Prediction transcript screening method and device for reference-genome-free species

Abstract

The invention belongs to the field of bioinformatics, and particularly relates to a prediction transcript screening method and device for reference-genome-free species. In order to solve the problems that existing software and methods are insufficient in pseudo transcript recognition capacity and high in false positive, the novel transcript screening method is provided and comprises the steps of obtaining input data, preprocessing the data, conducting BUSCO conserved gene matching, conducting ORF integrity detection, conducting ORF homology judgment, conducting ORF length detection and conducting statistical scoring through a fusion confidence coefficient function. And obtaining a final score F of the predicted transcript. The invention provides an efficient, universal and low-use-threshold prediction transcript screening method for reference-free genome species, and the screened transcript is real and reliable, has both efficiency and practicability, and is expected to become a key method for improving efficiency and gene research accuracy of bioinformatics and molecular biology workers in the future.

Description

A method and apparatus for predictive transcript screening of species without a reference genome Technical Field

[0001] This invention belongs to the field of bioinformatics, specifically relating to a method and apparatus for predictive transcript screening of species without a reference genome. Background Technology

[0002] For functional gene research in non-model plants and animals, the mainstream and commonly used strategy is to perform reference-free transcriptome analysis. Reference-free transcriptome analysis refers to the process of assembling raw mRNA sequencing data reads into "pseudo-transcripts" or "contigs" using various assembly algorithms (such as Trinity and SPAdes), and then performing quantitative analysis. This method has the advantage of not requiring reference genome back-linking, making it the primary choice for many researchers studying non-model species: because 99% of species on Earth have not yet completed whole-genome sequencing.

[0003] However, traditional parameter-free transcriptomics workflows suffer from significant technical drawbacks: a high false positive rate in the assembly results. The assembled so-called "transcripts" often contain numerous sequence artifacts, chimeras, or non-coding fragments; they are not actually transcription products present in the organism. When researchers select a gene and design primers, they may find that PCR amplification fails to produce any product, or amplifies other unknown products. This not only results in a huge waste of research reagents and time but also severely restricts the screening and identification of high-reliability functional genes, becoming a technical bottleneck in non-model organism research.

[0004] Currently, existing software and methods are insufficient in identifying pseudotranscripts. In complex or highly repetitive genomes, preliminary transcript assembly often generates a large number of pseudotranscripts from non-true sources, including misassemblies caused by repetitive sequences and misassemblies between members of homologous gene families. Existing software such as TRANSDECODER relies primarily on preliminary assembly results and general coding region prediction models, without adequately considering the identification and removal of pseudo-coupling sequences. Therefore, erroneous open reading frames may be misclassified as potential coding sequences, leading to the retention of low-quality or biologically insignificant candidate genes in subsequent analyses. Such errors not only affect the accuracy of functional annotation but also increase the risk and cost of subsequent experimental validation. Furthermore, existing technologies lack systematic quality control and comprehensive judgment mechanisms. Current software typically assesses transcript quality from a single dimension (such as coding potential or homology), lacking a comprehensive judgment strategy that incorporates multi-source evidence (such as sequence characteristics, evolutionary patterns, domain predictions, and cross-species conservation). In species with incomplete genomes or containing a large number of unknown sequences, this unilinear judgment method is prone to errors, further weakening the reliability and coverage of prediction results.

[0005] Therefore, there is an urgent need for a method that can effectively, accurately, and conveniently screen for predictive transcripts in species without a reference genome. Summary of the Invention

[0006] The purpose of this invention is to provide a method and apparatus for predictive transcript screening of species without a reference genome.

[0007] To achieve the above-mentioned objectives, the technical solution adopted by this invention is: a method for predictive transcript screening of a species, characterized by comprising the following steps:

[0008] S1. Obtain input data: Use pseudo-transcripts to be analyzed as input data;

[0009] S2. Data preprocessing: Merge transcripts with high similarity;

[0010] S3. The data processed in S2 were evaluated for conserved gene matching using BUSCO software. The matching degree was calculated using the following formula:

[0011]

[0012] Genes that output "Complete" receive 100% confidence; genes that output "Fragmented" receive 40% confidence; and genes that output "Duplicated" receive 0% confidence.

