NGS Library Prep Failure: Low-Biomass mNGS Triage
Challenges in Library Preparation from Low-Biomass and Host-Depleted Samples
Metagenomic sequencing is increasingly applied to sample types that yield limited microbial DNA, such as cerebrospinal fluid, vitreous humor, pediatric plasma, and host-depleted skin or nasal swabs. These low-biomass specimens frequently produce DNA concentrations that fall below the threshold for reliable metagenomic next-generation sequencing (mNGS) library preparation.
When this occurs, many labs are unable to proceed with library construction—resulting in lost sequencing runs, missed research or diagnostic opportunities, and unrecoverable specimens. Existing library prep protocols adapted from whole-genome sequencing are often unreliable below 10 nanograms, especially when microbial content is low or unevenly distributed.
Why Library Prep Fails Below 1 ng
- Low microbial input: CSF, vitreous humor, BAL, swabs, pediatric plasma, and host-depleted specimens may fall below Qubit detection. Use an ultralow-input kit such as Unison rather than a standard WGS kit.
- Host depletion reduces total DNA: host-cell removal can improve microbial fraction while lowering total recoverable DNA. Pair Devin host depletion with a low-input library prep strategy.
- Inhibitor carryover: extraction or sample matrix effects can reduce amplification efficiency. Re-check extraction cleanup and controls before blaming the sequencer.
- Amplification bias: kits that work at high input can distort community composition at trace input. Review the direct Unison vs NEBNext vs Illumina comparison.
- Wrong kit type: a general-purpose DNA kit may not fit host-depleted clinical mNGS or ONT workflows. Use the library-prep selection guide or the optimized ONT protocol.
This page is the failure-triage page. For a head-to-head benchmark, use the comparison page. For sample-type and platform selection, use the selection guide.
Study Design: Benchmarking Library Prep Performance at Low Inputs
To evaluate performance at ultralow levels, a benchmarking study was conducted using serial dilutions of stool-derived microbial DNA. While not a low-biomass matrix, stool provides a compositionally rich and consistent input source for controlled comparisons across input concentrations.
DNA extracted from a single healthy donor was diluted to generate input levels from 100 ng to 0.01 ng. For each level, triplicate libraries were prepared using four commercial kits:
- Unison Ultralow DNA NGS Library Preparation Kit (Micronbrane Medical)
- NEBNext Ultra II (New England Biolabs)
- Illumina DNA Prep
- Illumina TruSeq DNA Nano
Libraries were sequenced using the Illumina NovaSeq platform (~20 million paired-end reads per sample). After quality control with KneadData, taxonomic profiling was performed using MetaPhlAn4. Alpha diversity (Shannon, Simpson), beta diversity (Aitchison distance), and phylum-level composition were compared.
Comparative Kit Results
Results: Preservation of Diversity and Replicate Consistency at Ultralow Input Unison demonstrated consistent alpha diversity metrics down to 1 ng input, with stable Shannon and Simpson indices across replicates.
Alpha Diversity Results
Replicate libraries prepared with Unison also clustered tightly in beta diversity analysis, indicating minimal amplification bias and high reproducibility.
Beta Diversity PCoA Results
At 0.1 ng, Unison preserved phylum-level community structure without overamplification artifacts. NEBNext and Illumina kits exhibited distortion, including enrichment of Actinobacteria—a hallmark of low-input amplification bias.
Phylum-Level Composition Results
Illumina TruSeq DNA Nano could not be tested below 100 ng due to protocol input limitations.
Implications for mNGS of Trace and Irreplaceable Samples
For mNGS laboratories working with clinical or research samples that cannot be recollected, failed library prep has significant implications—delayed reporting, lost sequencing runs, unrecoverable data, and frustrated customers. DNA inputs that fall below Qubit detection are frequently discarded, even when microbial signal is present.
This evaluation demonstrates that library prep at ultralow inputs is both feasible and reliable. Among the kits tested, only Unison maintained taxonomic fidelity and replicate consistency at 1 ng, without introducing detectable bias. These findings are particularly relevant for labs handling samples after Devin host depletion, low-volume fluids, and other Qubit-negative specimens, where every nanogram counts.
Conclusion
Low-biomass and irreplaceable samples pose a persistent challenge for metagenomic sequencing due to DNA inputs that frequently fall below standard library preparation thresholds. This study highlights the feasibility of performing reliable NGS library prep at ultralow DNA concentrations. Among the kits evaluated, the Unison Ultralow DNA NGS Library Preparation Kit demonstrated superior performance, maintaining taxonomic accuracy and reproducibility at 1 ng input. These findings support its use as a dependable solution for preserving sequencing viability in trace-input workflows where sample conservation is critical.
