A third-generation full-length transcriptome-assisted gene prediction method
By combining the second-generation sequencing data and homologous species information, the intronic shear sites of the third-generation full-length transcriptome data are solved, and the problem of difficulty in correcting intronic shear sites in gene prediction in the prior art is achieved. Comprehensive and accurate prediction of gene structure and improve the reliability of gene prediction.
Patent Information
- Application Number
- CN202110322129.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-03-25
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2041-03-25
AI Technical Summary
The existing second-generation transcriptome sequencing data has assembly errors and incompleteness in gene prediction. Although the third-generation full-length transcriptome sequencing data is accurate, it is difficult to correct the intron shear site, resulting in limited accuracy of gene prediction and lack of Software compatible with three generations of full-length data can effectively correct intron shear sites.
By combining the second-generation sequencing data of the species and the gene information of homologous species, the intersection of intronic shear sites is obtained and combined with the third-generation full-length transcriptome data to correct intronic shear sites and obtain high reliability transcripts, thereby achieving comprehensive and accurate prediction of gene structure. Specific steps include outputting the intersection of intronic splicing sites of second-generation and homologous species, merging intronic splicing sites of third-generation transcriptomes, and aligning and assembly using specific software to obtain complete transcripts.
Improve the integrity and accuracy of gene prediction, can be applied in different species such as animals and plants, and realize high reliability prediction of genomic gene structure, especially the correction of intron shear sites, which enhances the reliability of gene prediction. sex.
Smart Images

Figure HDA0002993288810000011 
Figure HDA0002993288810000021 
Figure HDA0002993288810000022
Abstract
Description
Technical Field
[0001] The present invention relates to the field of bioinformatics, and in particular to a method for assisting in whole-genome gene structure prediction using third-generation full-length transcriptome sequencing data. Background Art
[0002] During the transcription process, eukaryotic genes trim introns and splice exons to form transcripts. It is precisely because of this splicing form that eukaryotic genes can adopt different splicing forms (variable splicing) to form different transcripts, thereby playing a more extensive and precise role. This also makes it difficult to predict the gene structure of eukaryotic organisms.
[0003] Currently, the following three different strategies are mainly used for eukaryotic gene prediction: homology-based prediction, de novo prediction, and transcriptome-based prediction. Since a large number of genomes have been published, the splicing sites can be determined by using the principle of good conservation of gene sequences between homologous species. Transcriptome prediction refers to the use of RNA-seq and three-generation full-length transcript data mixed from various tissues to assist gene prediction. Since transcriptome data is the most direct reflection of the gene structure of the species, this type of data can be used to determine the exon region and splicing site more accurately and realistically, and it is the most reliable strategy among the three gene prediction strategies.
[0004] The commonly used transcriptome-assisted gene prediction method currently uses the second-generation transcriptome sequencing data. However, the second-generation transcriptome sequencing usually breaks the extracted RNA fragments into small fragments for sequencing, and then assembles the small fragment sequencing data (read) (such as using Tirnity software) to obtain relatively complete transcripts. However, due to the short sequencing fragments, there may be assembly errors or incomplete assembly, resulting in the inability to accurately obtain complete transcripts, which in turn has a serious impact on the integrity and accuracy of gene prediction.
[0005] The third-generation sequencing platform currently under development, especially the Nanopore platform, is low-cost and can directly obtain high-quality full-length transcript sequences without assembly. One sequencing read can span the full-length transcript. Therefore, by aligning the read with the genome, it is very easy to determine the location of the gene on the genome and its complete structure, which is very conducive to gene annotation and has high accuracy. However, considering that the accuracy of all bases of the full-length transcripts obtained by the current third-generation sequencing platform is about 85%, there are some insertion and deletion errors, especially errors occurring at intron splicing sites (non-splicing sites, i.e., exon internal sequence errors can be directly corrected by aligning the genome, and there will be no errors; and the intron splicing site sequence cannot distinguish whether it is an intron or an error, resulting in difficulty in correcting this area) will have a more serious impact on gene prediction, thus limiting the use of most gene prediction software developed based on the second-generation transcriptome. New software must be compatible with this third-generation full-length data with a small number of errors.
