Metagenomics for Pathogen Surveillance in Public Health

When a cluster of patients presents with unexplained febrile illness, standard diagnostic panels — malaria smears, typhoid cultures, targeted PCR — often come back negative or incomplete. Metagenomics pathogen surveillance using metagenomic next-generation sequencing (mNGS) offers a hypothesis-free alternative that can identify bacteria, viruses, fungi, and parasites in a single assay without prior knowledge of the etiological agent. This webinar summary draws on field studies from West Africa and Southeast Asia to illustrate how mNGS is reshaping public health outbreak investigation and clinical diagnostics.

Key Takeaways

• mNGS identified the causative agent in outbreaks where standard PCR and culture failed, including monkeypox and yellow fever in Nigeria
• In Lassa fever patients, mNGS revealed secondary viral co-infections in nearly 8% of cases and malaria co-infection in two-thirds — information invisible to targeted diagnostics
• mNGS assembled complete Lassa virus genomes from two samples that tested PCR-negative, exposing gaps in existing primer-based surveillance
• Negative mNGS results have clinical value: excluding infectious etiology redirected one outbreak investigation toward environmental poisoning
• Rigorous quality control — including NTC thresholds, 10% genome assembly cutoffs, and ERCC spike-ins — is essential to distinguish true signal from contamination

Why Unbiased Sequencing Changes Outbreak Response

Traditional diagnostics test for what clinicians suspect. In regions where malaria and typhoid fever dominate differential diagnoses, atypical pathogens go undetected until clinical deterioration forces broader investigation. This hypothesis-driven approach introduces dangerous delays.

mNGS reverses the logic. By sequencing all nucleic acids in a sample and performing taxonomic classification computationally, the method can detect known pathogens, novel viruses, and unexpected co-infections in a single workflow. The typical approach involves RNA extraction from plasma, a DNase step to enrich viral content and reduce host background, conversion to cDNA with random primers, library construction, and sequencing on platforms selected for the required depth and sample throughput.

Real-World Outbreak Investigations Using mNGS

Monkeypox — Confirming a Novel Outbreak in Nigeria (2017)

In 2017, an outbreak of unknown cause emerged in Nigeria. Working with plasma samples, researchers used mNGS to confirm the presence of monkeypox virus. The viral load in plasma was extremely low — the initial coverage plot showed only sparse genome fragments. A targeted metagenomic approach using hybrid capture enrichment recovered additional fragments, enough to classify the clade responsible for the outbreak. The finding was critical for public health response even though a complete genome assembly was not possible from plasma alone.

The case illustrates an important lesson for outbreak work: using multiple sample types (skin lesion swabs in addition to plasma) increases the probability of recovering sufficient viral material for complete genome characterization.

Yellow Fever — From Sequencing to Vaccination Campaign

A separate outbreak primarily affecting young individuals showed no initial clinical suspicion of yellow fever. mNGS identified yellow fever virus as the sole pathogen. The sequencing data was reported to public health agencies in real time, triggering a mass vaccination campaign that curtailed further spread. The outcome underscores the value of rapid, unbiased detection — a vaccine-preventable disease was causing a devastating outbreak because clinicians were not looking for it.

Ruling Out Infection — Pesticide Poisoning Disguised as Outbreak

Not every outbreak has an infectious cause. In one investigation, mNGS found no specific pathogen despite thorough sequencing. Rather than treating this as a failure, the team reported the negative finding to public health authorities, who expanded the investigation beyond infectious etiologies. The root cause turned out to be pesticide poisoning in the community — a presentation that mimicked infectious disease. mNGS provided the confidence to exclude infection and redirect the investigation.

Clinical Diagnostics for Unresolved Febrile Illness

Beyond outbreak settings, mNGS has resolved individual diagnostic puzzles where conventional testing failed. In one case, a two-year-old child returned from a rural area with unusual symptoms that did not match malaria or typhoid. mNGS identified enterovirus B3 as the causative agent. In another, a three-year-old with atypical hepatic presentation was diagnosed with hepatovirus A via metagenomic sequencing.

