Identification, construction and application of tandem repeat sequences of plant centromere
By employing multi-feature fusion scoring and phase correction methods, the problem of accurate identification and quantification of plant centromere tandem repeat sequences was solved, enabling high-confidence candidate sequence screening and consensus sequence construction, supporting plant chromosome research and functional genomics analysis.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NANTONG UNIV
- Filing Date
- 2026-03-30
- Publication Date
- 2026-06-09
AI Technical Summary
Existing technologies struggle to accurately identify and quantify plant centromere tandem repeat sequences, resulting in high false positive rates, lack of interpretability, and instability in consensus sequence construction.
A multi-feature fusion scoring system combined with clustering and phase correction methods was adopted. High-confidence candidate monosomal sequences were screened by scoring periodic consistency, genome enrichment, sequence complexity, and GC content shift. High-quality consensus sequences were constructed by using a phase correction strategy that combines exhaustive cyclic shift and reverse complementation.
It achieves high-confidence candidate sequence screening and consensus sequence construction, improves the interpretability and stability of results, and can accurately distinguish between centromere and noncentromere repeats, supporting plant chromosome research and functional genomics analysis.
Smart Images
Figure CN122177228A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of bioinformatics technology, and in particular relates to the identification, construction and application of plant centromere tandem repeat sequences. Background Technology
[0002] Centromeres are key structures in eukaryotic chromosomes, responsible for segregation and genetic stability. Plant centromere regions are typically rich in tandem repeat satellite DNA, which, along with retrotransposons and other sequences, constitutes a complex repetitive landscape. Accurate identification and quantitative evaluation of centromere tandem repeat units are fundamental to chromosome structure analysis, epigenetic regulation research, and molecular identification of species and varieties.
[0003] In bioinformatics practice, the identification of centromere tandem repeat sequences mainly faces the following technical challenges:
[0004] (1) Existing methods usually start with repeat detection tools (such as Tandem Repeats Finder), which identify repeat sequences and provide candidate intervals, but it is difficult to uniformly rank a large number of candidate monomers with confidence. Some screening strategies introduce features such as periodic consistency and GC content for coarse screening. However, due to the limited feature dimensions, it is easy to misclassify tandem repeats (such as ribosomal DNA) that are not related to the centromere but have high periodicity as candidate sequences, resulting in insufficient false positive control.
[0005] (2) Existing centromere prediction tools (such as CentIER, quarTeT, etc.) usually output centromere region coordinates or candidate region rankings, which cannot directly quantify the probability that a candidate sequence belongs to a centromere, nor can they provide sub-item scoring criteria for users to trace the source and adjust the threshold. In recent years, classification methods represented by deep learning have made progress in discrimination accuracy, but their decision-making process lacks interpretability and cannot provide clear sub-item scoring criteria, thus failing to meet the need for traceability of results in scientific research scenarios.
[0006] (3) In the absence of a specific motif, multiple sequence alignment methods (such as MAFFT) are often used to construct consensus sequences and assist in discrimination. However, when dealing with highly similar cyclic repeating sequences with phase drift, traditional multiple sequence alignment tools struggle to correctly handle phase misalignment between sequences, easily introducing systematic biases that affect the accuracy and stability of the consensus sequence. Although alignment algorithms specifically designed for long tandem repeating sequences have been proposed recently, they have not yet achieved phase correction at the individual level and autonomous construction of consensus sequences.
[0007] Therefore, there is an urgent need to develop a new method for the identification and scoring of plant centromere tandem repeat sequences, to achieve reliable sequencing and efficient screening of candidate sequences under denovo conditions, and to provide an interpretable and attributable quantitative evaluation system without relying on species-specific motifs and external alignment tools, while also enabling the autonomous construction of high-fidelity consensus sequences. Summary of the Invention
[0008] This invention proposes a method, construction, and application for the identification of tandem repeat sequences in plant centromeres. The method first obtains candidate monomers through efficient tandem repeat detection and noise filtering. Then, it quantifies the centromere assignment probability using a multi-feature fusion scoring system, and further employs clustering and phase correction to obtain high-quality consensus sequences. This improves the robustness and interpretability of centromere candidate identification in complex repeat contexts. The technology proposed in this invention can provide new analytical tools for research on plant centromeres and chromosomes, helping to reveal the distribution and structural characteristics of tandem repeats in the genome, and providing a theoretical foundation and technical support for plant functional genomics and chromosome engineering.
