Bioinformatic analysis of shotgun metagenomic data for bacterial pathogen detection

Shotgun metagenomics is revolutionizing the field of microbiology by offering a comprehensive, high-throughput method for detecting pathogens. Unlike traditional culture-based techniques or PCR-based assays, which are often limited by prior knowledge of pathogens, shotgun metagenomics allows us to identify known and novel pathogens without the need for predefined hypotheses. In this blog post, we dive deep into how shotgun metagenomics works, its bioinformatic analysis, and how it is applied in bacterial pathogen detection, shedding light on its strengths and limitations.

What is Shotgun Metagenomics?

Metagenomics involves sequencing the total DNA from a microbial community without prior culturing or amplification of specific organisms. In shotgun metagenomics, DNA is randomly fragmented and sequenced, allowing for a broad spectrum of microbial diversity to be captured. This method is invaluable in uncovering microorganisms that may be hard to isolate using traditional techniques.

The Advantages of Shotgun Metagenomics Over Traditional Methods

Traditional pathogen detection methods like culturing or PCR rely on known sequences or organisms. If an organism is novel or unculturable, these methods can fail to detect it. Metagenomics, however, doesn’t require prior knowledge and can discover previously unrecognized pathogens. Additionally, it can identify co-infections, allowing researchers to detect multiple pathogens simultaneously—a feature traditional methods often miss.

Moreover, shotgun metagenomics provides insights into the genomic composition of microbial populations, giving researchers a more holistic view of the microbiome, including pathogenic and non-pathogenic organisms. This contrasts with PCR, which is limited to detecting specific pathogens defined by the primers used in the assay.

Key Metrics in Shotgun Metagenomic Analysis

To understand how metagenomics works in pathogen detection, it’s essential to grasp several core bioinformatic metrics:

• Sequencing Depth: This metric indicates how many times a particular base pair or feature in a genome is sequenced. Higher depth increases the likelihood of detecting low-abundance pathogens but also requires more sequencing resources.
• Sequencing Breadth: This metric represents the fraction of a gene or genome covered by at least one sequencing read. A higher breadth suggests a more comprehensive representation of the microbial community.
• Relative Abundance: This refers to the proportion of a particular sequence relative to the total sequencing effort. It gives insights into the commonality of different features within the sample.

Defining Species and Strains Using Average Nucleotide Identity (ANI)

In metagenomic studies, defining species and strains is crucial for pathogen identification. Average Nucleotide Identity (ANI) is a key metric used to establish genomic relationships between different organisms. ANI values above 95% typically indicate that organisms belong to the same species. The concept of strain and genomovars is further refined within species, with the 99.5% ANI threshold often being used to differentiate strains, especially in outbreak scenarios.

This advancement in ANI-based resolution is a significant improvement over traditional taxonomic classification, which often relied on morphological characteristics and less precise genetic markers.

Detection and Identification of Pathogens

Metagenomics allows for the detection of pathogens in a sample by using bioinformatic tools and methods that analyze the DNA sequences generated. These tools can match sequencing reads to reference genomes, profile taxa, and estimate the relative abundance of different organisms in the sample. Popular taxonomic profilers, such as MetaPhlAn, Kaiju, and Kraken, leverage ANI and k-mer matching to identify species and strains based on known sequences in reference databases.

More sophisticated approaches like read recruitment plots and TAD-80 (Truncated Average Depth) help estimate the relative abundance of pathogens with higher accuracy, offering more reliable pathogen detection, even in samples with low pathogen abundance.

Estimating Detection Limits and Pathogen Abundance

Understanding the limitations of shotgun metagenomics is crucial for interpreting results. Pathogen detection is dependent on sequencing depth and breadth, and it’s important to know the limit of detection (LOD)—the smallest fraction of a microbial community at which a pathogen can be reliably detected. By normalizing sequencing data with Genome Equivalents (GEQ) and using advanced metrics like TAD-80, researchers can more accurately estimate the fraction of the microbial community represented by a target pathogen.

For example, metagenomic sequencing efforts at 5 Gbp have been shown to detect pathogens like E. coli at a level as low as 0.01% of the total microbial community, demonstrating the high sensitivity of metagenomics.