While these cases ended tragically — delays in sample processing and analysis meant results came too late to change outcomes — they provided clinicians with actionable diagnostic feedback. Future patients presenting with similar symptoms now have enterovirus and hepatovirus on the differential diagnosis, potentially enabling earlier targeted treatment.

Co-Infections in Lassa Fever — Hidden Complexity

A large cohort of Lassa fever patients in Nigeria during the 2018 surge provided the most striking demonstration of mNGS value. The key questions: had the virus mutated to enable sustained human-to-human transmission? What co-infections were present that might affect patient outcomes?

mNGS analysis revealed that nearly 8% of Lassa-positive individuals carried secondary viral co-infections, including enterovirus, hepatitis B, and HIV. HIV co-infection in Lassa patients, given the immunocompromised state, has direct implications for clinical management and prognosis.

The sequencing also detected Pegivirus C (formerly GBV-C) at high frequency among Lassa patients. Individuals with Pegivirus C appeared to have lower Lassa viral loads — a finding that parallels Ebola cohort observations and warrants investigation into viral interference effects.

Perhaps most concerning, mNGS assembled complete Lassa virus genomes from two samples that tested PCR-negative by the standard diagnostic assay — meaning existing surveillance primers are missing positive cases.

Approximately two-thirds of Lassa-positive individuals also had concurrent malaria — in endemic settings, identifying one pathogen does not exclude others.

Interpreting mNGS Data Without Generating False Positives

The power of mNGS comes with a responsibility: unbiased detection means unbiased noise. The webinar emphasized a structured approach to quality control and interpretation that laboratories adopting mNGS for pathogen surveillance should follow.

Non-template control (NTC) thresholds. Every sequencing run must include water controls. Any organism detected in the NTC establishes a contamination baseline. Reads in clinical samples must exceed this threshold to be considered potentially real.

Genome assembly requirement. A minimum of 10% genome coverage is required to call a detection as a true positive. Single reads or low read counts alone are insufficient — they may represent index hopping, environmental contamination, or computational misclassification.

ERCC spike-in controls. External RNA Controls Consortium spike-ins enable quantitative tracking of sample-to-sample variability and library preparation efficiency.

Reads per million (RPM) normalization. Raw read counts are misleading because total sequencing depth varies between samples. Converting to RPM provides a comparable metric across samples and runs.

Pipeline validation. No single bioinformatic pipeline is perfect. In the Lassa fever study, the initial pipeline misclassified Pegivirus C as Pegivirus A — an error caught only because the team added an additional QC step of phylogenetic analysis. Multiple classification tools and manual verification of unexpected findings are essential.

The PaRTI-Seq mNGS assay addresses several of these challenges with built-in features: zwitterionic host depletion via the Devin Filter to increase microbial signal, mNGS-grade reagents with a certificate of analysis (COA) for contaminant tracking, and a cloud-based bioinformatic pipeline designed to reduce false positive rates.

Workflow Next Step for Surveillance Labs

For public health or clinical teams moving from outbreak investigation to routine sample-to-report testing, the practical workflow question is where signal is lost. If host reads dominate, start with the Devin Host Depletion Filter. If library input is low after depletion, route to Unison Ultralow Library Prep. If the challenge is interpretation, use PaRTI-Seq Analysis to standardize taxonomic calls, QC thresholds, and reporting.

About the Speakers

Judith Uche Agu, DVM, PhD is a postdoctoral fellow at the University of Texas Medical Branch investigating emerging pathogens using metagenomic approaches. Her research spans H5N1, Lassa fever, bovine tuberculosis, and dengue fever.

Hannah Lucio is the global commercial lead at Micronbrane Medical with prior roles at Oxford Nanopore Technologies, Natera, and Thermo Fisher Scientific.

