NGS has revolutionized clinical microbiology by offering new ways to identify and surveil pathogens without prior knowledge of the target. One popular approach is 16S rRNA and ITS amplicon sequencing, which targets specific microbial markers. But studies are increasingly incorporating shotgun metagenomics, a technique that sequences all DNA from a sample without targeting specific microbial markers. This approach allows more detailed taxonomic classification, detection of antimicrobial resistance genes, and reconstruction of microbial genomes (metagenome-assembled genomes, MAGs).
Relying on the whole DNA of a sample, however, also makes shotgun sequencing more sensitive to host DNA, environmental contamination, and analytical choices. So, if you are transitioning from amplicon to shotgun sequencing, there are a few key considerations to keep in mind. In this article, you will learn about:
• The challenges of extracting microbial DNA and conserving DNA length.
• How to manage host background and control contamination.
• How to choose library preparation methods, sequencing technologies, and sequencing depth.
• When to use read-based classification or genome assembly, and how to assess the results.
The workflow challenges of shotgun metagenomics
Many of the workflow challenges of shotgun metagenomics stem from one key difference: “while targeted 16S/ITS amplification largely bypasses human DNA, in shotgun metagenomics host DNA competes directly for sequencing capacity,” explained Armaan Mehra, Product Manager at Novogene Europe.
Thus, in low-microbial-biomass specimens, “most reads may therefore provide little microbial information,” Dr. Mehra said. Insufficient microbial coverage can limit genome reconstruction and increase costs if deep sequencing is needed. He added that, with little microbial material in the sample, reagent and environmental contamination can be harder to distinguish from the true biological signal.
Careful choices around nucleic acid extraction, host depletion/targeted enrichment, sequencing, and analysis are therefore key.
DNA extraction for metagenomics: balancing lysis and DNA integrity
Effective microbial lysis allows for an accurate representation of a microbial community, as some microorganisms, including Gram-positive bacteria, are harder to lyse than others. Using harsher lysis buffers or mechanical disruption methods can help, but these methods also risk shearing DNA. This is a particular challenge of working with shotgun metagenomics, as “preserving longer DNA fragments is a priority for genome assembly and understanding genomic context,” explained Anna Maria Niewiadomska, Associate Director for Micro & Infectious Diseases at Oxford Nanopore Technologies (ONT). Lysis must therefore be balanced with DNA integrity.
According to Dr. Mehra, these are some important things to take into account to achieve effective nucleic acid extraction while ensuring DNA quality is preserved:
• Keep sample collection and storage consistent, and avoid unnecessary freeze–thaw cycles.
• Match extraction protocol to the sample matrix and downstream objective. For stool and other complex matrices, an appropriate bead-beating step can improve recovery of difficult-to-lyse organisms. Depending on the target organisms, enzymatic or chemical lysis may also help, like cell-wall-disrupting enzymes for fungal-rich samples.
• Include extraction blanks to detect contamination and whole-cell mock communities (mixtures of known microbes) to check whether the method recovers organisms as expected.
• Don’t judge extraction only by total DNA yield: much of that DNA could be human. Also check microbial recovery, fragment length, host-DNA proportion, and inhibitors that could interfere with downstream steps.
Managing host DNA in shotgun metagenomics
During nucleic acid extraction, host DNA is also liberated from cells. For instance, in different respiratory samples, 94.1%, 99.2%, and 99.7% of the reads were human. Reducing host reads can be tackled at different stages of the workflow, from extraction to sequencing itself.
Host DNA depletion
Dr. Niewiadomska explained that “Host depletion through selective host-cell lysis is a valuable technique for host-rich samples, although this can also affect the microbial composition recovered and thus should be carefully considered.” Dr. Mehra added that selective host-cell lysis can work well when intact microbial cells are the target but may also lose extracellular or cell-free microbial DNA. This distinction is crucial in plasma, cerebrospinal fluid (CSF), and other specimens where cell-free pathogen DNA can be informative.
Target enrichment: when to narrow your search
Instead of reducing host DNA, you can enrich your target DNA. Choosing between targeted and untargeted shotgun metagenomics is a “matter of breadth versus sensitivity,” noted Dr. Mehra. As he explained, untargeted sequencing is useful when the causative organism is unknown or unexpected organisms are important. Targeted enrichment becomes attractive when the organisms of interest are reasonably well defined but present at very low abundance relative to host DNA, for instance, when working with metagenomic surveillance.
There are different approaches to targeted shotgun metagenomics, including multiplex PCR and hybridization capture. “Multiplex PCR can provide very high sensitivity for a defined panel but is constrained by primer design and sequence variation,” said Dr. Mehra. Hybridization capture, where DNA or RNA probes bind and capture sequences of interest, can instead “target broader pathogen groups or genomic regions while retaining more sequence diversity,” making it useful for genome recovery, antimicrobial resistance (AMR) analysis, and surveillance. Other methods like CRISPR-based enrichment are also emerging.
