Platon: identification and characterization of bacterial plasmid contigs in short-read draft assemblies exploiting protein-sequence-based replicon distribution scores

Oliver Schwengers, Patrick Barth, Linda Falgenhauer, Torsten Hain, Trinad Chakraborty, Alexander Goesmann

TL;DR

This paper introduces Platon, a new bioinformatics tool that accurately identifies and characterizes bacterial plasmid DNA fragments within draft genome assemblies. By leveraging the distinct distribution patterns of bacterial proteins between chromosomes and plasmids, Platon offers a high-throughput, taxon-independent solution that outperforms existing methods in both accuracy and balanced classification.

What Problem Does This Paper Address?

The increasing volume of bacterial whole-genome sequencing data presents a challenge: distinguishing plasmid DNA from chromosomal DNA within draft assemblies. Plasmids are crucial for bacterial adaptation, carrying genes for antibiotic resistance and virulence, making their accurate identification vital for understanding bacterial evolution and combating infectious diseases. Existing tools often struggle with high-throughput analysis, require taxon-specific databases, or provide biased classifications (either favoring sensitivity or specificity). This paper addresses the need for a robust, scalable, and taxon-independent method to reliably separate plasmid contigs from chromosomal ones in draft assemblies.

Methods

Platon employs a novel approach based on "replicon distribution scores" (RDS). This metric quantifies the bias of protein-coding genes' presence on either plasmids or chromosomes. The core idea is that essential housekeeping genes are typically chromosome-bound, while genes conferring adaptive advantages (like antibiotic resistance) are more frequently found on plasmids.

The workflow involves:

  1. Marker Protein Sequence (MPS) Database Construction: A large database of bacterial protein sequences is created and their distribution across known plasmid and chromosomal replicons is analyzed.
  2. Replicon Distribution Score (RDS) Calculation: For each protein in a contig, an RDS is calculated based on its alignment hit frequencies to plasmid and chromosome sequences, incorporating statistical significance.
  3. Higher-Level Contig Characterization: Platon performs additional analyses, including testing for circularization, detecting incompatibility groups, identifying replication and mobilization proteins, and searching for oriT sequences.
  4. Classification: Contigs are classified as chromosomal or plasmid based on their mean RDS and the results of the higher-level characterizations, using defined sensitivity, specificity, and conservative thresholds.

Platon is implemented as a Python 3 command-line application, designed for high-throughput analysis and easy integration into larger pipelines.

Datasets

  • Training Data:
    • UniProt's UniRef90 protein clusters (n=69,803,841)
    • NCBI RefSeq database (complete bacterial genomes)
    • NCBI Genomes plasmid section (complete plasmid sequences)
  • Benchmark Datasets:
    • Simulated Short-Read Data: Generated from all complete bacterial NCBI RefSeq genomes (n=13,930) using ART simulator, then assembled with SPAdes. This dataset aimed for taxonomic diversity.
  • Real Data: 24 multidrug-resistant Escherichia coli isolates sequenced using Illumina MiSeq and Oxford Nanopore GridION. Hybrid assemblies were used to establish ground truth for 21 isolates.

Key Insights and Ideas

  • Protein Distribution as a Plasmid Signature: The fundamental insight is that the differential distribution of protein-coding genes between bacterial chromosomes and plasmids provides a powerful, inherent signature for identifying plasmid contigs.
  • RDS as a Robust Metric: The developed Replicon Distribution Score (RDS) effectively captures this bias, offering a more nuanced classification than simple presence/absence of specific motifs.
  • Heuristics Enhance Performance: Combining the RDS with higher-level contig characterizations (like circularization and presence of specific protein families) significantly improves classification accuracy and sensitivity, addressing limitations of the RDS alone.
  • Taxon-Independent Approach: Platon's design, relying on general protein distribution patterns rather than specific taxon databases, makes it broadly applicable across diverse bacterial species.

Key Results

  • High Accuracy: Platon achieved an accuracy of 96.6% on simulated data and 91.5% on real E. coli isolates.
  • Improved Performance over Competitors:
    • Simulated Data: Platon outperformed PlasFlow with higher accuracy (97.6% vs. 87.1%), specificity (99.3% vs. 88.3%), and significantly fewer false positives (5,277 vs. 88,712). Platon also showed better F1 score (0.828 vs. 0.465) and MCC (0.818 vs. 0.440).
  • Real E. coli Data: Platon achieved higher accuracy (0.915 vs. 0.841 for PlasmidFinder, 0.901 for PlaScope), sensitivity (0.699 vs. 0.223 for PlasmidFinder, 0.684 for PlaScope), and better F1 score (0.758 vs. 0.349 for PlasmidFinder, 0.725 for PlaScope) and MCC.
  • Balanced Classification: Platon demonstrated more balanced predictions (higher F1 scores) due to a lower false positive rate compared to other tools.
  • Scalability: The tool is designed for high-throughput analysis and is implemented as a command-line application.