[0006] Currently, there is only one software, LoReAn, that analyzes third-generation sequencing data to assist in gene prediction. However, this software can only correct the overall transcript sequence, and does not focus on the correction of the most core intron splicing sites in gene prediction, resulting in large correction errors. At the same time, it simply relies on its own and second-generation transcriptome predictions, without relying on homologous species information, which will lead to incomplete correction problems. Therefore, it is urgent to develop an intron splicing site correction method that combines multi-level data (homologous and transcriptome) and has a greater impact on gene prediction, so as to achieve a comprehensive and accurate prediction of the gene structure of the genome. Summary of the invention
[0007] The purpose of the present invention is to provide a method for predicting the whole genome gene structure using a third-generation full-length transcriptome.
[0008] In order to achieve the purpose of the present invention, the present invention provides a method for predicting gene structure at the whole genome level in a first aspect, comprising:
[0009] Use the species' second-generation sequencing data and homologous species gene information to predict the species' gene structure, and obtain the intersection of the species' second-generation sequencing data and the intron splicing site information in the homologous species gene information;
[0010] The intersection is merged with the intron splicing sites in the three-generation full-length transcriptome data of the species;
[0011] The merged intron splicing sites are used to identify and correct the intron splicing sites obtained by predicting the gene structure from the three-generation full-length transcriptome data of the species, and obtain transcripts.
[0012] Specifically, the method for predicting gene structure at the whole genome level provided by the present invention includes:
[0013] (1) Outputting the species second generation sequencing data and the intron splicing sites of the homologous species gene information of the species, and taking the intersection;
[0014] (2) outputting the intron splicing sites of the three-generation full-length transcriptome data of the species;
[0015] (3) The intersection described in (1) and the intron splicing site described in (2) are combined to identify and correct the intron splicing site obtained when predicting the gene structure of the species using the third-generation full-length transcriptome data to obtain the transcript of the gene.
[0016] In the method provided by the present invention, obtaining the intersection includes:
[0017] The gene structure of the species to be predicted is predicted based on the transcriptome data obtained from the second-generation Illumina platform, or the gene structure of the species to be predicted is predicted based on the gene information data of the homologous species, and the intron splicing sites with a read number greater than 2 are obtained, and the ratio of the intron splicing sites located in the intron to the exon is ensured to be greater than 0.5.
[0018] Specifically, in the method provided by the present invention, obtaining the intersection includes:
[0019] S1 predicts the gene structure of the species based on transcriptome data obtained from the second-generation Illumina platform;
[0020] S2 predicts the gene structure of the species using gene information data of homologous species of the species;
[0021] S3 outputs the intron splicing sites with read numbers greater than 2 in S1 and S2 respectively; and ensures that the ratio of intron splicing sites located in introns to those located in exons is greater than 0.5;
[0022] Compare the two intron splicing sites obtained in S3 and take the intersection.
[0023] In the method provided by the present invention, obtaining the intron splicing site in the three-generation full-length transcriptome data of the species includes:
[0024] The gene structure of the species to be predicted is predicted based on the full-length transcriptome data obtained by the third-generation Nanopore platform, and the spliced introns in the third-generation full-length transcriptome data start with GT and end with AG, or the spliced introns start with GC and end with AG; the intron splicing sites with a read number greater than 5 are obtained.
[0025] In the method provided by the present invention, obtaining the transcript comprises:
[0026] The merged intron splicing site extends 10-30 bases to the left and right, and the intron splicing site obtained by predicting the species gene structure from the third-generation full-length transcriptome data of the species is intersected, and the intron splicing site falling in the intersection is replaced with the merged intron splicing site. The method provided by the present invention uses genemaker-ST to realize genome-wide gene structure prediction after comparing the genome information of the transcript and the species.