Enrichment during sequencing with ONT adaptive sampling
For nanopore sequencing, enrichment can also happen during sequencing itself through what is known as adaptive sampling. “Thanks to the real-time sequencing capability of ONT, MinKNOW can decide whether to continue sequencing individual DNA molecules or not based on supplied reference information,” noted Dr. Niewiadomska. This can be especially useful when additional sample preprocessing or physical enrichment is undesirable, Dr. Niewiadomska added.
Planning sequencing depth and choosing a technology
Compared with amplicon profiling, where most reads target microbial markers, shotgun profiling may require substantially more sequencing to obtain enough microbial reads. Thus, when planning sequencing depth, “the amount of usable microbial sequence data can be more informative than total read numbers,” explained Dr. Niewiadomska. This should be considered after host removal.
Dr. Mehra added that the required depth also depends on the endpoint. A 2026 benchmark using complex mixtures of DNA from cultured bacteria found useful reference-based strain-level information at approximately 0.5–1 Gb. In contrast, de novo MAG reconstruction may require substantially deeper sequencing, depending on community complexity, microbial abundance, and analytical objectives.
Choosing short-read, long-read, or hybrid sequencing
In terms of the sequencing technology of choice, “short reads remain cost-efficient and provide deep multiplexed coverage,” which is especially important when working with large sample cohorts, noted Dr. Mehra. Their limitation, however, comes down to genomic context. Dr. Niewiadomska pointed out that, “Long reads can be valuable for assembling more complete microbial genomes, resolving plasmids and other mobile genetic elements, and understanding the genomic context of antimicrobial resistance genes.”
Some researchers also opt for a combination of both. Hybrid sequencing can be useful when both high-confidence base calls and long-range linkage are needed, “particularly for reference-quality genomes or complex plasmids,” explained Dr. Mehra.
Shotgun metagenomics analysis: read classification and genome assembly
Bioinformatic analysis can be one of the most time-consuming parts of shotgun metagenomics due to the volume and complexity of the data generated, especially when reconstructing MAGs. Genome assembly requires piecing reads together into longer sequences, followed by binning and quality assessment.
When to use read-based classification
But if your goal is to identify which microorganisms are present, read-based classification can be a good option. “Read-based classification is particularly useful when speed, sensitivity, or low-abundance detection matters, because organisms can be identified directly from sequence reads without waiting for successful genome assembly,” explained Dr. Mehra. This is the case for pathogen detection or surveillance workflows, where requiring successful assembly could exclude organisms present at low abundance.
Dr. Mehra noted that established tools include Kraken2 with Bracken for classification and abundance refinement, and MetaPhlAn 4 or mOTUs for marker-based community profiling. However, “for a clinically relevant low-abundance signal, read count alone is insufficient,” Dr. Mehra added. Confidence increases when reads map uniquely across multiple regions of a reference genome, show appropriate breadth and depth of coverage, and are absent from negative controls.
Read-based classification and MAG reconstruction can also be combined to get the best of both. As Mehra explained, “read classification can identify organisms too rare to assemble, while MAGs provide genomic context for strain variation, functional genes, and mobile elements.” In a recent pilot study using nanopore sequencing to investigate bloodstream infections, initial taxonomic assignments were obtained within approximately one hour of sequencing, but runs continued for up to 72 hours to improve genome coverage, assembly continuity, and reconstruction of plasmid-associated sequences. The authors noted that extended sequencing primarily enhanced genomic completeness and contextual analyses but did not alter initial species identification or major AMR predictions.
Assessing MAG quality beyond completeness and contamination
When reconstructing genomes, both Dr. Mehra and Dr. Niewiadomska agreed that MAG quality needs to be assessed using several complementary metrics. “These include completeness, contamination, fragmentation, and assembly continuity,” noted Dr. Niewiadomska. Data show that passing thresholds such as >90% estimated completeness and <5% contamination does not automatically make a MAG biologically reliable. As Dr. Mehra noted, complementary checks could include GUNC for taxonomic chimerism; genome size; contig count; N50 and read-mapping coverage for fragmentation and structural consistency; and GTDB-Tk and ANI comparisons for taxonomic context.
About the author: Nuria M. Wentzien got her Ph.D. from the University of Granada in 2023. After finishing her Ph.D., she discovered that science communication brings together her two favorite worlds: science and writing. Her career reflects her passion for learning and exploring new topics, from a Bachelor’s degree in Biochemistry (with a strong focus on human sciences) to a Ph.D. in sustainable agriculture. She brings this same curiosity to every project, making readers feel as inspired by science as she does. Now dedicated to helping life-science companies communicate rigorous science in an engaging way, she is the co-founder of Helixa Communications.