[0009] This invention provides a method for identifying tandem repeat sequences of plant centromeres, comprising the following steps:
[0010] (1) Obtain the sequence information of the target plant genome, perform tandem repeat detection on the sequence information, and obtain candidate repeat regions and their corresponding cycle lengths P;
[0011] (2) Extract repeat units as candidate monomer sequences based on the genomic coordinates of the candidate repeat regions, and remove candidate monomers whose base composition complexity is lower than a preset threshold.
[0012] (3) Calculate a multi-feature fusion score for each candidate monomer, wherein the fusion score includes at least two of the following: cycle consistency score, genome enrichment score, sequence complexity score, and GC content offset score, and screen candidate monomer sequences based on the fusion score;
[0013] (4) Cluster the candidate monomer sequences after screening in step (3) to obtain at least one monomer subtype;
[0014] (5) Perform phase correction on the monomer sequence within each monomer subtype and construct a consensus sequence based on the corrected sequence.
[0015] Preferably, in step (3), the fusion score is calculated using the following formula:
[0016]
[0017] in, Candidate monomeric sequences, For sequence Length, The period length; , , , Candidate monomer sequences The periodic consistency score, genome enrichment score, sequence complexity score, and GC content offset score are given, with w1, w2, w3, and w4 being the weight coefficients of each score, and w1 + w2 + w3 + w4 = 1.
[0018] Preferably, the weighting coefficients w1, w2, w3, and w4 are 0.35, 0.30, 0.20, and 0.15, respectively.
[0019] Preferably, in step (1), the concatenation repetition detection employs a prefix sum vectorized O(n) algorithm, including:
[0020] Encode the DNA sequence as an integer array; construct a matching array such that the first... Position and First Record 1 if the positions are equal, otherwise record 0; calculate the prefix sum of the matching array to obtain the array Pfx; calculate the sliding start position according to the following formula. Similarity at location :
[0021]
[0022] in, The starting position index of the sliding window. The candidate period length, and For prefix sum vectors at position and The element value at that location.
[0023] Preferably, the periodic consistency score Calculate using the following formula:
[0024]
[0025] in, Candidate monomeric sequences Length, The period length, This is the position index for alignment after the sequence has been shifted. For indicator functions, when (Right now Upper Position and First When the bases are the same, the value is 1; otherwise, the value is 0.
[0026] The genome enrichment score Calculate according to the following three formulas in sequence:
[0027]
[0028]
[0029]
[0030] in, Candidate monomeric sequences, For sequence Length, For k-mer length, Index of the window's starting position. For sequence From the first consecutive starting bits The k-mer corresponding to each base. and They are respectively Frequency of occurrence in the whole genome background and candidate haplogroup To avoid a preset positive small constant at zero frequency, For the first The log-odds ratio of sliding windows For all The arithmetic mean, and These represent taking the maximum value and taking the minimum value, respectively.
[0031] The sequence complexity score Calculate using the following formula:
[0032]
[0033] in, Candidate monomeric sequences, For traversing the set The base types in base In sequence The frequency in;
[0034] The GC content offset score Calculate using the following formula:
[0035]
[0036] in, For sequence GC base ratio, To preset the target GC content, For the GC deviation penalty width parameter, To obtain the maximum value.
[0037] Preferably, in step (5), the phase correction includes: adjusting the period of... Each single enumeration positive chain Cyclic shift and reverse complementary chain Cyclic shift Candidates are identified; the candidate with the highest positional consistency rate is selected as the optimal phase for that entity, based on the current consensus sequence; at least three rounds of iteration guided by consensus updates are performed; the consensus sequence is enumerated. Cyclic shift, in which Given the period length, the offset with the highest average positional consistency rate with all original units is selected as the globally optimal phase; finally, the consensus sequence is reconstructed by majority vote in each column.
[0038] Preferably, before constructing the consensus sequence in step (5), an optimal period selection step is also included:
[0039] For each candidate period in the candidate period set Calculate the consistency rate of each column separately. Average internal consistency between monomers And select the optimal period according to the following formula:
[0040]
[0041] in, For period The corresponding column consistency rate For period The corresponding average internal consistency between individual units; take The maximum value corresponding to This is the optimal period.