[0027] As a specific embodiment of the present invention, a method for predicting the gene structure of the whole genome includes: Figure 1 :
[0028] (1) Download the genome file and gff3 file of the species to be predicted to be homologous from the NCBI website (https: / / www.ncbi.nlm.nih.gov / );
[0029] (2) Using the GeMoMa software, the gene information of the homologous species downloaded was used to predict the gene information of the genome of the species, and a gff3 file containing the required intron splicing site information was generated;
[0030] (3) Use the Illumina platform to perform next-generation transcriptome sequencing of the species to be predicted. The sequencing materials are derived from mixed samples of various tissues and organs to ensure that as many transcripts as possible are covered;
[0031] (4) Use HISAT2 to compare the sequencing data (read) with the genome of the species to be predicted, and obtain a comparison result file in bam format;
[0032] (5) Use the groupby command in the bedtools tool to convert the bam format to the bed format and extract the location of intron shearing;
[0033] (6) Count the number of reads supporting each intron splicing position, and retain those with a support greater than 2; at the same time, count the number of splicing sites located in introns and exons, ensuring that the ratio of splicing sites located in introns to exons is less than 0.5;
[0034] (7) Taking the intersection of the splicing sites obtained from the second-generation transcriptome and the homologous species, a highly reliable intron splicing site can be obtained;
[0035] (8) The Nanopore platform was used to perform third-generation transcriptome sequencing of this species. The sequencing materials were derived from mixed samples of various tissues and organs to ensure that as many transcripts as possible were covered;
[0036] (9) Use GMAP software to align the sequencing data (read) to the genome and output it in gff3 format. This file contains the original uncorrected intron splicing site information;
[0037] (10) Screen out the original intron splicing sites that meet the GT-AG and GC-AG rules and have at least 5 reads to support the intron splicing sites as candidate splicing sites with higher reliability in the third-generation transcriptome prediction, and merge them with the high-reliability intron splicing sites described in (7) to form a set of non-redundant high-reliability intron splicing sites; GT-AG represents that the intron starts with GT and ends with AG, and GC-AG represents that the intron starts with GC and ends with AG;
[0038] (11) Extend the non-redundant high-reliability intron splicing site by 20 bases to the left and right, use bedtools to take the intersection with the intron splicing site identified in (9), and use the non-redundant high-reliability splicing site to replace the splicing site identified by the third-generation transcriptome in the intersection, thereby completing the correction of the intron splicing site predicted by the third-generation full-length transcriptome sequencing data;
[0039] (12) After correcting the splicing sites predicted by the third-generation transcriptome sequencing data, if two transcripts correspond to the same splicing sites on the genome, the transcripts are merged to finally obtain a set of non-redundant and highly accurate candidate transcripts;
[0040] (13) Using PASA software, the candidate transcripts are further assembled, and the remaining incomplete transcripts are further assembled into complete transcripts to obtain a set of non-redundant transcripts;
[0041] (14) Based on the above set of non-redundant transcripts, genemaker-ST was used to obtain the gene structure of the species to be predicted.
[0042] The second aspect of the present invention provides three generations of full-length transcripts for whole genome gene structure prediction, and the three generations of full-length transcripts for whole genome gene structure prediction are obtained by the method provided by the present invention.
[0043] The third-generation full-length transcripts for whole-genome gene structure prediction provided by the present invention also include those obtained by assembling non-full-length transcripts by PASA.
[0044] In the three generations of full-length transcripts for whole genome gene structure prediction provided by the present invention, if the three generations of full-length transcripts correspond to the same splicing sites on the genome, the three generations of full-length transcripts are merged to remove duplicated three generations of full-length transcripts.