Related Reading

• What Is Metagenomic Pathogen Detection?
• The Hidden Crisis in Infectious Disease Diagnostics
• Host Depletion Validation for Public Health TB Labs
• Future of mNGS Diagnostics: Expert Panel on Barriers

ICCMg 2024 in Geneva: The Power of mNGS for Infectious Disease Diagnosis

ICCMg 2024 brought together clinical microbiologists and infectious disease specialists in Geneva, Switzerland in November 2024. Micronbrane Medical was represented by Dr. Amos Adler, director of clinical microbiology at a major medical center in Tel Aviv, Israel, who presented clinical validation data for metagenomic next-generation sequencing (mNGS) in sterile body fluids using the PaRTI-Seq™ workflow.

About This Presentation

Dr. Adler’s research focused on challenging sample types where traditional diagnostics frequently fail: cerebrospinal fluid (CSF), synovial fluid, and vitreous humor — collectively called sterile body fluids. These specimens are drawn once, in small volumes, from patients with serious or sight-threatening infections. Culture and PCR miss a substantial proportion of cases; Dr. Adler’s team set out to quantify that gap and demonstrate where mNGS closes it.

The PaRTI-Seq workflow evaluated in this study paired the Devin™ Microbial Enrichment Kit with the Unison Ultralow DNA NGS Library Prep Kit and the PaRTI-Seq Analysis bioinformatics pipeline. Head-to-head comparisons were performed against conventional culture and PCR reference standards across 40 positive samples and 11 true-negative controls.

The conference poster is available for download: Tel Aviv Sourasky Medical Center — ICCMg 2024 Poster (PDF)

Key Highlights from the ICCMg 2024 Talk

• 92.1% overall sensitivity, 100% specificity across 40 culture- or PCR-confirmed positive samples; the three false-negative cases were all Staphylococcus aureus from samples with suspected DNA degradation — not a platform limitation
• Endophthalmitis and uveitis emerged as a high-value clinical niche: vitreous taps yield tiny sample volumes, the differential spans bacterial, fungal, viral (HSV, VZV, CMV), and parasitic causes, and there is usually only one sample collection opportunity — exactly the scenario where unbiased mNGS provides the most clinical leverage
• RPM ratio threshold of 10 (RPMR = reads per million in sample ÷ RPM in no-template control) provided effective separation of true pathogens from reagent background, enabling clinically actionable reports
• Host depletion is critical for high-human-DNA samples: without effective host removal, biopsies and tissue samples consume tens of millions of reads without recovering microbial signal — making host depletion a prerequisite for cost-effective mNGS in challenging sample types

Key Takeaways

• Up to half of meningitis patients receive no confirmed microbiological diagnosis with standard methods — mNGS addresses this diagnostic gap directly
• A single clinical mNGS result can cover bacteria, fungi, viruses, and parasites simultaneously, replacing multiple separate assays that would otherwise exhaust a small sample
• Lot-matched negative controls and spike-in process controls are essential: reagent background varies by brand and lot, and false positives are most reliably excluded by system control subtraction rather than database filtering alone
• The combination of high sensitivity and broad pathogen range makes mNGS particularly valuable for ophthalmologic infections, CNS infections, and any setting where repeat sampling is not feasible

Related Resources

• mNGS Validation for Sterile Body Fluids: Methods and Results — extended summary of Dr. Adler’s validation study
• ICCMg 2024 Poster — Tel Aviv Sourasky (PDF)
• Devin Microbial Enrichment Kit
• PaRTI-Seq Analysis Platform
• TVGH Sepsis mNGS Study: PaRTI-Seq Detects 74.4% of Cases
• Host Depletion for Pathogen Identification in mNGS

Blood Microbiome Evidence: 9,770-Person Study

The concept of a resident blood microbiome — a stable community of microorganisms living in the bloodstream of healthy individuals — has generated both excitement and controversy in biomedical research. If real, it would fundamentally change how we interpret metagenomic next-generation sequencing (mNGS) results from blood samples. If false, the reported signals are artifacts that contaminate clinical decision-making. This webinar summary covers a landmark Nature Microbiology study that analyzed whole genome sequencing data from 9,770 healthy individuals in Singapore and found no evidence for a common blood microbiome — but did find something more nuanced than simple contamination.