[0013] S4. Perform ORF integrity checks on the data processed in S2; the ORF integrity confidence scoring function is as follows:

[0014]

[0015] Genes that output "Complete" have a 100% confidence level; genes that output "Partial" have a 60% confidence level; and genes that output "Internal" have a 0% confidence level.

[0016] S5. Perform ORF homology assessment on the data processed in S2; the homology confidence scoring function for ORF homology assessment is as follows:

[0017]

[0018] Where Z is the homology score, defined as -log 10 (e-value).

[0019] S6. Perform ORF length detection on the data processed in S2; the ORF length confidence scoring function is as follows:

[0020]

[0021] Where L is the transcript length, and the mean and standard deviation are determined by statistics of the entire dataset.

[0022] S7. Scoring is done based on the BUSCO conserved gene matching, ORF integrity, ORF homology, and ORF length to obtain the final score F for the predicted transcript.

[0023] The formula for calculating the rating F is as follows:

[0024] ,

[0025] in, Confidence level of BUSCO conserved gene matching results For ORF integrity confidence, For ORF homology confidence, The confidence level for the ORF length.

[0026] Alternatively, the formula for calculating the score F is as follows:

[0027]

[0028] Where j is the mathematical traversal variable; P j As j is traversed, the probability of credibility of biological structural evidence and biologically conserved evidence will be represented.

[0029] Accordingly, a computer device / apparatus / system includes a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method.

[0030] Accordingly, a computer-readable storage medium stores a computer program / instructions thereon, which, when executed by a processor, implement the steps of the method.

[0031] Accordingly, a computer program product includes a computer program / instructions that, when executed by a processor, implement the steps of the method.

[0032] This invention offers the following advantages: It provides a novel integrated transcript screening and annotation system. Addressing the shortcomings of existing technologies, such as insufficient accuracy in transcript prediction, limited ability to identify pseudo-transcripts, and reliance on reference genomes for non-model species annotation, this invention constructs a unified bioinformatics analysis workflow. This workflow can automatically complete transcript quality assessment, authenticity determination, functional annotation, and candidate gene screening under conditions of no reference genome or insufficient reference information, ultimately designing reliable PCR primers.

[0033] This invention organically unifies complex bioinformatics software by encapsulating existing powerful software and employing an intelligent result merging and voting paradigm. By aligning the standard outputs of each software, it adjusts the outputs of upstream software. This ensures that the output files of software A are correctly and seamlessly transmitted to software B, while simultaneously allowing for the storage of results from the analysis software A. It unifies the communication bridge between software programs and truly enables each software to leverage its strengths.

[0034] The intelligent voting mechanism of this invention is a key innovation. It treats each piece of evidence as an independent supporting event for the authenticity of the target transcript, and obtains the final fusion confidence value by calculating the joint confidence score when at least one piece of evidence is valid. This method aligns with biological reality: in the absence of a reference genome, the presentation of evidence for true genes is highly diverse; it is difficult for a gene to have excellent homology, BUSCO conservation, ORF length, and other indicators, making the judgment too complex to be reasonably fused using a simple linear model.

[0035] In addition to considering the "false positive" scenario where a fake transcript is identified as a real one, this invention also considers the "false negative" scenario where a large number of real genes are systematically missed. As long as one piece of highly credible evidence exists, there is reason to strongly suspect that the transcript is a real gene. Each type of evidence is obtained through different biological principles, different algorithmic paths, and different data sources, providing independent support for the authenticity of the transcript.

[0036] Moreover, this invention is easy to operate; any researcher can conveniently obtain all screening results with a single short command, and the screening score for each gene is clear and readily apparent. Using the method of this invention, the time for a researcher proficient in bioinformatics software to screen genes (referring to the time spent using the integrated software and manually judging the results) can be reduced from 20 genes selected in one week to 50 genes in 10 minutes. The time for manually designing primers—10 minutes for designing one gene—can be reduced to 50 genes in one second.