Limitations and Future Work

  • Limitations:
    • Small contigs with few protein-coding genes, or contigs encoding proteins with equal distribution across chromosomes and plasmids, can be harder to classify correctly.
  • Contigs larger than 500 kbp are heuristically assigned to the chromosome to optimize runtime, which could potentially misclassify very large plasmids (though this was found to be rare, affecting only 0.19% of plasmid contigs in simulations).
  • Future Work:
    • The authors envision adding new higher-level contig characterizations.
  • Further enhancements to the applied heuristics are planned.
  • Platon is designed for integration into larger analysis pipelines and as an upfront tool for subsequent plasmid-specific analyses.

Summary updated

Abstract

ABSTRACT Plasmids are extrachromosomal genetic elements replicating independently of the chromosome which play a vital role in the environmental adaptation of bacteria. Due to potential mobilization or conjugation capabilities, plasmids are important genetic vehicles for antimicrobial resistance genes and virulence factors with huge and increasing clinical implications. They are therefore subject to large genomic studies within the scientific community worldwide. As a result of rapidly improving next generation sequencing methods, the amount of sequenced bacterial genomes is constantly increasing, in turn raising the need for specialized tools to (i) extract plasmid sequences from draft assemblies, (ii) derive their origin and distribution, and (iii) further investigate their genetic repertoire. Recently, several bioinformatic methods and tools have emerged to tackle this issue; however, a combination of both high sensitivity and specificity in plasmid sequence identification is rarely achieved in a taxon-independent manner. In addition, many software tools are not appropriate for large high-throughput analyses or cannot be included into existing software pipelines due to their technical design or software implementation. In this study, we investigated differences in the replicon distributions of protein-coding genes on a large scale as a new approach to distinguish plasmid-borne from chromosome-borne contigs. We defined and computed statistical discrimination thresholds for a new metric: the replicon distribution score (RDS) which achieved an accuracy of 96.6%. The final performance was further improved by the combination of the RDS metric with heuristics exploiting several plasmid specific higher-level contig characterizations. We implemented this workflow in a new high-throughput taxon-independent bioinformatics software tool called Platon for the recruitment and characterization of plasmid-borne contigs from short-read draft assemblies. Compared to PlasFlow, Platon achieved a higher accuracy (97.5%) and more balanced predictions (F1=82.6%) tested on a broad range of bacterial taxa and better or equal performance against the targeted tools PlasmidFinder and PlaScope on sequenced E. coli isolates. Platon is available at: platon.computational.bio Data Summary Platon was developed as a Python 3 command line application for Linux. The complete source code and documentation is available on GitHub under a GPL3 license: https://github.com/oschwengers/platon and platon.computational.bio . All database versions are hosted at Zenodo: DOI 10.5281/zenodo.3349651. Platon is available via bioconda package platon Platon is available via PyPI package cb-platon Bacterial representative sequences for UniProt’s UniRef90 protein clusters, complete bacterial genome sequences from the NCBI RefSeq database, complete plasmid sequences from the NCBI genomes plasmid section, created artificial contigs, RDS threshold metrics and raw protein replicon hit counts used to create and evaluate the marker protein sequence database are hosted at Zenodo: DOI 10.5281/zenodo.3759169 24 Escherichia coli isolates sequenced with short read (Illumina MiSeq) and long read sequencing technologies (Oxford Nanopore Technology GridION platform) used for real data benchmarks are available under the following NCBI BioProjects: PRJNA505407, PRJNA387731 Impact Statement Plasmids play a vital role in the spread of antibiotic resistance and pathogenicity genes. The increasing numbers of clinical outbreaks involving resistant pathogens worldwide pushed the scientific community to increase their efforts to comprehensively investigate bacterial genomes. Due to the maturation of next-generation sequencing technologies, nowadays entire bacterial genomes including plasmids are sequenced in huge scale. To analyze draft assemblies, a mandatory first step is to separate plasmid from chromosome contigs. Recently, many bioinformatic tools have emerged to tackle this issue. Unfortunately, several tools are implemented only as interactive or web-based tools disabling them for necessary high-throughput analysis of large data sets. Other tools providing such a high-throughput implementation however often come with certain drawbacks, e . g . providing taxon-specific databases only, not providing actionable, i . e . true binary classification or achieving biased classification performances towards either sensitivity or specificity. Here, we introduce the tool Platon implementing a new replicon distribution-based approach combined with higher-level contig characterizations to address the aforementioned issues. In addition to the plasmid detection within draft assemblies, Platon provides the user with valuable information on certain higher-level contig characterizations. We show that Platon provides a balanced classification performance as well as a scalable implementation for high-throughput analyses. We therefore consider Platon to be a powerful, species-independent and flexible tool to scan large amounts of bacterial whole-genome sequencing data for their plasmid content.
BibTeX
@inproceedings{schwengers2020platon,
  author = {Oliver Schwengers and Patrick Barth and Linda Falgenhauer and Torsten Hain and Trinad Chakraborty and Alexander Goesmann},
  title = {{Platon: identification and characterization of bacterial plasmid contigs in short-read draft assemblies exploiting protein-sequence-based replicon distribution scores}},
  booktitle = {Microbial},
  year = {2020},
  doi = {10.1099/mgen.0.000398},
}
Figure