[0042] Preferably, in step (4), the clustering adopts Jaccard distance hierarchical clustering based on k-mer, including:
[0043] Candidate monomer sequences are encoded as k-mer existence vectors, and the Jaccard distance between each pair is calculated. Hierarchical clustering is performed based on the Jaccard distance, and monomer subtypes are divided according to a preset similarity threshold.
[0044] The present invention also provides a set of high-confidence plant centromere candidate monomers obtained by screening using any of the methods described above, or a consensus sequence of plant centromere tandem repeat sequences constructed therefrom.
[0045] This invention also provides an application of the method described in any of the above-mentioned methods in centromere region annotation of plant genomes, centromere function research, plant variety identification, chromosome fluorescence in situ hybridization probe design, or comparative genomics analysis.
[0046] One technical solution provided in this application embodiment has at least the following technical effects:
[0047] 1. This invention proposes the CenScore scoring system, which maps four dimensions—cycle consistency, genome enrichment, sequence complexity, and GC shift—to the same probability scale and fuses them according to their weights. Each sub-score can be output independently, facilitating result traceability and threshold setting. Compared to binary judgment methods relying solely on tandem repeat detection or single GC / cycle rules, this invention's multi-dimensional quantitative scoring strategy effectively distinguishes between centromere tandem repeats and highly periodic non-centromere repeats (such as rDNA and microsatellites). Taking the Arabidopsis Col-CEN genome as an example, there are 80,194 high-confidence monomers (CenScore ≥ 0.45), with an average GC content of approximately 0.38, consistent with typical plant centromere satellite sequence characteristics, validating the accuracy and low false-positive characteristics of the scoring system.
[0048] 2. This invention employs a phase search strategy combining exhaustive cyclic shifts and reverse complementation, along with iterative consensus guidance and global self-consistent alignment. This allows for the acquisition of highly consistent consensus sequences without requiring external multiple sequence alignment tools such as MAFFT or relying on species-specific known motifs. Taking the Arabidopsis Col-CEN genome as an example, the optimal period was found to be 178 bp, with a column consistency rate of 0.9272, an internal consistency rate of 0.8764, and a consistency rate exceeding 98% with the full-length reference sequence of the known centromere. This result significantly outperforms traditional multiple sequence alignment tools such as MAFFT (consistency rate approximately 91.3%), validating the superiority of this method in phase correction and consensus sequence construction.
[0049] 3. This invention transforms four independent biological characteristics into a unified probability score through weighted summation, allowing each candidate to obtain a clear sub-score and overall score. Users can adjust the threshold or trace the contribution of each sub-score according to actual needs, solving the problems of "black box" decision-making and difficulty in attribution in existing deep learning methods, and providing traceable quantitative evidence for subsequent research and result verification.
[0050] 4. This invention combines tandem repeat detection with prefix sum vectorization to achieve rapid scanning at the whole genome scale, which can usually be completed within tens of seconds; at the same time, it combines k-mer distance calculation based on matrix vectorization with hierarchical clustering to efficiently process thousands of candidate monomers, meet the routine analysis needs of high-throughput genomic data, and has good scalability and application value.