[0037] Compared to no screening, the success rate of PCR for pseudo-transcripts screened by this invention is increased by 100%, and the probability of correct PCR products is increased by 120%. Compared to using only TRANSDECODER software for judgment, the success rate of PCR for pseudo-transcripts screened by this invention is increased by 60%, and the probability of correct PCR products is also increased by 60%.

[0038] In summary, this invention provides an efficient, universal, and low-barrier-to-use method for predicting transcripts of species without reference genomes. The selected transcripts are accurate and reliable, combining efficiency and practicality. It is expected to become a key method for bioinformaticians and molecular biologists to improve efficiency and accuracy in gene research in the future. Attached Figure Description

[0039] Figure 1 is a schematic diagram of the quality of transcript assembly detected by BUSCO using the software designed by the method of the present invention.

[0040] Figure 2 is a schematic diagram of the software designed using the method of the present invention to predict ORF;

[0041] Figure 3 is a schematic diagram of the software output results designed using the method of the present invention;

[0042] Figure 4 is one of the schematic diagrams of PCR primer design using the software output results designed by the method of the present invention;

[0043] Figure 5 is a second schematic diagram of PCR primer design using the software output results designed by the method of the present invention.

[0044] Figure 6 shows the gel electrophoresis diagram obtained by amplification using the designed primers. Detailed Implementation

[0045] This invention provides a method for predictive transcript screening of species without a reference genome, specifically including the following steps:

[0046] S1. Obtain Input Data: Use pseudo-transcripts obtained from existing technologies as input data. For example, Trinity software can be used to analyze the target and obtain pseudo-transcripts. Trinity software is the most widely used software for reference-free transcriptome analysis in existing technologies. Pseudo-transcript results from Trinity or other reference-free transcriptome assembly software can be used as input files.

[0047] S2. Data Preprocessing: Transcripts with high similarity are merged to reduce subsequent computation time. If using Trinity software, this step can be completed directly using its CD-HIT function. Transcripts with ≥98% nucleic acid sequence similarity or ≥95% amino acid sequence similarity can be considered highly similar.

[0048] After transcript merging, BUSCO (Benchmarking Universal Single-Copy Orthologs) integrity assessment is performed using BUSCO software. If the BUSCO integrity C-value (i.e., the proportion of intact conserved genes) is greater than 90%, the transcript assembly quality is considered good, and step S3 is performed. If the BUSCO C-value is less than 80%, it indicates that a certain number of conserved genes were not detected by BUSCO during transcript assembly, and the transcript is deemed unqualified and cannot be used; it must be returned to S1 to obtain a pseudo-transcript. If the C-value is between 80% and 90%, the transcript quality is considered average, but does not meet the standard for direct rejection. Step S3 can continue, but 0.1 should be deducted from the final score F calculated according to the formula.

[0049] S3. Perform BUSCO conserved gene matching on the data processed in S2:

[0050]

[0051] This step uses the BUSCO software. BUSCO (Benchmarking Universal Single-Copy Orthologs) detects conserved single-copy orthologous genes, with a weighting that can be set to 40%. If a transcript matches a BUSCO gene, it indicates that it is an important core functional gene. It's important to note that step S2 performs a BUSCO integrity assessment, evaluating the quality of the entire transcriptome (how many core genes were assembled), and the results of step S2 directly affect whether to proceed to the next step. This step, however, performs a BUSCO gene matching assessment, evaluating whether a single transcript is a core conserved gene. The result of this assessment only affects the score; regardless of the assessment result, the next steps must be performed.

[0052] In the method provided by this invention, only genes in the "Complete", "Fragmented", and "Duplicated" states are counted. When the BUSCO software outputs genes in the "Complete" state, it will obtain 100% confidence; while genes in the "Fragmented" and "Duplicated" states will obtain 40% and 0% confidence, respectively.