Key Takeaways

• Analysis of 9,770 healthy adults found microbial signatures in fewer than 16% of individuals, with most showing only a single species — far below what constitutes a microbiome
• The most prevalent bacteria detected (Cutibacterium acnes, Lactobacillus, Bifidobacterium, Streptococcus species) are common skin, gut, and oral commensals, consistent with sporadic translocation rather than blood colonization
• Batch-based decontamination analysis identified known reagent contaminants (Sphingomonas, Methylobacterium) that prior smaller studies had reported as blood microbiome members
• DNA replication signatures confirmed some detected bacteria were recently viable, ruling out purely dead or artifactual signal
• No ecological community structure (co-occurrence networks) was found among blood-detected species, unlike established microbiomes at other body sites

Why the Blood Microbiome Question Matters for mNGS

Blood has traditionally been considered sterile in healthy individuals, with microbial presence indicating infection. This view has been challenged over the past decade by studies using PCR, culture, and sequencing to report bacteria in the blood of apparently healthy people. The largest pre-existing study used quantitative PCR (qPCR) to argue that microbes are present in blood across a large fraction of the population, though at low resolution without species-level identification.

If a stable blood microbiome exists, every mNGS blood test must account for a “normal flora” baseline — a major bottleneck in biomedical research for clinical metagenomics. If it does not, the detected signals in prior studies are likely reagent contamination, computational artifacts, or transient bacteremia that does not constitute a microbiome. Distinguishing between these possibilities requires a study at population scale with rigorous contamination controls.

Study Design — Repurposing a Precision Medicine Cohort

Rather than collecting new samples (which introduces its own handling variables), the study repurposed whole genome sequencing data from a multi-ethnic precision medicine cohort in Singapore comprising Chinese, Indian, and Malay individuals. The data was originally generated for human genetics research (genome-wide association studies), providing shotgun sequencing at depth exceeding 10x that of previously published blood microbiome studies.

This design offered two advantages. First, sequencing depth averaged 373 million reads per sample — far beyond typical microbiome studies. Second, the cohort spanned multiple collection batches, extraction kits, and reagent lot numbers with detailed metadata, enabling systematic contamination detection.

Despite the depth, only about 6,000 reads per sample (by median) mapped to microbial genomes — a ratio of roughly 1 in 60,000. Computational filtering had to remove artifacts without discarding genuine signal.

Separating Signal from Contamination in Low-Biomass Samples

The study’s decontamination approach is its most methodologically valuable contribution, directly applicable to any laboratory running mNGS on low-biomass sample types.

Batch-Based Contaminant Detection

Reagent contaminants follow a predictable pattern: they appear in samples processed with a specific extraction kit lot or library prep batch and are absent from samples processed with different lots. The study exploited this by comparing microbial detection across all known batches. Species present in only one or a few batches — but not others — were flagged as probable contaminants.

For example, Sphingomonas species YG-1 appeared exclusively in a small number of batches and was absent from all others — a classic contamination signature. Notably, some prior smaller studies had reported Sphingomonas as a blood microbiome member, illustrating how studies without batch-level controls can mistake contaminants for biological signal.

Correlation-Based Filtering

Contaminant species introduced by the same reagent batch tend to co-vary in abundance across samples — their concentrations rise and fall together because they originate from the same source. The study used correlation analysis to identify these co-varying clusters, catching contaminants that batch analysis alone might miss. This approach identified additional known reagent contaminants including Methylobacterium and Acinetobacter.

After filtering, the proportion of contaminant species dropped from 21% to 10%, while the proportion of human-associated microbes increased from 40% to 70% — a meaningful improvement in signal-to-noise ratio.

What the Data Actually Shows — Sporadic Translocation, Not a Microbiome

After decontamination, fewer than 16% of healthy individuals showed any microbial signal in their blood. Most of those had only a single species detected. The species list read like a catalog of commensals from other body sites: Cutibacterium acnes (skin), Lactobacillus and Bifidobacterium species (gut), and Streptococcus species (oral cavity).

Three lines of evidence argued against calling this a microbiome.