[0051] 5. This invention provides a robust, interpretable, and reproducible scoring-consensus integrated technical approach for identifying tandem repeats of plant centromeres. It solves the technical problems of insufficient false positive control, lack of interpretable quantitative evaluation, and easy introduction of systematic bias in consensus sequence construction in existing technologies, and has clear application value. Attached Figure Description
[0052] Figure 1 is a chromosome distribution diagram of whole-genome tandem repeat region annotation in an embodiment of the present invention;
[0053] Figure 2 is a histogram of the CenScore distribution of candidate monomers in an embodiment of the present invention;
[0054] Figure 3 is a k-mer enrichment analysis diagram of an embodiment of the present invention;
[0055] Figure 4 is a t-SNE dimensionality reduction visualization of the hierarchical clustering results of candidate single-unit k-mer binary vectors in an embodiment of the present invention;
[0056] Figure 5 shows the consensus sequence and column conservation diagram of Arabidopsis centromeres in an embodiment of the present invention;
[0057] Figure 6 shows the score(P) and column consistency rate and internal consistency rate scoring chart of the embodiment of the present invention. Detailed Implementation
[0058] This application provides a method for scoring and identifying plant centromere tandem repeat sequences. The general approach is as follows: First, the sequence information of the target plant genome is obtained. An efficient prefix sum vectorized tandem repeat detection method is used to rapidly scan the entire genome, identify candidate repeat regions, and extract repeat units as candidate monosomal sequences, while eliminating simple repeat noise with low base composition complexity. Then, a multi-feature fusion score (CenScore) is calculated for each candidate monosomal. This score is weighted and fused from four dimensions: periodic consistency, genome enrichment, sequence complexity, and GC content shift, quantifying the probability that a candidate monosomal belongs to a centromere into a unified probability scale. Next, high-scoring candidate monosomal sequences are hierarchically clustered based on the k-mer existence vector and Jaccard distance to classify different monosomal subtypes. Finally, to address the phase drift problem that easily occurs in tandem repeat sequences, a phase correction strategy combining exhaustive cyclic shift and reverse complementation is adopted. Through iterative consensus guidance and global self-consistent alignment, a high-fidelity consensus sequence is constructed. This embodiment uses the Arabidopsis thaliana Col-CEN genome as the test object to provide a detailed explanation and performance verification of the above method.
[0059] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.
[0060] Example 1
[0061] This embodiment provides a method for scoring and identifying plant centromere tandem repeat sequences, including the following steps:
[0062] (1) Data download
[0063] The target genome used in this embodiment is the Arabidopsis thaliana Col-CEN assembly version (Col-CEN_v1.2), which was released by the Schatz laboratory. This high-quality assembly is specifically optimized for the centromere region of Arabidopsis and can be downloaded from https: / / github.com / schatzlab / Col-CEN. The genome contains 5 chromosome sequences (Chr1–Chr5), with a total length of approximately 132 Mb.
[0064] (2) Tandem repeat detection and candidate monomer extraction
[0065] For the target genomic DNA sequence in the candidate cycle set ( Within a range of bp, a prefix sum and vectorized similarity scan are performed to identify consecutive highly similar segments and merge adjacent candidate regions, constructing a set of candidate repeating intervals. The following sliding window similarity formula is used to determine contiguous repeating regions:
[0066]
[0067] in, For sliding window similarity, The starting position index of the sliding window. The candidate period length, and For prefix sum vectors at position and The formula calculates the proportion of bases matching period P within a window using prefix sums and differences, achieving a fast scan with O(n) time complexity.
[0068] Subsequently, repetitive units were extracted from the original sequence based on the genomic coordinates of each candidate region as candidate monomers, and low-complexity monomers were eliminated according to the following AT base ratio formula to reduce the impact of poly(A) / poly(T) noise on subsequent scoring:
[0069]
[0070] in, The ratio of AT bases in the candidate monomer sequence; Candidate monomeric sequences Length; For sequence The total number of bases A and T in the medium. When The candidate monomer is then removed.
[0071] In this embodiment, a total of 11,006 tandem repeat regions were obtained on 5 chromosomes, and 86,286 candidate monochromatic regions were extracted. Among them, the 178 bp periodicity region had a significant advantage in both quantity and cumulative length. Figure 1 As shown, it is highly consistent with the known biological scale of Arabidopsis centromere satellite monomers.
[0072] (3) CenScore multi-feature fusion score
[0073] For the candidate monomers retained after step (2), the CenScore fusion score is calculated using the following formula:
[0074]
[0075] in, Candidate monomeric sequences, For sequence Length, The period length; , , , Candidate monomer sequences The periodic consistency score, genome enrichment score, sequence complexity score, and GC content offset score are used.
[0076] Among them, periodic consistency score :
[0077]
[0078] in, Candidate monomeric sequences Length, The period length, This is the position index for alignment after the sequence has been shifted. For indicator functions, when (Right now Upper Position and First When the bases are the same, the value is 1; otherwise, the value is 0.
[0079] Genome enrichment score Calculate using the following three formulas:
[0080]
[0081]
[0082]
[0083] in, Candidate monomeric sequences, For sequence Length, For k-mer length, Index of the window's starting position. For sequence From the first consecutive starting bits The k-mer corresponding to each base. and They are respectively Frequency of occurrence in the whole genome background and candidate haplogroup To avoid a preset positive small constant at zero frequency, For the first The log-odds ratio of sliding windows For all The arithmetic mean, and These represent taking the maximum value and taking the minimum value, respectively.