[0053] S4. Perform ORF integrity checks on the data processed in S2:

[0054] Perform this step using the TransDecoder software. A complete ORF (containing both start and stop codons) is more likely to represent the true encoded region. The weight for this item can be optionally set to 20%.

[0055] Complete: Start codon + CDS + stop codon; this type is optimal. Partial: 5prime_partial (5' end incomplete): Missing start codon, possibly indicating incomplete assembly or a true N-terminal truncated protein; 3prime_partial (3' end incomplete): Missing stop codon; these two types are suboptimal. Internal: Both ends are incomplete; this type is the worst and is not recommended. The ORF integrity confidence scoring function is a segmentation function that accepts a string:

[0056]

[0057] Genes output by TransDecoder in the "Complete" state will receive 100% confidence; while genes in the "Partial" and "Internal" states will receive 60% and 0% confidence, respectively.

[0058] S5. Perform ORF homology assessment on the data processed in S2: ORFs supported by homologous proteins are more likely to be genuine. The weight of this item can be set to 20%. If there is homology evidence, i.e., a significant match is found in the protein database, its homology score (e-value) is extracted and the confidence score is calculated; if there is no homology evidence, although there is a possibility of a new gene, its confidence score is also recorded as 0. The homology confidence scoring function is:

[0059]

[0060] Where Z is the homology score, which is defined as -log 10 (e-value); e-value is the most important indicator for software such as BLAST to determine homologous proteins. The smaller the e-value, the higher the degree of homology. In one optional approach, an e-value less than 0.01 is considered homologous. e-value is the expected value, referring to the expected number of sequences that match the query sequence with a degree no less than the current result. Therefore, when the e-value is very small, the protein being searched has a very small probability of being false ("false" means not homologous to the target protein). The protein alignment function built into the TransDecoder software automatically outputs the e-value. α is an adjustment parameter (range (-1, 0)) used to control the gradient of the entire function.

[0061] S6. Perform ORF length detection on the data processed in S2: The ORF length can be set with a weight of 20%: longer ORFs are usually more reliable because the probability of randomly generated long ORFs is extremely low. Assuming that the length of the protein-encoding mRNA (or transcript) follows a normal distribution, the probability of transcripts that are too long or too short is low.

[0062] The ORF length confidence score function is a normal distribution function.

[0063]

[0064] Where L is the transcript length, μ is the mean, and σ is the standard deviation; the mean and standard deviation are determined by statistics of the entire dataset.

[0065] S7. Use the fusion confidence function to calculate the score and obtain the final score F of the predicted transcript.

[0066] In one alternative approach, F is calculated using the following formula:

[0067]

[0068] in, The confidence level of the BUSCO conserved gene matching results. For ORF integrity confidence, For ORF homology confidence, The confidence level for the ORF length.

[0069] Preferably, the formula for calculating F is:

[0070]

[0071] Here, j is the traversal variable, which iterates through the elements in the value (BIO, struct) of j, first transforming them into biological structural evidence (struct), then into biological conserved evidence (BIO), and finally multiplying the confidence level by a factorial. Thus, P j This becomes the probability of credibility of biological structural evidence and biologically conserved evidence.

[0072] The formula is constructed as follows:

[0073] The evidence obtained in the above steps is divided into two groups: (1) Biological Evidence, which includes the matching degree of BUSCO conserved genes and ORF homology; and (2) Structure Evidence, which includes ORF length and ORF integrity.

[0074] For a structural reliability evidence set, define a gene structure confidence function. :

[0075]

[0076] In the formula, M represents any transcript, i.e., the independent variable in the mathematical formula. i refers to the traversal variable in mathematics. In this formula, i will first become ORF (Open Reading Frame Integrity), at which point φ ORF (M) then naturally becomes φ mentioned above. ORF (Y), which represents the ORF integrity confidence; subsequently, i will become LEN (ORF length), at which point φ LEN (M) then naturally becomes φ mentioned above. ORF (L) represents the ORF length confidence score. This formula calculates the ORF integrity confidence score and the ORF length confidence score separately, then multiplies the two confidence scores together, subtracts the result from 1, and finally obtains the biological structure reliability.