No community structure. Gut and oral microbiomes show dense co-occurrence networks — species whose abundances correlate because they interact ecologically. Blood-detected species showed no such network, even in individuals with multiple species.

Distinct from infection profiles. Hospital blood cultures are dominated by Staphylococcus, E. coli, and Klebsiella — species rare in the healthy cohort’s blood metagenomes. Sporadic translocation is biologically different from infectious bacteremia.

Evidence of recent replication. DNA replication origin analysis confirmed that some detected bacteria — including Fusobacterium nucleatum — showed signatures of recent cell division, while contaminant species like Acinetobacter showed flat coverage. The non-contaminant detections represent bacteria that were recently alive, consistent with transient translocation.

No associations were found between microbial detection and host phenotypes including gender, ancestry, age, or BMI. One apparent association between Cutibacterium acnes and ancestry was inconsistent across cohorts and likely not biologically meaningful.

Implications for Clinical Metagenomics and Host Depletion

The study’s conclusions have direct practical implications for laboratories using mNGS on blood samples.

Baseline definition. The sporadic translocation signatures — primarily skin, gut, and oral commensals at very low abundance — provide a reference baseline for what “normal” looks like in blood mNGS data. Any pathogen detection pipeline should account for these background species to avoid false positive calls.

Custom depletion strategies. Different patient populations may have different baselines. Individuals with gum disease likely show more oral bacteria; those with inflammatory bowel disease may show more gut translocation. The Devin Filter from Micronbrane Medical enables efficient host cell depletion from blood samples, and understanding the expected background helps laboratories set appropriate detection thresholds.

Contamination control is non-negotiable. Without batch-level metadata and systematic decontamination, low-biomass mNGS studies will inevitably report reagent contaminants as biological findings. Laboratories should track extraction kit lot numbers, library prep batches, and include no-template controls in every run.

Deeper sequencing has diminishing returns. As detection thresholds drop, the proportion of contaminants and computational artifacts increases. Below approximately 0.1% relative abundance, even stool samples carry uncertain signals. For blood, the challenge is greater still.

About the Speakers

Niranjan Nagarajan, PhD is the senior group leader at the Laboratory of Metagenomic Technologies and Microbial Systems and associate director of Genome Architecture at the Genome Institute of Singapore. He is also associate professor at the Yong Loo Lin School of Medicine, National University of Singapore. His research focuses on antimicrobial resistance transmission, microbiome ecology, and computational metagenomics.

Mengchu Wu, PhD is the co-founder, CEO, and chairwoman of Micronbrane Medical. She led the discussion on clinical implications of the study’s findings for metagenomic diagnostics.

Related Reading

• Cell-Free or Genomic DNA-Based mNGS Assays?
• Reagent Contamination Shown to Distort Pathogen Detection in mNGS
• How DNA Extraction Reagent Contamination Distorts mNGS
• Benchmarking Host Depletion: 5 Methods Compared for mNGS
• Managing Surface Contamination in Sequencing Workflows
• Oral Microbiome Profiling with mNGS: PaRTI-Seq Workflow
• Metagenomics Sample Prep: How to Reduce Contamination at Every Step

Contamination Detection in NGS: Methods, Controls & QC Protocol

DNA extraction reagent contamination is one of the most serious yet underappreciated threats to accurate metagenomic next-generation sequencing (mNGS) results — and one of the hardest to eliminate. Reliable contamination detection in NGS depends on understanding where reagent-derived microflora enters the workflow and how to distinguish it from true pathogen signal. While surface contamination and environmental sources receive considerable attention, the microflora inherent in extraction kits themselves can introduce false-positive signals that mimic real pathogens, triggering misidentification and costly retesting. Eliminating cross-contamination in NGS therefore requires controls that capture the reagent background, not just the laboratory environment. In a micro-webinar featuring research conducted at the China Medical University Hospital (CMUH) in Taiwan, Dr. Maurice Chan and Dr. Zi-Lun Lai presented systematic data comparing contamination profiles across extraction reagent brands, lots, and laboratory sites.