[0045] The present invention also claims to protect the method provided by the present invention and the application of the three-generation full-length transcript provided by the present invention in the prediction of alternative splicing.
[0046] The beneficial effects of the present invention include at least:
[0047] (1) The method provided by the present invention is applicable to the method of predicting the whole genome gene structure by using the universal third-generation full-length transcriptome sequencing data of different species such as animals and plants;
[0048] (2) The method provided by the present invention can be used to identify genes in animal and plant genomes at the whole genome level using third-generation full-length transcriptome sequencing data, which is beneficial to improving the integrity and accuracy of gene prediction;
[0049] (3) The method provided by the present invention obtains the splicing sites of three types of data (second-generation transcriptome data, homologous species data, and third-generation transcriptome data), identifies and corrects the splicing sites in the third-generation transcriptome data, obtains highly reliable transcripts, and assists in gene prediction (including variable splicing prediction) based on the highly reliable transcript information, thereby improving the reliability of gene prediction. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] Figure 1 It is a specific roadmap of the present invention.
[0051] Figure 2 This is an example diagram of Arabidopsis gene prediction assisted by three-generation full-length transcriptome data in Example 1 of the present invention.
[0052] Figure 3 This is an example diagram of zebrafish gene prediction assisted by three-generation full-length transcriptome data in Example 2 of the present invention. DETAILED DESCRIPTION
[0053] The preferred embodiments of the present invention will be described in detail below in conjunction with examples. It should be understood that the following examples are provided only for the purpose of illustration and are not intended to limit the scope of the present invention. Those skilled in the art may make various modifications and substitutions to the present invention without departing from the purpose and spirit of the present invention.
[0054] The experimental methods used in the following examples are conventional methods unless otherwise specified. The materials, reagents, etc. used in the following examples are all commercially available unless otherwise specified.
[0055] Example 1 Using Nanopore third-generation full-length transcriptome data to assist in the prediction of Arabidopsis genome gene structure
[0056] This example uses the gene prediction of the Arabidopsis genome as the analysis object, and provides a method for using the third-generation full-length transcriptome sequencing data to assist in the prediction of the whole genome gene structure, as follows:
[0057] 1. Second and third generation transcriptome data sequencing
[0058] The sequencing materials came from mixed samples of Arabidopsis roots, stems, leaves, flowers and seeds. After extracting mRNA, they were sequenced on the second-generation Illumina and third-generation Nanopore platforms to obtain the original offline data (reads).
[0059] 2. Comparison of transcriptome data and prediction of splicing sites based on gene information of homologous species
[0060] (1) Use HISAT2 to align the second-generation transcriptome raw data downloaded from the machine to the Arabidopsis genome and obtain a bam file;
[0061] (2) Use GMAP software to align the downloaded three-generation transcriptome data to the Arabidopsis genome and output the alignment results in gff3 format;
[0062] (3) Download the genome and gff3 file of the homologous species of Arabidopsis thaliana, Brassica rapa (NCBI No. GCF_000309985.1), then use GeMoMa software to predict the gene information of Arabidopsis thaliana with the help of the Brassica rapa gene information, output the file in gff3 format, and extract the splicing site information.
[0063] 3. Correction of intron splicing sites predicted by third-generation full-length transcriptome sequencing data
[0064] (1) Use the groupby command in the bedtools tool to convert the bam format generated by the second-generation sequencing alignment into the bed format, extract the intron splicing position, and use the Linux command sort and uniq to count the number of reads supporting each intron splicing position, and retain those with a support greater than 2; at the same time, count the number of splicing sites located in introns and exons, and ensure that the ratio of splicing sites located in introns and exons is less than 0.5, so as to obtain high-reliability intron splicing sites;
[0065] (2) Screen out the original intron splicing sites of the third-generation transcriptome data that conform to the GT-AG and GC-AG rules and have at least 5 reads to support the intron splicing sites as candidate splicing sites with higher reliability in the third-generation transcriptome prediction, and merge them with the high-reliability intron splicing sites into a set of non-redundant high-reliability intron splicing sites;
[0066] (3) The above non-redundant high-reliability intron splicing site was extended by 20 bases to the left and right, and the intersection with the candidate splicing sites in the gene structure predicted by the third-generation transcriptome was taken using bedtools. The candidate splicing sites in the gene structure predicted by the third-generation transcriptome that fell in the intersection were replaced by the non-redundant high-reliability intron splicing site, thereby completing the correction of the intron splicing site predicted by the third-generation full-length transcriptome sequencing data.
