PlasGUN: gene prediction in plasmid metagenomic short reads using deep learning.
Ontology highlight
ABSTRACT: Summary
We present the first tool of gene prediction, PlasGUN, for plasmid metagenomic short-read data. The tool, developed based on deep learning algorithm of multiple input Convolutional Neural Network, demonstrates much better performance when tested on a benchmark dataset of artificial short reads and presents more reliable results for real plasmid metagenomic data than traditional gene prediction tools designed primarily for chromosome-derived short reads.Availability and implementation
The PlasGUN software is available at http://cqb.pku.edu.cn/ZhuLab/PlasGUN/ or https://github.com/zhenchengfang/PlasGUN/.Supplementary information
Supplementary data are available at Bioinformatics online.
SUBMITTER: Fang Z
PROVIDER: S-EPMC7214025 | biostudies-literature | 2020 May
REPOSITORIES: biostudies-literature
Publications
PlasGUN: gene prediction in plasmid metagenomic short reads using deep learning.
Fang Zhencheng Z Tan Jie J Wu Shufang S Li Mo M Wang Chunhui C Liu Yongchu Y Zhu Huaiqiu H
Bioinformatics (Oxford, England) 20200501 10
<h4>Summary</h4>We present the first tool of gene prediction, PlasGUN, for plasmid metagenomic short-read data. The tool, developed based on deep learning algorithm of multiple input Convolutional Neural Network, demonstrates much better performance when tested on a benchmark dataset of artificial short reads and presents more reliable results for real plasmid metagenomic data than traditional gene prediction tools designed primarily for chromosome-derived short reads.<h4>Availability and implem ...[more]
Similar Datasets
Project description:BACKGROUND: One of the major open challenges in next generation sequencing (NGS) is the accurate identification of structural variants such as insertions and deletions (indels). Current methods for indel calling assign scores to different types of evidence or counter-evidence for the presence of an indel, such as the number of split read alignments spanning the boundaries of a deletion candidate or reads that map within a putative deletion. Candidates with a score above a manually defined threshold are then predicted to be true indels. As a consequence, structural variants detected in this manner contain many false positives. RESULTS: Here, we present a machine learning based method which is able to discover and distinguish true from false indel candidates in order to reduce the false positive rate. Our method identifies indel candidates using a discriminative classifier based on features of split read alignment profiles and trained on true and false indel candidates that were validated by Sanger sequencing. We demonstrate the usefulness of our method with paired-end Illumina reads from 80 genomes of the first phase of the 1001 Genomes Project ( http://www.1001genomes.org) in Arabidopsis thaliana. CONCLUSION: In this work we show that indel classification is a necessary step to reduce the number of false positive candidates. We demonstrate that missing classification may lead to spurious biological interpretations. The software is available at: http://agkb.is.tuebingen.mpg.de/Forschung/SV-M/.
| S-EPMC3614465 | biostudies-other
Project description:Next generation sequencing technologies used in metagenomics yield numerous sequencing fragments which come from thousands of different species. Accurately identifying genes from metagenomics fragments is one of the most fundamental issues in metagenomics. In this article, by fusing multifeatures (i.e., monocodon usage, monoamino acid usage, ORF length coverage, and Z-curve features) and using deep stacking networks learning model, we present a novel method (called Meta-MFDL) to predict the metagenomic genes. The results with 10 CV and independent tests show that Meta-MFDL is a powerful tool for identifying genes from metagenomic fragments.
| S-EPMC5698827 | biostudies-literature
Project description:Background FragGeneScan is currently the most accurate and popular tool for gene prediction in short and error-prone reads, but its execution speed is insufficient for use on larger data sets. The parallelization which should have addressed this is inefficient. Its alternative implementation FragGeneScan+ is faster, but introduced a number of bugs related to memory management, race conditions and even output accuracy. Results This paper introduces FragGeneScanRs, a faster Rust implementation of the FragGeneScan gene prediction model. Its command line interface is backward compatible and adds extra features for more flexible usage. Its output is equivalent to the original FragGeneScan implementation. Conclusions Compared to the current C implementation, shotgun metagenomic reads are processed up to 22 times faster using a single thread, with better scaling for multithreaded execution. The Rust code of FragGeneScanRs is freely available from GitHub under the GPL-3.0 license with instructions for installation, usage and other documentation (https://github.com/unipept/FragGeneScanRs). Supplementary Information The online version contains supplementary material available at 10.1186/s12859-022-04736-5.