Key Takeaways

• Every extraction reagent brand tested carried a distinct contaminating microflora profile, confirming that most background organisms originate from the reagents themselves
• Contaminant profiles vary significantly between lots of the same brand, making lot-specific negative controls essential
• Spike-in process controls (SICP) serve as both positive and negative controls without introducing additional contaminants
• A consistent blood microbiome was not detected in 10 healthy controls, eliminating the need for healthy blood as a negative control
• Purpose-built bioinformatics filtering reduced raw organism calls from hundreds to two confirmed pathogens in a clinical sample
• Lot-matched negative controls and SICP spike-ins are the most effective way to eliminate cross-contamination in NGS workflows that use commercial extraction kits

Where Reagent Contamination Enters the mNGS Workflow

Contamination can enter at virtually every stage of the mNGS workflow: sample acquisition (skin microflora during blood draws), DNA extraction (reagent-inherent microflora), library preparation (reagent contamination and barcode hopping), and sequencing (alignment errors caused by conserved microbial sequences). However, research has increasingly identified extraction reagents as the most serious contamination source.

Published studies have traced organisms long assumed to be disease-associated back to spin column microflora. For example, Parvovirus sequences associated with certain diseases were ultimately found to originate from extraction column contamination. These findings underscore that reagent contamination does not merely add noise — it can generate clinically misleading signals.

Despite growing awareness, no established criteria exist for identifying or excluding contaminants from extraction reagents, and manufacturers of extraction kits generally do not guarantee the absence of contaminating DNA in their products.

Brand-to-Brand Variation in Extraction Kit Microflora

The CMUH study compared extraction reagents from multiple brands (designated M, Q, R, and Z) using two types of input material: molecular biology grade (MBG) water (representing an extraction blank) and a spike-in process control (SICP) containing two known organisms, Allobacillus halotolerans and Imtechella halotolerans.

Heat map and principal component analysis (PCA) clustering revealed that each brand carried a distinct contaminant profile. The backgrounds were non-overlapping between brands, confirming that the organisms were inherent to the reagents rather than introduced from a shared environmental source.

Critically, MBG water and SICP samples produced indistinguishable contamination profiles within each brand, demonstrating that the spike-in control organisms did not introduce additional contaminants. This finding validates SICP as a dual-purpose control: positive (the two spike-in organisms should be detected) and negative (all other organisms represent reagent background).

Among the brands tested, Devin Microbial Enrichment Kit reagents showed the lowest background contamination, appearing predominantly as dark blue (low frequency) on the heat map while other brands displayed large blocks of red and light blue (high frequency) for numerous organisms.

Lot-to-Lot Inconsistency Within the Same Brand

The study also examined multiple lots from the same brand and found that contaminant profiles varied substantially between lots. This has direct operational implications: a negative control processed with a different reagent lot than the test sample may fail to capture the relevant background organisms, leading to incorrect pathogen calls.

The practical recommendation is straightforward. Negative controls must come from the same reagent lot as the test samples and should be processed simultaneously. Manufacturers should provide lot-specific quality control certificates documenting the background microflora for each reagent lot.

Cross-Site Reproducibility of Contaminant Profiles

To determine whether contaminant profiles are site-dependent, the study compared results from reagents quality-controlled at the Micronbrane Medical laboratory and then used at CMUH. The heat maps showed largely consistent profiles between sites, with only sporadic environmental contamination events appearing at the CMUH laboratory.

This cross-site consistency means that the manufacturer’s QC certificate for reagent background is broadly applicable regardless of where the reagents are used. However, each laboratory should still run its own negative controls to capture any site-specific environmental contamination.

Does a Blood Microbiome Complicate Negative Controls?

Earlier publications suggested that healthy blood might harbor a baseline microbiome that would need to be accounted for in negative controls. If true, laboratories would need to maintain healthy blood samples as control material, creating significant logistical burden.

The CMUH study tested 10 healthy blood controls against extraction blanks and found no consistent microflora. None of the healthy controls produced significant organism calls. This eliminates the need for healthy blood as negative control material. Standard extraction blanks, MBG water, or SICP controls are sufficient for background subtraction.