Identifying the Etiological Agent of Disease

Metagenomics is not only useful for pathogen detection but also for identifying the cause of an outbreak or disease. By analyzing relative abundance, sequence clonality (ANIr), and the presence of virulence genes, metagenomics can pinpoint the pathogen responsible for a disease, even in complex environments like foodborne outbreaks. This approach has the potential to be faster and more accurate than traditional culture-based methods, which can be time-consuming and may fail to detect novel or fastidious pathogens.

Challenges in Metagenomic Pathogen Detection

Despite its potential, metagenomics faces several challenges:

• Cost: Sequencing can be expensive, especially for large datasets or low-abundance pathogens.
• Computational Resources: Analyzing metagenomic data requires specialized software and high-performance computing resources.
• Detection Limits: The sensitivity of metagenomics can be impacted by the abundance of the pathogen relative to the total microbial community.
• Live vs. Dead Cells: Distinguishing between viable pathogens and non-viable DNA fragments remains a challenge, although emerging techniques are helping to address this.

The Future of Shotgun Metagenomics in Pathogen Detection

While metagenomics has already proven its value in pathogen detection, there is still much to learn. Ongoing advancements in bioinformatic tools, sequencing technologies, and data interpretation are continually improving the accuracy and efficiency of pathogen detection in clinical, environmental, and food safety contexts. Additionally, the development of more cost-effective sequencing methods will make this powerful technology more accessible for routine use.

The future of pathogen detection is increasingly being shaped by metagenomics, offering an unprecedented ability to detect known and unknown pathogens with high accuracy and speed.

Conclusion

Shotgun metagenomics is transforming our approach to bacterial pathogen detection. By harnessing the power of sequencing technologies and bioinformatic tools, researchers can now identify pathogens with greater sensitivity and precision than ever before. While challenges remain, the potential of metagenomics to improve disease detection, surveillance, and outbreak management is immense. As technology continues to advance, metagenomics may soon become a cornerstone in our efforts to understand and control microbial threats.

FAQ on Shotgun Metagenomics

What is shotgun metagenomics, and why is it important for pathogen detection?

Shotgun metagenomics involves sequencing all the DNA present in a sample, providing a comprehensive view of the microbial community. This is crucial for pathogen detection because it allows for the identification of pathogens, including those that are difficult or impossible to culture. Metagenomics also enables the study of the pathogen within the context of its community, providing a better understanding of its behavior and impact. Furthermore, this approach can detect multiple pathogens in a single analysis, offering a more holistic view of infectious agents.

What are key metrics to consider when analyzing metagenomic data?

The key metrics in metagenomic analysis include sequencing depth, sequencing breadth, and sequencing effort. Sequencing depth is the number of reads covering a specific base pair, or the average depth across the feature, while sequencing breadth refers to the fraction of base pair positions covered by at least one read. Sequencing effort is the total amount of sequence data generated, often measured in gigabases (Gbp). The ratio between sequencing depth and sequencing effort generates relative abundance of the feature. Additionally, “genome equivalents” (GEQ), the average depth of single copy genes, is used to normalize the data for community size.

How do metagenomic studies support the concept of bacterial species as discrete units?

Metagenomic studies, combined with genomic analyses of bacterial isolates, have shown that bacteria primarily form sequence-discrete species, typically ranging from 95% to 100% average nucleotide identity (ANI) among members. ANI represents the average nucleotide identity of shared genes between two genomes. Distinct species usually exhibit ANI values lower than 90%. This observation contrasts with older views suggesting bacteria do not form distinct species because of horizontal gene transfer and large population sizes. Recent findings using larger datasets and novel methods indicate these are discrete units.

How are “strains” defined in metagenomics, and what is the significance of the 99.5% ANI threshold?

In metagenomics, the term “strain” is challenging to define precisely, especially without culture data. Traditionally, a strain is a group of genetically similar descendants of a single cell or colony. However, within metagenomic data, a 99.5% ANI threshold has been identified that effectively distinguishes between groups of closely related organisms within a species. Below 99.2% and above 99.8% ANI are often observed with a scarcity of genomes between these values creating a gap. Genomes sharing >99.99% are proposed as ‘strains’ but this is somewhat flexible depending on the genes that distinguish them. The 99.5% ANI gap aligns well with traditional sequence types (STs) used in medical microbiology but with much greater precision across the entire genome vs the 6-7 selected loci that STs use. Therefore, the 99.5% threshold is proposed for defining a more robust and data-driven “genomovar” for intra-species groups, while traditional STs may be kept if there is a necessity to adhere to that system.

