Viabin: A novel method for Overlapped Binning of Metagenomic Contigs using ZINB-autodecoder and Assembly Graphs: Viabin: a metagenomic contigs binning tool
{"title":"Viabin: A novel method for Overlapped Binning of Metagenomic Contigs using ZINB-autodecoder and Assembly Graphs: Viabin: a metagenomic contigs binning tool","authors":"Fangyu Yu, Ziming Yan, Fu Liu","doi":"10.1145/3592686.3592742","DOIUrl":null,"url":null,"abstract":"Binning the metagenomic contigs into potential genomes is an essential step to investigate and characterize various genetic materials of different species and understand microbial populations. Existing tools have achieved considerable success with a variety of datasets, especially in long-length sequences, while precise classification of short contigs in complex metagenomics remains a challenging task. Since different species may have same sequences in their genomes, a contig sequence can be assigned to multiple species. However, current contig binning tools merely support single label binning. In other words, each contig is assigned to just one species. This article introduced Viabin which can refine the binning results using ZINB-autoencoder and assembly graphs. Viabin used Gaussian mixture model to preliminarily classify the unclassified contigs encoded by autoencoder based on ZINB. Subsequently, assembly graph and sequence feature were combined with the binning results of other tools to complete the final binning. More importantly, it had the capacity to assign contigs to numerous bins. Experimental 3 datasets demonstrated that Viabin could not only improve binning performance of short contig sequences but also support assignment of contigs to multiple bins.","PeriodicalId":366447,"journal":{"name":"Proceedings of the 2023 3rd International Conference on Bioinformatics and Intelligent Computing","volume":"237 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2023-02-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Proceedings of the 2023 3rd International Conference on Bioinformatics and Intelligent Computing","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1145/3592686.3592742","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
Binning the metagenomic contigs into potential genomes is an essential step to investigate and characterize various genetic materials of different species and understand microbial populations. Existing tools have achieved considerable success with a variety of datasets, especially in long-length sequences, while precise classification of short contigs in complex metagenomics remains a challenging task. Since different species may have same sequences in their genomes, a contig sequence can be assigned to multiple species. However, current contig binning tools merely support single label binning. In other words, each contig is assigned to just one species. This article introduced Viabin which can refine the binning results using ZINB-autoencoder and assembly graphs. Viabin used Gaussian mixture model to preliminarily classify the unclassified contigs encoded by autoencoder based on ZINB. Subsequently, assembly graph and sequence feature were combined with the binning results of other tools to complete the final binning. More importantly, it had the capacity to assign contigs to numerous bins. Experimental 3 datasets demonstrated that Viabin could not only improve binning performance of short contig sequences but also support assignment of contigs to multiple bins.