Contamination Detection in NGS: Filtering Hundreds of Hits to True Pathogens

The webinar demonstrated the critical role of bioinformatics filtering using data from an endometrial sample. Raw sequencing produced hundreds of organism hits. Applying a minimum threshold of 10 reads still yielded 129 organisms. Filtering by microbial sequence percentage (at least 0.1% of total microbial reads) reduced the list to 39. Applying system control subtraction (comparing against the negative control) brought the count to 11. Final computational filtering identified just two confirmed pathogens.

PaRTI-Seq Analysis implements this multi-stage filtering approach automatically. The platform uses a curated database of over 1,400 pathogens and applies system control subtraction followed by fold-change thresholds to distinguish true signal from reagent background. The final report includes genome coverage uniformity plots, allowing the interpreting clinician to verify that the identified organism’s genome was uniformly sequenced, as expected for whole-genome mNGS.

Contamination QC Checklist for NGS Laboratories

The CMUH research distills into a set of practical QC steps that any laboratory running mNGS can implement regardless of which extraction kit they use. Each item maps directly to one of the contamination failure modes described above.

1. Use lot-matched negative controls. Process a negative control (MBG water or extraction blank) from the same reagent lot as your test samples, in the same run. A negative control from a different lot will not capture the relevant lot-specific background organisms and may miss real contamination.
2. Add a spike-in process control (SICP). Include SICP organisms (Allobacillus halotolerans and Imtechella halotolerans, or equivalent) in every sample including the no-template control. SICP serves simultaneously as a positive control (confirming the spike-in organisms are recovered) and a negative control (all other organisms represent reagent background). It adds no new contamination of its own.
3. Apply system control subtraction before pathogen calling. Cross-reference all organism calls against the matched negative control. Only organisms present in the sample at a read count significantly above the negative control background should advance to pathogen-calling. Raw hit lists without this subtraction step will contain dozens of reagent-origin organisms.
4. Request lot-specific QC certificates from your reagent supplier. The CMUH data show that manufacturers’ background profiles are broadly reproducible across sites, so a manufacturer-provided lot certificate is meaningful — but only if it is lot-specific, not a generic product-level statement.
5. Verify genome coverage uniformity on final pathogen calls. A high read count mapping to a narrow genomic region may indicate misclassification rather than true detection. Coverage uniformity plots confirm that reads span the organism’s genome broadly, as expected for whole-genome sequencing of a genuine sample constituent.

Applying all five steps addresses the reagent contamination QC problem end to end: lot-matched controls prevent false negatives in the control comparison, SICP provides a built-in positive/negative standard, system control subtraction eliminates reagent background from the final call list, lot certificates extend QC accountability to the manufacturer, and coverage verification catches any residual misclassification. For broader context on contamination sources and mNGS standards, see Why mNGS Standards Matter for Microbiome Analysis and Removing Surface Contamination.

About the Speakers

Dr. Maurice Chan is the RD Lead at Micronbrane Medical. He has a distinguished record of developing high-value, high-complexity NGS-based diagnostic assays, with prior roles at MiRXES, Angsana Molecular & Diagnostics, Changi General Hospital, the National Cancer Centre Singapore, Vela Diagnostics, and the U.S. Centers for Disease Control and Prevention. He earned his bachelor’s, master’s, and doctorate in microbiology and molecular biology at the National University of Singapore.

Dr. Zi-Lun Lai is a post-doctoral research fellow in the Department of Laboratory Medicine at the China Medical University Hospital in Taiwan. She led this research and has over 20 peer-reviewed publications.

Related Reading

• How to Reduce Contamination at Every Step
• Reagent Contamination Shown to Distort Pathogen Detection in mNGS
• Removing Surface Contamination
• Blood Microbiome: Does It Exist? 9,770-Person Study
• Why mNGS Standards Matter for Microbiome Analysis
• Download the ASM 2024 reagent contamination poster (PDF)