[0077] The advantage of this formula is that as long as the transcript performs well in either "ORF integrity" or "ORF length reasonableness," its structural reliability is significantly improved. It also has a certain tolerance for any low-probability evaluation item, thus avoiding false negatives.

[0078] For a set of evidence of biological conservation, define a gene conservation confidence function. :

[0079]

[0080] In the formula, N also represents any transcript, that is, the independent variable in the mathematical formula. This is the confidence score of the BUSCO conserved gene matching results; It is the protein ORF homology confidence score.

[0081] β refers to the penalty amplification parameter; when the confidence level of a piece of evidence is high... , The β value approaches 0 more quickly, indicating that the larger the β value, the higher the value of a successful match and the heavier the cost of a failed match. The penalty amplification parameter for the confidence level of BUSCO conserved gene matching is set to twice the value of ORF homology. This is to ensure that the overall confidence level should show a non-linear jump once a BUSCO conserved gene match is found. This is because BUSCO conserved genes are very strong biological evidence; once a conserved gene is detected, the corresponding transcript almost certainly exists. However, the β value for ORF homology evidence should not be set too large. This is because the unreferenced transcriptome may contain species-specific genes, rapidly evolving genes, etc., which may not have significant homology. Therefore, setting the β value for ORF homology evidence too large would cause the entire formula to systematically exclude new genes, leading to false negatives.

[0082] The preferred value range for β is 0.5 to 1.5, more preferably 0.8 to 1.2, and the most preferred value for β is 1. From a mathematical perspective, for exponential functions, the magnitude of the exponent (i.e., β in the formula) greatly affects the function value. If it is greater than 1, the penalty is too severe; if it is less than 1, the penalty is almost negligible. The exponential function curve with an exponent of 1 is relatively mild.

[0083] Confidence function of integrated gene structure and gene conservatism confidence function The calculation results ultimately yield the formula for calculating the score F mentioned above, namely... .

[0084] If the BUSCO integrity C value is between 80% and 90% in step S2, then F will have 0.1 points deducted from the calculated score.

[0085] It should be noted that in the above operations, if Trinity software is used to obtain pseudo-transcripts, the pseudo-transcripts output by this software are in FASTA format, which can be used as direct input for CD-HIT and BUSCO software. CD-HIT and BUSCO software also output FASTA files. However, when the FASTA file is input into TransDecoder software required for subsequent steps, the output of TransDecoder software is more complex and difficult to distinguish. An example of TransDecoder software output is shown below:

[0086] transcripts.fasta.transdecoder_dir /

[0087] ├── longest_orfs.pep

[0088] ├── longest_orfs.cds

[0089] ├── longest_orfs.gff3

[0090] ├── longest_orfs.bed

[0091] └── longest_orfs.mRNA

[0092] ├── longest_orfs.pep

[0093] ├── transcripts.fasta.transdecoder.cds

[0094] ├── transcripts.fasta.transdecoder.pep

[0095] ├── transcripts.fasta.transdecoder.gff3

[0096] └── transcripts.fasta.transdecoder.bed

[0097] To address this issue, the preferred solution is as follows: In subsequent steps of ORF detection and judgment, this invention discards files related to the longest ORF, because in organisms, even the longest ORF may not necessarily translate into a protein. Furthermore, this invention comprehensively considers BUSCO scoring, ORF length, and homology scores. This selection effectively avoids being misled by the incomplete or biased output of TransDecoder software.

[0098] Given the output of the TransDecoder software, the preferred file extracted in this invention is transcripts.fasta.transdecoder.gff3. The gff3 file shows the position of each ORF within its own transcript. Based on this positional information, the transcripts.fasta.transdecoder.cds transcript and its ORF sequence are then extracted from the gff3 file.

