 541) For any leadSNP in the test leadSNP subset, the GWAS data and the QTL data are cut according to the corresponding preset window to obtain the snps group of the corresponding leadSNP; the snps group includes the site corresponding to each snp, the p value in the GWAS data and the p value in the QTL data.
[0063] More specifically, based on any leadsnp (which can be represented as leadsnp) in the aforementioned subset of test leadsnp iFor each i ∈ n, i is a specific i-th lead NP in the set of lead NPs. The GWAS data and the QTL data are then segmented according to the corresponding window based on pos (the position of the lead NP), thereby obtaining a set of all snps of the corresponding lead NP within a certain window, i.e., an snps group, which can be expressed by the following formula:
[0064]
[0065] Among them, the number of SNPs in the snps group is Nsnps i,j ; any snp ∈ snps in the snps group i,j , can be denoted as x i,j,k , Consider a collection of windows For specific The corresponding leadSNP is denoted as This represents the j-th window in the window set.
[0066] 542) For any SNP in the SNP group, normalize the p-value in the GWAS data and the p-value in the QTL data corresponding to the SNP, and then calculate the GWAS-QTL correspondence of the leadSNP using the following formula:
[0067]
[0068] Where, θ i,j The GWAS-QTL correspondence of the leadsnp is given by θ. 1,1 This represents the GWAS-QTL correspondence of the first leadsnp in the set when the window is 100K. and These are the p-values in the GWAS data and QTL data (taken as -log10) corresponding to the normalized SNP; For indicator functions, For a set of weighting factors: [] represents the Gaussian integer function; Nsnps i,j The number of SNPs in the snps group; and These are the p-values in the GWAS data and QTL data corresponding to SNP (taken as -log10), respectively. and These are the loci in the GWAS data and the loci in the QTL data corresponding to SNP, respectively.
[0069] The GWAS-QTL correspondence score is a value greater than 0. The higher the value, the better the co-localization attribute of the leadSNP. In subsequent comparisons, it needs to be used in conjunction with p-value ranking to obtain the highest accuracy; both are indispensable. That is to say, this application sets two thresholds: one is the p-value threshold (i.e., the target p-value), and the other is the correspondence score threshold (i.e., the target correspondence score). Only leadSNPs that meet the requirements will be retained, and others will be discarded.
[0070] 55) Determine the target correspondence score based on the GWAS-QTL correspondence and colocalization analysis results of each of the aforementioned leadsnp subsets.
[0071] In a specific practical application, step 55) includes:
[0072] 551) For any of the aforementioned test leadnp subsets, obtain multiple preset correspondence scores at equal intervals. For example, if the equal interval is set to 5, then set multiple preset correspondence scores to be 30, 35, and 40 respectively.
[0073] 552) For any of the preset correspondence scores, based on the GWAS-QTL correspondence of each of the leadsnp in the test leadnp subset, select multiple leadnp sites from the test leadnp subset whose GWAS-QTL correspondence is less than the preset correspondence score.
[0074] Taking a preset correspondence score of 30 as an example, multiple leadSNPs with a GWAS-QTL correspondence score of less than 30 can be obtained through step 552).
[0075] 553) Based on the colocalization analysis results, multiple leadsnp sites with GWAS-QTL correspondence scores less than the preset correspondence scores are screened again, and leadsnp sites with good colocalization attributes are retained, and the proportion of retained sites is calculated.
[0076] For the multiple leadSNPs with a GWAS-QTL correspondence of less than 30 obtained in the previous step, based on the colocalization analysis results for each leadSNP, when its p-value is less than the third preset p-value threshold, it is characterized as a leadSNP site with good colocalization attributes. All leadSNP sites with good colocalization attributes are retained, and their proportion among all multiple leadSNPs with a GWAS-QTL correspondence of less than 30 is calculated. For example, if the number of leadSNPs with a GWAS-QTL correspondence of less than 30 obtained in the previous step is 30, and 15 of them have good colocalization attributes, then the corresponding retention rate is 50%.
[0077] Similarly, the proportion of retained sites when the preset correspondence scores are 35 and 40 can be obtained.
[0078] 554) Based on the proportion of retained sites corresponding to the multiple preset correspondence scores, select the preset correspondence score corresponding to the largest proportion of retained sites and mark it as the correspondence score to be used; that is, select the largest proportion of retained sites from the proportions of retained sites corresponding to the preset correspondence scores of 30, 35 and 40 respectively, and use the preset correspondence score corresponding to this proportion, such as 35, as the correspondence score to be used for subsequent processing.