[0084] Sequence complexity score :
[0085]
[0086] in, Candidate monomeric sequences, For traversing the set The base types in base In sequence The frequency in.
[0087] GC content offset score :
[0088]
[0089] in, For sequence GC base ratio, To preset the target GC content, For the GC deviation penalty width parameter, To obtain the maximum value.
[0090] In this embodiment, the k-mer length Take 5 (corresponding to) (k-mer feature space), whole-genome background frequency sampling size per chromosome Magnitude Take 0.38, We set the value to 0.30. The mean CenScore of the candidate monomers is approximately 0.52, mainly distributed in the range of 0.45 to 0.80 (e.g., ...). Figure 2 As shown); when sequences with CenScore(M) ≥ 0.45 were identified as high-confidence centromere candidates, 80,194 high-confidence candidate monomers were retained (accounting for approximately 93% of the total candidates), with an average GC content of approximately 0.38 (as shown). Figure 2 As shown), and exhibits a specific k-mer enrichment pattern (e.g. Figure 3 As shown in the figure, it matches the sequence characteristics of typical plant centromere satellite repeats, indicating that multi-feature fusion is beneficial for effectively distinguishing highly periodic rDNA, microsatellites and other non-centromere tandem repeats at the sequencing level.
[0091] (4) Hierarchical clustering, phase correction and optimal period selection
[0092] Encode high-confidence candidate monomers into k-mer binary vectors and calculate the fully vectorized Jaccard distance matrix. Then, use an average link-based hierarchical clustering strategy to classify subtypes (e.g., ...). Figure 4 (As shown in the figure) Exhaustive cyclic shift and reverse complementary phase search are performed within each subtype. After three rounds of consensus-guided iteration and global cyclic shift self-alignment of the consensus sequence, a consensus sequence with high column consistency rate and internal consistency is obtained.
[0093] In the periodic fine search interval Within bp, for each candidate Calculate the overall score using the following formula and take the period corresponding to the maximum value as the optimal period:
[0094]
[0095] in, For period The corresponding column consistency rate For period The corresponding average internal consistency between individual units; take The maximum value corresponding to This is the optimal period.
[0096] like Figure 5 and Figure 6 As shown, in this embodiment, the optimal period is determined to be 178 bp. It is approximately 0.9119 (obtained by weighted combination of column consistency rate 0.9272 and internal consistency 0.8764).
[0097] (5) Comparison of the effects with existing processing approaches
[0098] To objectively demonstrate the technical effectiveness of this embodiment, it is compared with two common processing approaches. The first approach relies solely on tandem repeat detection and heuristic threshold screening, without introducing multi-feature probabilistic scoring, thus making it difficult to provide a unified and interpretable confidence ranking for a large number of candidate sequences. The second approach directly calls external multiple sequence alignment tools (such as MAFFT) to generate consensus sequences, but it is prone to systematic bias when dealing with highly similar tandem repeats with phase drift. This embodiment, without calling external alignment tools or relying on preset motifs, performs a full-length BLAST alignment of the generated consensus sequence with known reference sequences, achieving a consistency rate exceeding 98%, indicating that this invention balances the interpretability of scoring and the quality of consensus sequence construction in de novo scenarios.
[0099] In summary, the plant centromere tandem repeat sequence scoring and identification method of the present invention can stably output cycle and GC features consistent with biological priors at the whole genome scale, while obtaining consensus sequences with high column consistency rates and verifiable reference sequence consistency, thereby supporting the quantitative characterization of centromere regions and downstream applications.
[0100] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention.
[0101] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.
Claims
1. A method for identifying tandem repeat sequences of plant centromeres, characterized in that, Includes the following steps: (1) Obtain the sequence information of the target plant genome, perform tandem repeat detection on the sequence information, and obtain candidate repeat regions and their corresponding cycle lengths P; (2) Extract repeat units as candidate monomer sequences based on the genomic coordinates of the candidate repeat regions, and remove candidate monomers whose base composition complexity is lower than a preset threshold. (3) Calculate a multi-feature fusion score for each candidate monomer, wherein the fusion score includes at least two of the following: cycle consistency score, genome enrichment score, sequence complexity score, and GC content offset score, and screen candidate monomer sequences based on the fusion score; (4) Cluster the candidate monomer sequences after screening in step (3) to obtain at least one monomer subtype; (5) Perform phase correction on the monomer sequence within each monomer subtype and construct a consensus sequence based on the corrected sequence.