[0099] Finally, PCR was performed on the high-scoring, qualified results. Another advantage of choosing the gff3 file is that the location information provided in it helps the primer designer understand the ORF at the specific location required.

[0100] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Unless otherwise specified, the technical means used in the embodiments are conventional means well known to those skilled in the art, and the data obtained are all average values ​​obtained after at least three repetitions, and each repetition yields valid data.

[0101] Example: Screening and validation of Leymus chinensis transcripts

[0102] The method provided in this invention was used to test the mature-stage no-reference transcriptome of *Leymus secalinus*. *Leymus secalinus* is a complex allotetraploid plant with an extremely complex genome and abundant transcripts, making the assembly of no-reference transcripts in this group of plants extremely difficult; therefore, this species was chosen for testing.

[0103] After assembly using the existing software Trinity, the number of reliable pseudotranscripts for a single mature stage of *Leymus chinensis* reached as high as 180,232, far exceeding the number of transcripts with reference genomes for other grasses (e.g., the wheat transcriptome has approximately 10,000 transcripts with reference genomes). This order-of-magnitude difference suggests that 90% of these sequences are non-existent spurious sequences, and the screening process is extremely difficult.

[0104] Using the aforementioned dataset of 180,232 transcripts, the design process for this invention was carried out, resulting in approximately 1,600 reliable transcripts. However, this dataset contained too many genes, which was not conducive to visualization. Therefore, we selected a small sample of *Ligustrum lucidum* unreferenced transcripts (20 transcript sequences) for validation of this invention. The outputs of this invention are shown in Tables 1 and 2.

[0105] Table 1 Output File Table (First 5 Rows)

[0106]

[0107] Table 2 Output File Table (First 5 rows, continued)

[0108]

[0109] For example, the transcript TRINITY_DN2332_c0_g1_i5_2 is a key candidate gene, scoring a high 0.6. This is because: its ORF (open reading frame, the basic unit for mRNA to be translated into protein) is intact (corresponding to "ORF_Type" as complete), its length is 1404 base pairs, which is the standard length required for a protein, and its protein product has been validated in other species. Although this gene is not present in the BUSCO database (i.e., it lacks the 0.4 points required for BUSCO gene matching), other indicators support the transcript TRINITY_DN2332_c0_g1_i5_2 as a key candidate gene.

[0110] PCR primers were designed and validated for all key candidate genes, and it was found that the key genes such as the transcript TRINITY_DN2332_c0_g1_i5_2 screened in this invention actually exist.

[0111] The specific process of testing a small dataset of 50 random transcripts from the root system of Leymus chinensis using the software constructed using the method of this invention is as follows:

[0112] S1. Input a reliable pseudo-transcript obtained after assembly using the existing software Trinity;

[0113] S2 and BUSCO were used to assess the quality of transcript assembly, and the results are shown in Figure 1. The results indicate that BUSCO used the transcription mode to identify the integrity of parameter-free transcript assembly, adapting to sequence characteristics at the transcript level and avoiding misjudgments of intron structure and gene integrity as seen in the genome mode. This mode comprehensively considers factors such as splice variants and transcript fragmentation, more accurately reflecting the ability of transcript assembly to restore functional genes, i.e., the "integrity assessment" mentioned earlier. In this dataset, the 50 randomly selected transcripts do not represent the complete whole-tissue transcript situation; therefore, this step was manually skipped.

[0114] It should be noted that the BUSCO software is designed to detect the integrity of conserved genes in an entire living organism (e.g., plant tissue). If these genes are incomplete, the plant tissue cannot perform its full biological functions, which is impossible in nature. Therefore, if conserved genes are incomplete, and the BUSCO integrity C-value is below 80%, it indicates a problem in the sequencing or computation steps. In this case, the pseudo-transcript should be discarded according to the judgment criteria in step S2 of the specific implementation method described above. However, in this embodiment, for demonstration purposes, only 50 random transcripts from the roots of *Leymus chinensis* were selected for the process. This inevitably leads to incomplete conserved genes in the entire root tissue, resulting in a BUSCO integrity C-value below 80%. Since this is due to the manual selection of a small number of transcripts and is only for demonstration purposes, the subsequent processes continue. The dataset of 50 random transcripts in this embodiment is just a small use case. In real-world scenarios, researchers would input transcripts from the entire tissue into the software, making BUSCO integrity assessment valuable.