[0079] 555) The mean of the candidate correspondence scores corresponding to multiple test leadnp subsets is calculated to obtain the target correspondence score.
[0080] In step 205 above, the maximum p-value and minimum correspondence score of the key are obtained in advance based on a test set randomly selected from the dataset. Then, these two parameters are used to filter the total leadsnp dataset, i.e., step 206 below.
[0081] Step 206: Based on the target p-value and the target correspondence score, select target leads NPs from the second lead NP set as target points. Specifically, when selecting target leads NPs from the second lead NP set based on the target p-value and the target correspondence score, those that satisfy... The conditional lead NP is used as the target lead NP; where maxP is the target p-value and minscore is the target correspondence score. Let p be the p value of the leadSNP in the second set of leadSNPs. This represents the GWAS-QTL correspondence of the leads in the second set of leadsnp.
[0082] The colocalization analysis mentioned in the text is an analytical method used when studying the relationship between a specific SNP and a trait. The aim is to determine whether the SNP influences the trait by regulating the expression of a particular gene. Specifically, there are four hypotheses for colocalization analysis:
[0083] H0: Phenotype 1 (GWAS) and phenotype 2 (QTL, eQTL) are not significantly associated with any of the snp sites in a genomic region.
[0084] H1 / H2: Phenotype 1 (GWAS) or phenotype 2 (QTL, eQTL) is significantly associated with snp sites in a genomic region.
[0085] H3: Phenotype 1 (GWAS) and phenotype 2 (QTL, eQTL) are significantly associated with snp sites in a genomic region, but are driven by different causal variant sites.
[0086] H4: Phenotype 1 (GWAS) and phenotype 2 (QTL, eQTL) are significantly associated with snp sites in a genomic region and are driven by the same causal variant site.
[0087] By using Bayesian inference-based co-localization methods, considering GWAS and QTL data, and observing the obtained PP.H4.abf value, a value greater than 0.7 is considered to indicate that the locus has "co-localized," meaning it possesses good co-localization properties. Statistically speaking, this value represents the probability that this SNP will influence traits by regulating gene expression levels (eQTL as an example).
[0088] Compared to existing technologies, the target sites selected in this application exhibit better co-localization screening results, improving the performance, efficiency, and accuracy of co-localization screening. In a specific test, the data processing steps 201-204 of this application are sufficient to support rapid result acquisition, resulting in a significant performance improvement. Building upon this, the parameter analysis and data screening in steps 205-206 further enhance performance and accuracy. From an accuracy perspective, this application, combining p-value ranking and correspondence, selected 200 lead NPs, of which 112 possessed good co-localization attributes. Compared to existing technologies, the accuracy was improved from 1.47% to 56%. From a performance perspective, the test leadSNP set of this application includes 193 leadSNP sites with good co-localization properties among 13,599 leadSNPs. After filtering using maxP as the threshold, 3,000 leadSNPs with smaller p-values were obtained. These leadSNPs include 191 of the aforementioned 193 leadSNP sites with good co-localization properties. From this perspective, the method of this application can eliminate nearly 80% of the leadSNPs, as these leadSNPs cannot obtain good co-localization results. Through this method, a significant performance improvement can be achieved.
[0089] Finally, to reiterate, the reason this application achieves excellent performance and a significant improvement in accuracy is due to the following improvement steps:
[0090] Steps 201-204: A novel data processing architecture is proposed. This architecture can rapidly screen lead SNPs and use them for colocalization analysis, exhibiting good colocalization properties. In a validation experiment, the method described in this application was used to analyze 29 blood phenotypic features, 5 autoimmune disease phenotypes, and 8 cell types. The GWAS data contained a total of 1,281,221,061 SNP sites, and the evQTL data contained a total of 345,972,761 SNP sites. The novel architecture provided in this application can perform colocalization analysis on all data within 48 hours.