2. The identification method as described in claim 1, characterized in that, In step (3), the fusion score is calculated using the following formula: in, Candidate monomeric sequences, For sequence Length, The period length; , , , Candidate monomer sequences The periodic consistency score, genome enrichment score, sequence complexity score, and GC content offset score are given, with w1, w2, w3, and w4 being the weight coefficients of each score, and w1 + w2 + w3 + w4 = 1.
3. The identification method as described in claim 2, characterized in that, The weighting coefficients w1, w2, w3, and w4 are set to 0.35, 0.30, 0.20, and 0.15, respectively.
4. The identification method as described in claim 1, characterized in that, In step (1), the concatenation repetition detection employs a prefix sum vectorized O(n) algorithm, including: Encode the DNA sequence as an integer array; construct a matching array such that the first... Position and First Record 1 if the positions are equal, otherwise record 0; calculate the prefix sum of the matching array to obtain the array Pfx; calculate the sliding start position according to the following formula. Similarity at location : in, The starting position index of the sliding window. The candidate period length, and For prefix sum vectors at position and The element value at that location.
5. The identification method as described in claim 2, characterized in that, The periodic consistency score Calculate using the following formula: in, Candidate monomeric sequences Length, The period length, This is the position index for alignment after the sequence has been shifted. For indicator functions, when (Right now Upper Position and First When the bases are the same, the value is 1; otherwise, the value is 0. The genome enrichment score Calculate according to the following three formulas in sequence: in, Candidate monomeric sequences, For sequence Length, For k-mer length, Index of the window's starting position. For sequence From the first consecutive starting bits The k-mer corresponding to each base. and They are respectively Frequency of occurrence in the whole genome background and candidate haplogroup To avoid a preset positive small constant at zero frequency, For the first The log-odds ratio of sliding windows For all The arithmetic mean, and These represent taking the maximum value and taking the minimum value, respectively. The sequence complexity score Calculate using the following formula: in, Candidate monomeric sequences, For traversing the set The base types in base In sequence The frequency in; The GC content offset score Calculate using the following formula: in, For sequence GC base ratio, To preset the target GC content, The GC deviation penalty width parameter, To obtain the maximum value.
6. The identification method as described in claim 1, characterized in that, In step (5), the phase correction includes: adjusting the period to... Each single enumeration positive chain Cyclic shift and reverse complementary chain Cyclic shift Candidates are identified; the candidate with the highest positional consistency rate is selected as the optimal phase for that entity, based on the current consensus sequence; at least three rounds of iteration guided by consensus updates are performed; the consensus sequence is enumerated. Cyclic shift, in which Given the period length, the offset with the highest average positional consistency rate with all original units is selected as the globally optimal phase; finally, the consensus sequence is reconstructed by majority vote in each column.
7. The identification method as described in claim 1, characterized in that, Before constructing the consensus sequence in step (5), the optimal period selection step is also included: For each candidate period in the candidate period set Calculate the consistency rate of each column separately. Average internal consistency between monomers And select the optimal period according to the following formula: in, For period The corresponding column consistency rate For period The corresponding average internal consistency between individual units; take The maximum value corresponding to This is the optimal period.
8. The identification method as described in claim 1, characterized in that, In step (4), the clustering adopts Jaccard distance hierarchical clustering based on k-mer, including: Candidate monomer sequences are encoded as k-mer existence vectors, and the Jaccard distance between each pair is calculated. Hierarchical clustering is performed based on the Jaccard distance, and monomer subtypes are divided according to a preset similarity threshold.
9. A set of high-confidence plant centromere candidate monomers obtained by screening using the method described in any one of claims 1 to 8, or a consensus sequence of plant centromere tandem repeat sequences constructed therefrom.
10. The application of the method as described in any one of claims 1 to 8 in centromere region annotation of plant genomes, centromere function research, plant variety identification, chromosome fluorescence in situ hybridization probe design, or comparative genomics analysis.