ENVirT: inference of ecological characteristics of viruses from metagenomic data

Duleepa Jayasundara, Damayanthi Herath, Damith A. Senanayake, Isaam Saeed, Cheng-Yu Yang, Yuan Sun, Bill C. Chang, Sen-Lin Tang, Saman K. Halgamuge

Abstract

Metagenomic analysis of viral communities, or "viromes," is often hindered by a reliance on reference databases, which are inherently biased toward known sequences. We present ENVirT (Ecological Characteristics of Viruses from Metagenomic Data), a database-independent algorithm that simultaneously estimates viral richness and average genome length. By employing a 4-dimensional Genetic Algorithm with a specialized niching strategy, ENVirT achieves significantly higher accuracy than existing tools, particularly in characterizing novel viral "dark matter."

Introduction

Viruses represent the most abundant biological entities on Earth, yet the vast majority remain uncatalogued. In metagenomic studies, researchers aim to determine the richness (number of distinct types) and average genome length of viral communities. Traditional estimation methods, such as PHACCS or CatchAll, suffer from a "database bias": they require an initial estimate of average genome length, typically derived from known sequences. This circular dependency leads to significant inaccuracies when studying environments dominated by novel viruses.

Methodology

ENVirT addresses the limitations of database-dependent methods by treating average genome length as an unknown variable to be optimized alongside richness.

The Model

The algorithm is based on the Lander-Waterman model, which relates DNA fragment (read) overlaps to the formation of contigs. ENVirT utilizes the "contig spectrum"—the distribution of reads across contigs of varying sizes—as its primary input.

Optimization and Niching

The estimation process involves minimizing a non-linear error function comparing observed and theoretical contig spectra. To navigate the complex search space and avoid local minima, ENVirT utilizes a 4-dimensional Genetic Algorithm (GA). A critical component of this approach is the "niching" strategy, which partitions the search space into sub-regions. This ensures a comprehensive exploration of potential richness and genome length values, preventing premature convergence on sub-optimal solutions.

Comparison of parameter estimation between ENVirT and a standard Genetic Algorithm without niching.

Results

ENVirT was benchmarked using over 200 simulated datasets and 20 experimental metaviromes from diverse environments, including freshwater lakes, the human gut, and fermented foods.

Accuracy and Robustness

On simulated data, ENVirT demonstrated an average estimation error of only 0.91%. In comparison, the industry-standard PHACCS tool exhibited an error of 585.87%, even when provided with the true average genome length. ENVirT maintained stability across a wide range of community complexities, whereas previous tools often failed to produce estimates for highly complex samples.

Correlation between actual and estimated genome lengths (bp) across simulated datasets.

Real-World Application

In practical applications, ENVirT revealed previously "invisible" ecological trends. For instance, in the Feitsui Reservoir, the tool identified an increase in average viral genome length following a typhoon, likely indicating the influx of terrestrial viruses. In samples from the human gut and sauerkraut, ENVirT estimated larger average genome lengths than traditional methods, suggesting the presence of large, uncatalogued viruses like Mimiviruses.

Discussion

While ENVirT represents a significant advancement, it currently provides a single average genome length estimate, which may not fully capture multi-modal distributions in highly diverse communities. Additionally, estimation accuracy for genomes exceeding 100 kbp shows a slight decline, suggesting a need for improved handling of long-read assembly data. Despite these limitations, ENVirT's database-independent approach and computational efficiency make it a powerful tool for exploring the viral unknown.

Reference: Jayasundara et al. (2019). ENVirT: inference of ecological characteristics of viruses from metagenomic data. BMC Bioinformatics. DOI: 10.1186/s12859-018-2398-5

Summary updated

Abstract

Background Estimating the parameters that describe the ecology of viruses,particularly those that are novel, can be made possible using metagenomic approaches. However, the best-performing existing methods require databases to first estimate an average genome length of a viral community before being able to estimate other parameters, such as viral richness. Although this approach has been widely used, it can adversely skew results since the majority of viruses are yet to be catalogued in databases. Results In this paper, we present ENVirT, a method for estimating the richness of novel viral mixtures, and for the first time we also show that it is possible to simultaneously estimate the average genome length without a priori information. This is shown to be a significant improvement over database-dependent methods, since we can now robustly analyze samples that may include novel viral types under-represented in current databases. We demonstrate that the viral richness estimates produced by ENVirT are several orders of magnitude higher in accuracy than the estimates produced by existing methods named PHACCS and CatchAll when benchmarked against simulated data. We repeated the analysis of 20 metavirome samples using ENVirT, which produced results in close agreement with complementary in virto analyses. Conclusions These insights were previously not captured by existing computational methods. As such, ENVirT is shown to be an essential tool for enhancing our understanding of novel viral populations.
BibTeX
@article{jayasundara2019envirt,
  author = {Duleepa Jayasundara and Damayanthi Herath and Damith A. Senanayake and Isaam Saeed and Cheng-Yu Yang and Yuan Sun and Bill C. Chang and Sen-Lin Tang and Saman K. Halgamuge},
  title = {{ENVirT: inference of ecological characteristics of viruses from metagenomic data}},
  journal = {BMC Bioinformatics},
  year = {2019},
  doi = {10.1186/s12859-018-2398-5},
}
Figure