What are read recruitment plots, and how are they used to study population diversity?

Read recruitment plots involve mapping metagenomic reads against a reference genome. These plots visually display the read mapping patterns, highlighting genomic regions with high or low coverage or areas of sequence discontinuity. Read recruitment plots provide a transparent, quantitative view of natural populations and are useful for identifying gene gains or losses over time and detecting variations. Additionally, tools have been developed to assess the average nucleotide identity of the reads relative to the reference (ANIr), which reflects the clonality of the population within the sample.

How can we estimate the relative abundance of a target genome within a metagenome?

Estimating relative abundance involves comparing the sequencing depth of a target genome to the overall sequencing effort of the sample. A sophisticated metric called TAD-80 (truncated average sequencing depth over the middle 80 percent of indices sorted by depth) is recommended over simple average sequencing depth to avoid spurious matches from non-diagnostic regions. Additionally, normalizing TAD-80 by prokaryotic genome equivalents (GEQ) helps to correct for genome size differences among samples. A non-zero TAD-80 value is used as the threshold to determine the limit of detection.

What is meant by the “limit of detection” in metagenomics, and how is it approximated?

The limit of detection (LOD) is the minimum amount of a target organism that can be reliably detected given a certain sequencing effort. The LOD is approximated by considering both sequencing breadth and depth. A sequencing breadth of at least 10% is a general guideline for a reliable detection. Additionally, there are formulas that relate sequencing effort to minimum detectable relative abundance depending on the sequencing breadth that is chosen for the limit of detection. However, this theoretical approach should complement rather than replace experimental controls such as spiked-in standards.

How is metagenomic data used to identify the etiological agent of a disease?

Identifying a disease’s etiological agent using metagenomics involves combining three key features: higher in-situ metagenomic abundance of the suspected pathogen in disease samples compared to control samples; higher level of clonality (lower intra-population diversity, i.e., higher ANIr) in the pathogen causing disease, and the detection of known virulence genes. Read recruitment plots and analysis of community composition provide additional useful information, even when the pathogen has not been isolated in culture. These approaches have even enabled the identification of the disease agent when traditional approaches have failed.

Glossary of Key Terms

• Metagenomics: The study of genetic material recovered directly from environmental samples, encompassing all the genomes present within a sample.
• Shotgun Sequencing: A sequencing method where DNA is randomly broken into fragments, sequenced, and then reassembled.
• Sequencing Depth: The number of sequence reads covering a specific position in a genome or feature.
• Sequencing Breadth: The fraction of base pair positions in a feature covered by at least one sequence read.
• Sequencing Effort: The amount of data produced by metagenomic sequencing, often measured in gigabases (Gbp).
• Genome Equivalents (GEQ): An estimate of the average sequencing depth of a single copy gene in a sample used to normalize relative abundance.
• Relative Abundance: The proportion of a specific feature (e.g., a genome, gene) in a metagenomic dataset, calculated as a ratio of sequencing depth to sequencing effort.
• Average Nucleotide Identity (ANI): A measure of the average percentage identity between the nucleotide sequences of two genomes, used to assess their relatedness.
• ANIr: The average nucleotide identity of mapped reads against a reference genome which is used to assess its in situ clonality.
• Sequence Type (ST): A genetic marker based on the sequences of a few housekeeping genes, used to distinguish closely related bacterial strains.
• Genomovar: A term proposed to describe discrete genomic groups within a bacterial species sharing 99.5% ANI, as an alternative to ST.
• TAD-80: Truncated average sequencing depth over the middle 80 percent of indices sorted by depth. Used to get an unbiased, robust measure of abundance.
• Read Recruitment Plot: A visual tool that shows how reads from a metagenome map to a reference genome, revealing sequence discontinuities and gene content diversity.
• Metagenome-Assembled Genomes (MAGs): Genomes reconstructed from metagenomic data, typically representing the average genome of a population.
• Limit of Detection (LOD): The lowest concentration of a target that can be reliably detected by a measurement method.

Reference

Lindner, B. G., Gerhardt, K., Feistel, D. J., Rodriguez-R, L. M., Hatt, J. K., & Konstantinidis, K. T. (2024). A user’s guide to the bioinformatic analysis of shotgun metagenomic sequence data for bacterial pathogen detection. International journal of food microbiology, 410, 110488.