[0115] S3. Predicting ORFs: The results are shown in Figure 2. The results indicate that 28 genes showed evidence of homology. This means these 28 transcripts are more likely to translate into the actual proteins. Additionally, 22 complete ORFs were detected; transcripts containing complete ORFs and homologous proteins are more reliable.

[0116] S4. Output the results and generate a report. Part of the report is shown in Figure 3. The results show that some genes scored greater than 0.5, but were not identified as conserved genes by BUSCO; however, they are generally usable (because their length matches the protein-coding sequence; they have homology information; and their ORFs are complete). As shown in Figure 2, although 5 complete conserved genes and 4 incomplete conserved genes were ultimately found in the BUSCO conserved gene search, these 9 genes had low scores due to incomplete ORFs and other reasons, making PCR unsuitable. This is precisely the multi-evidence confidence fusion scheme advocated in this invention.

[0117] S5. PCR primers were designed using the genes with the highest scores in the output results (such as TRINITY_DN2332_c0_g1_i5_2), and the results are shown in Figures 4 and 5. Amplification was then performed using the designed primers, and the gel electrophoresis image is shown in Figure 6. The results showed that the gene was successfully amplified, and the DNA marker indicated that the gene was longer than genes such as TRINITY_DN4824_c0_g1_i19_3, which is highly consistent with the ORF prediction results.

[0118] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Any modifications, alterations, substitutions, or variations made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention shall fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for screening predictive transcripts of a species, characterized in that, Includes the following steps: S1. Obtain input data: Use pseudo transcripts to be analyzed as input data; S2. Data preprocessing: Merge transcripts with high similarity; S3. Perform BUSCO conserved gene matching assessment on the data processed in S2; S4. Perform ORF integrity check on the data processed in S2; S5. Perform ORF homology check on the data processed in S2; S6. Perform ORF length check on the data processed in S2. S7. Scoring is done based on the BUSCO conserved gene matching, ORF integrity, ORF homology, and ORF length to obtain the final score F for the predicted transcript.

2. The method according to claim 1, characterized in that: In step S3, conserved gene matching is performed using BUSCO software. The matching degree is calculated using the following formula: Genes that output "Complete" receive 100% confidence; genes that output "Fragmented" receive 40% confidence; and genes that output "Duplicated" receive 0% confidence.

3. The method according to claim 1, characterized in that: In step S4, when performing ORF integrity testing, the ORF integrity confidence scoring function is as follows: Genes that output "Complete" have 100% confidence; genes that output "Partial" have 60% confidence; and genes that output "Internal" have 0% confidence.

4. The method according to claim 1, characterized in that: In step S5, when performing ORF homology determination, the homology confidence scoring function is: Where Z is the homology score, defined as -log 10 (e-value), where α is the adjustment parameter.

5. The method according to claim 1, characterized in that: In step S6, when performing ORF length detection, the ORF length confidence scoring function is as follows: Where L is the transcript length, μ is the mean, and σ is the standard deviation. The mean and standard deviation are determined by statistics of the entire dataset.

6. The method according to claim 1, characterized in that: In step S7, the formula for calculating the score F is as follows: in, Confidence level of BUSCO conserved gene matching results For ORF integrity confidence, For ORF homology confidence, The confidence level for the ORF length.

7. The method according to claim 1, characterized in that: In step S7, the formula for calculating the score F is as follows: Where j is the traversal variable; P j The probability of credibility of biological structural evidence and biological conservatism evidence is given. Biological structural evidence includes BUSCO conserved gene matching degree and ORF homology; biological conservatism evidence includes ORF length and ORF integrity.

8. A computer device / equipment / system, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1 to 7.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1 to 7.

Images