[0091] Steps 205-206: A novel set of judgment parameters was proposed: the target p-value and the target correspondence score. Combining these two parameters to filter lead SNPs significantly improved the accuracy and performance of colocalization. Building upon steps 201-204, the performance of removing lead SNPs, segmenting the data, and performing colocalization was greatly improved. The time taken for segmenting the data and performing colocalization was reduced from 24 hours to 6 hours, and the overall runtime was further optimized to perform colocalization analysis on valuable data within 30 hours. Furthermore, in the test dataset, by obtaining the correspondence score and p-value ranking of key parameters, 112 out of the selected 200 sites possessed good colocalization attributes. The accuracy improved from approximately 1.47% (193 sites with good colocalization attributes out of 13599 sites) to 56%, achieving a significant improvement in accuracy and a substantial optimization of overall performance.
[0092] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 4 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and databases. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a target screening method for predicting the occurrence of complex human disease phenotypes.
[0093] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.
[0094] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.
[0095] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).
[0096] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.
[0097] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0098] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.
Claims
1. A method of screening for targets that predict the onset of a complex disease phenotype in humans, characterized in that, The target screening method for predicting the occurrence of complex human disease phenotypes includes: Obtain genotype-phenotype association data for a target human complex disease; wherein the target human complex disease includes at least one human complex disease; Genome-wide association studies and quantitative trait locus mapping were performed on the genotype-phenotype association data to obtain GWAS data and QTL data. Based on a preset window, the GWAS data is filtered and grouped to obtain a first set of leadsnp; the first leadnp in each subset of the first set of leadsnp is the SNP with the smallest p-value in the GWAS data within the corresponding preset window range; the size of the preset window corresponding to the first leadnp in different subsets is different; the p-value represents the significance of the genotype-phenotype association. Based on the preset window and the first leadsnp set, the QTL data is matched, filtered, and grouped to obtain a second leadsnp set; the second leadsnp in each subset of the second leadsnp set is the SNP in the GWAS data and the QTL data with the smallest p-value within the corresponding preset window range; Based on the second leadsnp set, the GWAS data, and the QTL data, colocalization analysis and GWAS-QTL correspondence calculation are performed to determine the target p-value and target correspondence score. Based on the target p-value and the target correspondence score, target leads are selected from the second set of leads to serve as target points.
2. The target screening method for predicting the occurrence of complex human disease phenotypes according to claim 1, characterized in that, Based on a preset window, the GWAS data is filtered and grouped to obtain a first set of leadSNPs, including: The GWAS data is filtered based on a first preset p-value threshold. based on a plurality of preset r 2 values, the filtered GWAS data is grouped to obtain a plurality of to-be-processed snp groups; wherein, r 2 values represent the correlation degree of alleles between sites. For any of the SNP groups to be processed, the SNP with the smallest p-value is determined from the SNP groups to be processed and used as the benchmark SNP; Get preset windows of different sizes; For any size preset window, starting from the position of the reference SNP, expand in two directions respectively. When the reference SNP is the SNP with the smallest p value within the range of the preset window, mark the reference SNP as the first lead SNP corresponding to the preset window. Based on the size of the preset window, all the first leadSNPs are grouped to obtain multiple first leadSNP subsets; each first leadSNP subset corresponds to a preset window of a certain size; multiple first leadSNP subsets constitute a first leadSNP set.
3. The target screening method for predicting the occurrence of complex human disease phenotypes according to claim 1, characterized in that, Based on the preset window and the first leadSNP set, the QTL data is matched, filtered, and grouped to obtain a second leadSNP set, including: The QTL data is filtered based on a second preset p-value threshold. Based on the first leadSNP set, the QTL data after screening is matched to obtain significant SNP sites corresponding to different gene types in the QTL data; For each gene type, starting from the significant SNP site corresponding to the QTL data, expansion is carried out in two directions. When the significant SNP site is the SNP with the smallest p-value within a preset window of any size, the significant SNP site is marked as the second leadSNP corresponding to the preset window. Multiple second leadSNPs with the same preset window size constitute a second leadSNP subset. Multiple second leadSNP subsets constitute a second leadSNP set.
4. The target screening method for predicting the occurrence of complex human disease phenotypes according to claim 1, characterized in that, Based on the second leadSNP set, the GWAS data, and the QTL data, co-localization analysis and GWAS-QTL correspondence calculation are performed to determine the target p-value and target correspondence score, including: Extract a test leadnp set from the second leadnp set; the test leadnp set includes multiple test leadnp subsets, and each test leadnp subset includes multiple leadnp sets. For each subset of test leadsnp, colocalization analysis is performed to obtain colocalization analysis results; the colocalization analysis results include the test p value corresponding to each leadnp in the subset of test leadsnp; Calculate the target p-value based on the colocalization analysis results corresponding to multiple test leadSNP subsets; For each subset of the test leadsnp, the GWAS-QTL correspondence of each leadnp is calculated by combining the GWAS data and the QTL data; The target correspondence score is determined based on the GWAS-QTL correspondence and colocalization analysis results of each of the multiple test leadnp subsets.