| S-EPMC9148508 | biostudies-literature
Project description:UnlabelledBackgroundThe metagenomics approach allows the simultaneous sequencing of all genomes in an environmental sample. This results in high complexity datasets, where in addition to repeats and sequencing errors, the number of genomes and their abundance ratios are unknown. Recently developed next-generation sequencing (NGS) technologies significantly improve the sequencing efficiency and cost. On the other hand, they result in shorter reads, which makes the separation of reads from different species harder. Among the existing computational tools for metagenomic analysis, there are similarity-based methods that use reference databases to align reads and composition-based methods that use composition patterns (i.e., frequencies of short words or l-mers) to cluster reads. Similarity-based methods are unable to classify reads from unknown species without close references (which constitute the majority of reads). Since composition patterns are preserved only in significantly large fragments, composition-based tools cannot be used for very short reads, which becomes a significant limitation with the development of NGS. A recently proposed algorithm, AbundanceBin, introduced another method that bins reads based on predicted abundances of the genomes sequenced. However, it does not separate reads from genomes of similar abundance levels.ResultsIn this work, we present a two-phase heuristic algorithm for separating short paired-end reads from different genomes in a metagenomic dataset. We use the observation that most of the l-mers belong to unique genomes when l is sufficiently large. The first phase of the algorithm results in clusters of l-mers each of which belongs to one genome. During the second phase, clusters are merged based on l-mer repeat information. These final clusters are used to assign reads. The algorithm could handle very short reads and sequencing errors. It is initially designed for genomes with similar abundance levels and then extended to handle arbitrary abundance ratios. The software can be download for free at http://www.cs.ucr.edu/∼tanaseio/toss.htm.ConclusionsOur tests on a large number of simulated metagenomic datasets concerning species at various phylogenetic distances demonstrate that genomes can be separated if the number of common repeats is smaller than the number of genome-specific repeats. For such genomes, our method can separate NGS reads with a high precision and sensitivity.
| S-EPMC3537596 | biostudies-literature
Project description:MotivationMetagenomics is a recent field of biology that studies microbial communities by analyzing their genomic content directly sequenced from the environment. A metagenomic dataset consists of many short DNA or RNA fragments called reads. One interesting problem in metagenomic data analysis is the discovery of the taxonomic composition of a given dataset. A simple method for this task, called the Lowest Common Ancestor (LCA), is employed in state-of-the-art computational tools for metagenomic data analysis of very short reads (about 100 bp). However LCA has two main drawbacks: it possibly assigns many reads to high taxonomic ranks and it discards a high number of reads.ResultsWe present MTR, a new method for tackling these drawbacks using clustering at Multiple Taxonomic Ranks. Unlike LCA, which processes the reads one-by-one, MTR exploits information shared by reads. Specifically, MTR consists of two main phases. First, for each taxonomic rank, a collection of potential clusters of reads is generated, and each potential cluster is associated to a taxon at that rank. Next, a small number of clusters is selected at each rank using a combinatorial optimization algorithm. The effectiveness of the resulting method is tested on a large number of simulated and real-life metagenomes. Results of experiments show that MTR improves on LCA by discarding a significantly smaller number of reads and by assigning much more reads at lower taxonomic ranks. Moreover, MTR provides a more faithful taxonomic characterization of the metagenome population distribution.AvailabilityMatlab and C++ source codes of the method available at http://cs.ru.nl/gori/software/MTR.tar.gz.
| S-EPMC3018814 | biostudies-literature
Project description:Antimicrobial peptides (AMPs) are a valuable source of antimicrobial agents and a potential solution to the multi-drug resistance problem. In particular, short-length AMPs have been shown to have enhanced antimicrobial activities, higher stability, and lower toxicity to human cells. We present a short-length (≤30 aa) AMP prediction method, Deep-AmPEP30, developed based on an optimal feature set of PseKRAAC reduced amino acids composition and convolutional neural network. On a balanced benchmark dataset of 188 samples, Deep-AmPEP30 yields an improved performance of 77% in accuracy, 85% in the area under the receiver operating characteristic curve (AUC-ROC), and 85% in area under the precision-recall curve (AUC-PR) over existing machine learning-based methods. To demonstrate its power, we screened the genome sequence of Candida glabrata-a gut commensal fungus expected to interact with and/or inhibit other microbes in the gut-for potential AMPs and identified a peptide of 20 aa (P3, FWELWKFLKSLWSIFPRRRP) with strong anti-bacteria activity against Bacillus subtilis and Vibrio parahaemolyticus. The potency of the peptide is remarkably comparable to that of ampicillin. Therefore, Deep-AmPEP30 is a promising prediction tool to identify short-length AMPs from genomic sequences for drug discovery. Our method is available at https://cbbio.cis.um.edu.mo/AxPEP for both individual sequence prediction and genome screening for AMPs.