[0067] 4. Gene structure prediction method after obtaining high-reliability transcripts
[0068] (1) After the splicing sites predicted by the aforementioned three-generation transcriptome sequencing data are corrected, if two transcripts correspond to the same splicing sites on the genome, the transcripts are merged to finally obtain a set of non-redundant and highly accurate candidate transcripts;
[0069] (2) using PASA software, further assembling the candidate transcripts, and further assembling the remaining incomplete transcripts into complete transcripts to obtain a set of non-redundant transcripts;
[0070] (3) Based on the above transcripts, genemaker-ST was used to obtain the gene structure at the whole genome level, and IGV was used to observe the consistency of the gene structure with the reference gene structure.
[0071] The results are as follows: A total of 30Gb of data was downloaded from the third-generation sequencing data. This method predicted 50,531 transcripts, and the consistency with the reference gene set was 72.04% after evaluation, indicating that most genes were predicted, which plays an important role in gene-assisted prediction. Figure 2 The prediction result of one of the gene structures is given, and IGV is used to observe the consistency between the prediction result and the reference. It can be seen that among the three transcripts of the gene predicted by the present invention, one is completely consistent with the reference gene structure, and the other is a newly predicted transcript, which proves that the newly developed method of the present invention can not only ensure the accuracy of the predicted gene structure but also predict more variable splicing.
[0072] Example 2 Using Nanopore third-generation full-length transcriptome data to assist in the prediction of zebrafish genome gene structure
[0073] This example uses the same method as Example 1 to predict the whole genome gene structure of zebrafish using the 30Gb third-generation ONT full-length transcriptome. The results show that the consistency with the reference gene set reached 80.28%. Figure 3 This is the information from one of the transcripts.
[0074] Although the present invention has been described in detail above with general descriptions and specific embodiments, it is obvious to those skilled in the art that some modifications or improvements may be made thereto based on the present invention. Therefore, these modifications or improvements made without departing from the spirit of the present invention all fall within the scope of protection claimed by the present invention.
Claims
1. Methods for predicting gene structure at the whole genome level, It is characterized in that include: Predict the gene structure of species based on transcriptome data obtained from the second-generation Illumina platform and predict the gene structure of the species based on gene information data of homologous species, obtain intron splicing sites with a read number greater than 2, and ensure that the ratio of intron splicing sites located in introns to those located in exons is greater than 0.5 as the intersection; Based on the full-length transcriptome data obtained from the third-generation Nanopore platform, the gene structure of the species is predicted, and the spliced introns in the third-generation full-length transcriptome data are obtained to start with GT and end with AG, or the spliced introns start with GC and end with AG; the intron splicing sites with a read number greater than 5 are used as the intron splicing sites in the third-generation full-length transcriptome data of the species; The intersection is merged with the intron splicing sites in the three-generation full-length transcriptome data of the species; the merged intron splicing sites are extended to the left and right by 10-30 bases, and the intersection is taken with the intron splicing sites obtained by predicting the species gene structure from the three-generation full-length transcriptome data of the species, and the intron splicing sites falling in the intersection are replaced with the merged intron splicing sites to obtain transcripts; After the transcripts were compared with the genome information of the species, genemaker-ST was used to predict the gene structure at the whole genome level.
2. Application of the method according to claim 1 in alternative splicing prediction.