5. The target screening method for predicting the occurrence of complex human disease phenotypes according to claim 4, characterized in that, For each subset of test leadsnp, combining the GWAS data and the QTL data, calculate the GWAS-QTL correspondence of each leadnp, including: For any leadSNP in the test leadSNP subset, the GWAS data and the QTL data are cut according to the corresponding preset window to obtain the snps group of the corresponding leadSNP; the snps group includes the site corresponding to each SNP, the p value in the GWAS data and the p value in the QTL data; For any SNP in the SNP group, the p-values in the GWAS data and QTL data corresponding to the SNP are normalized respectively, and then the GWAS-QTL correspondence of the leadSNP is calculated using the following formula: Where, θ i,j The GWAS-QTL correspondence degree of the leadsnp; and These are the p-values in the GWAS data and QTL data corresponding to the normalized SNP, respectively; Let λ be the indicator function. r2 For a set of weighting factors: [] represents the Gaussian integer function; Nsnps i,j The number of SNPs in the snps group; and These are the p-values in the GWAS data and QTL data corresponding to SNP, respectively; and These are the loci in the GWAS data and the loci in the QTL data corresponding to SNP, respectively.
6. The target screening method for predicting the occurrence of complex human disease phenotypes according to claim 4, characterized in that, Based on the GWAS-QTL correspondence and colocalization analysis results of each of the aforementioned leadSNP subsets, the target correspondence score is determined, including: For any of the aforementioned test leadSNP subsets, obtain multiple preset correspondence scores at equal intervals; For any of the preset correspondence scores, based on the GWAS-QTL correspondence of each leadnp in the test leadnp subset, multiple leadnp sites with GWAS-QTL correspondence scores less than the preset correspondence scores are selected from the test leadnp subset. Based on the colocalization analysis results, multiple leadsnp sites with GWAS-QTL correspondence scores lower than the preset correspondence scores are further screened, and leadsnp sites with good colocalization attributes are retained. The proportion of retained sites is calculated. Based on the proportion of retained sites corresponding to multiple preset correspondence scores, select the preset correspondence score corresponding to the largest proportion of retained sites and mark it as the correspondence score to be used. The target correspondence score is obtained by averaging the candidate correspondence scores corresponding to multiple subsets of the test leadsnp.
7. The target screening method for predicting the occurrence of complex human disease phenotypes according to claim 4, characterized in that, Based on the colocalization analysis results corresponding to multiple test leadSNP subsets, the target p-value is calculated, including: For any colocalization analysis result corresponding to any of the test leadsnp subsets, based on the third preset p-value threshold, multiple leadsnp sites with good colocalization attributes are determined from all leadsnp in the test leadsnp subsets. From a plurality of leadsnp sites with good colocalization attributes, determine the site with the largest test p-value; The mean of the test p-values is calculated based on the sites with the largest test p-values corresponding to multiple subsets of the test leadsnp to obtain the target p-value.
8. The target screening method for predicting the occurrence of complex human disease phenotypes according to claim 1, characterized in that, Based on the target p-value and the target correspondence score, when selecting target leads from the second leadSNP set, those that satisfy... The conditional leader NP is used as the target leader NP; Where maxP is the target p-value, and minscore is the target correspondence score. Let p be the p value of the leadSNP in the second set of leadSNPs. This represents the GWAS-QTL correspondence score of the leads in the second set of leadsnp.
9. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the target screening method for predicting the occurrence of complex human disease phenotypes as described in any one of claims 1-8.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the target screening method for predicting the occurrence of complex human disease phenotypes as described in any one of claims 1-8.
Citation Information
Patent Citations
Method for detecting SNP combination on GWAS data based on gene pool
CN118522345A
Methods and Compositions for Treating, Ameliorating, and / or Preventing Chronic Kidney Disease (CKD) and Complications thereof by Regulating DPEP1, CASP9, ACSS2 and / or FASN
US20240229043A1