| S-EPMC7256447 | biostudies-literature
Project description:BackgroundNext-generation sequencing (NGS) has greatly facilitated metagenomic analysis but also raised new challenges for metagenomic DNA sequence assembly, owing to its high-throughput nature and extremely short reads generated by sequencers such as Illumina. To date, how to generate a high-quality draft assembly for metagenomic sequencing projects has not been fully addressed.ResultsWe conducted a comprehensive assessment on state-of-the-art de novo assemblers and revealed that the performance of each assembler depends critically on the sequencing depth. To address this problem, we developed a pipeline named InteMAP to integrate three assemblers, ABySS, IDBA-UD and CABOG, which were found to complement each other in assembling metagenomic sequences. Making a decision of which assembling approaches to use according to the sequencing coverage estimation algorithm for each short read, the pipeline presents an automatic platform suitable to assemble real metagenomic NGS data with uneven coverage distribution of sequencing depth. By comparing the performance of InteMAP with current assemblers on both synthetic and real NGS metagenomic data, we demonstrated that InteMAP achieves better performance with a longer total contig length and higher contiguity, and contains more genes than others.ConclusionsWe developed a de novo pipeline, named InteMAP, that integrates existing tools for metagenomics assembly. The pipeline outperforms previous assembly methods on metagenomic assembly by providing a longer total contig length, a higher contiguity and covering more genes. InteMAP, therefore, could potentially be a useful tool for the research community of metagenomics.
| S-EPMC4545859 | biostudies-literature
Project description:BackgroundThe recent development of metagenomic sequencing makes it possible to massively sequence microbial genomes including viral genomes without the need for laboratory culture. Existing reference-based and gene homology-based methods are not efficient in identifying unknown viruses or short viral sequences from metagenomic data.MethodsHere we developed a reference-free and alignment-free machine learning method, DeepVirFinder, for identifying viral sequences in metagenomic data using deep learning.ResultsTrained based on sequences from viral RefSeq discovered before May 2015, and evaluated on those discovered after that date, DeepVirFinder outperformed the state-of-the-art method VirFinder at all contig lengths, achieving AUROC 0.93, 0.95, 0.97, and 0.98 for 300, 500, 1000, and 3000 bp sequences respectively. Enlarging the training data with additional millions of purified viral sequences from metavirome samples further improved the accuracy for identifying virus groups that are under-represented. Applying DeepVirFinder to real human gut metagenomic samples, we identified 51,138 viral sequences belonging to 175 bins in patients with colorectal carcinoma (CRC). Ten bins were found associated with the cancer status, suggesting viruses may play important roles in CRC.ConclusionsPowered by deep learning and high throughput sequencing metagenomic data, DeepVirFinder significantly improved the accuracy of viral identification and will assist the study of viruses in the era of metagenomics.
| S-EPMC8172088 | biostudies-literature
Project description:Identifying therapeutic genes is crucial for developing treatments targeting genetic causes of diseases, but experimental trials are costly and time-consuming. Although many deep learning approaches aim to identify biomarker genes, predicting therapeutic target genes remains challenging due to the limited number of known targets. To address this, we propose HIT (Hypergraph Interaction Transformer), a deep hypergraph representation learning model that identifies a gene's therapeutic potential, biomarker status, or lack of association with diseases. HIT uses hypergraph structures of genes, ontologies, diseases, and phenotypes, employing attention-based learning to capture complex relationships. Experiments demonstrate HIT's state-of-the-art performance, explainability, and ability to identify novel therapeutic targets.
| S-EPMC11752618 | biostudies-literature
Project description:BACKGROUND: Investigation of metagenomes provides greater insight into uncultured microbial communities. The improvement in sequencing technology, which yields a large amount of sequence data, has led to major breakthroughs in the field. However, at present, taxonomic binning tools for metagenomes discard 30-40% of Sanger sequencing data due to the stringency of BLAST cut-offs. In an attempt to provide a comprehensive overview of metagenomic data, we re-analyzed the discarded metagenomes by using less stringent cut-offs. Additionally, we introduced a new criterion, namely, the evolutionary conservation of adjacency between neighboring genes. To evaluate the feasibility of our approach, we re-analyzed discarded contigs and singletons from several environments with different levels of complexity. We also compared the consistency between our taxonomic binning and those reported in the original studies. RESULTS: Among the discarded data, we found that 23.7 ± 3.9% of singletons and 14.1 ± 1.0% of contigs were assigned to taxa. The recovery rates for singletons were higher than those for contigs. The Pearson correlation coefficient revealed a high degree of similarity (0.94 ± 0.03 at the phylum rank and 0.80 ± 0.11 at the family rank) between the proposed taxonomic binning approach and those reported in original studies. In addition, an evaluation using simulated data demonstrated the reliability of the proposed approach. CONCLUSIONS: Our findings suggest that taking account of conserved neighboring gene adjacency improves taxonomic assignment when analyzing metagenomes using Sanger sequencing. In other words, utilizing the conserved gene order as a criterion will reduce the amount of data discarded when analyzing metagenomes.
| S-EPMC3098102 | biostudies-literature