Figure 1: Metagenomics and single-cell genomics take fundamentally different paths to microbial genomes — bulk community DNA versus individual cell resolution — and the choice between them determines which biological questions can be answered.
Shotgun metagenomics has become the default tool for culture-independent microbial community analysis. It delivers a census of who is present and what functional potential they carry, and it scales across large sample cohorts at a cost that most research budgets can accommodate. Single-cell microbial genomics — isolating, lysing, and amplifying the genome of one cell at a time — has historically been the more technically demanding and lower-throughput option.
That balance is shifting. Advances in microfluidics, hydrogel encapsulation, and combinatorial barcoding have pushed single-cell genomics from boutique method toward scalable platform. At the same time, a series of systematic comparisons published in 2024 and 2025 have clarified exactly where single-cell approaches add information that metagenomics cannot recover — and where metagenomics remains the more practical choice.
The decision is not about which method is better. It is about which biological question you need to answer, and whether the extra resolution of single-cell genomics justifies the added cost and complexity for that specific question.
Two Approaches, Different Gaps
Metagenomics starts with DNA extracted from an entire microbial community. That DNA is sequenced, assembled, and computationally sorted into bins — metagenome-assembled genomes, or MAGs — that represent population-level genome reconstructions. When a community is dominated by a modest number of abundant species, this works well. When diversity is high, closely related strains co-exist, or mobile genetic elements need to be traced to their hosts, MAG-based analysis hits structural limits.
Single-cell genomics starts at the opposite end. Individual cells are isolated — by droplet microfluidics, hydrogel encapsulation, or flow sorting — and their genomes are amplified, typically through multiple displacement amplification. The product is a single-amplified genome, or SAG: a genome recovered from one cell, with no computational binning required. The strengths and weaknesses are nearly mirror images of metagenomics.
Genome quality and chimerism. Chang and colleagues compared thousands of SAGs and MAGs recovered from the same global ocean samples and found that SAGs carried substantially less chimeric assembly — sequences from unrelated organisms erroneously joined together — and more accurately reflected the true relative abundance and pangenome content of microbial lineages. MAGs, however, were better at recovering genomes from rare, low-abundance taxa that cell sorting might miss.
Strain resolution. A 2025 study by Kifushi and colleagues compared SAGs and MAGs constructed from the same soybean field soil and rhizoplane samples. SAGs recovered 1,031 medium- or high-quality genomes spanning 22 bacterial phyla and 693 strains. MAGs from the same samples yielded 357 genomes across 17 phyla and 191 strains. The overlap between the two genome sets was less than one percent — they were capturing fundamentally different subsets of the community. Only SAGs resolved strain-level abundance changes in Bradyrhizobium populations across soybean growth stages.
Mobile element host-linking. The same study found that 11 percent of SAGs carried detectable plasmid sequences, spanning five phyla and nineteen genera. MAGs detected plasmids in 8.1 percent of genomes and across only two phyla. Critically, SAGs captured quorum-sensing genes on Bradyrhizobium plasmids across all three species present — evidence of interspecies plasmid transfer that MAG-based analysis was blind to. Metagenomic binning cannot reliably assign plasmids to host chromosomes because plasmid nucleotide composition and copy number differ from those of the host genome.
The 16S rRNA gap. Roughly 7 percent of MAGs from short-read metagenomes contain a full-length 16S rRNA gene, which means the vast majority of MAGs cannot be linked to the 16S-based taxonomic surveys that dominate microbiome literature. SAGs recover rRNA genes at much higher rates — 69 percent in the Kifushi study — making them far more useful for connecting genome-resolved information to existing amplicon-based datasets.
The table below summarizes the key differences at a glance.
| Capability | Metagenomics (MAGs) | Single-Cell Genomics (SAGs) |
|---|---|---|
| Resolution | Population consensus | Individual cell / strain |
| Chimerism risk | Higher — computational binning errors | Lower — no binning required |
| 16S rRNA recovery | ~7% of genomes | ~69% of genomes |
| Strain-level discrimination | Limited | Resolved |
| Plasmid/phage → host linkage | Cannot do reliably | Direct physical linkage |
| Novel lineage discovery | Good for abundant taxa | Excellent — captures phyla MAGs miss |
| Performance in high-diversity samples | Degrades — short, disconnected contigs | Unaffected by community complexity |
| Throughput and scalability | Very high — hundreds of samples | Lower but growing rapidly |
| Cost per sample | Lower | Higher — cell handling + amplification |
When Single-Cell Adds Value
Single-cell genomics is not a replacement for metagenomics. It is an answer to specific questions that bulk methods handle poorly or not at all.
Linking mobile elements to hosts. If your project tracks antibiotic resistance genes, plasmid dissemination, or phage-host interactions, single-cell genomics provides the physical linkage that metagenomic binning cannot. A 2025 study by Ling and colleagues used semi-permeable capsule-based single-cell sequencing on sewage and fecal samples and directly assigned the top antimicrobial resistance genes to their individual bacterial host species — information that guides risk assessment and intervention design in a way that community-level ARG abundance data cannot.
Accessing the uncultured majority in complex environments. In high-diversity samples — soil, sediment, rhizosphere — metagenomic assembly produces short, disconnected contigs that resist binning. The Kifushi study quantified this: metagenomic read information dropped 25-fold from raw reads to MAGs in bulk soil, while SAGs retained substantially more genomic information. Forty-three percent of soil SAGs belonged to genera with no existing annotation, compared to 18 percent of MAGs from the same samples.
Resolving strain-level heterogeneity. Metagenomics produces population-consensus genomes. If your question concerns within-population variation — which strains carry which functions, how individual genotypes respond to treatment, whether a pathogenic variant exists within a commensal population — single-cell resolution is necessary. The Kifushi study's Bradyrhizobium example is instructive: different strains of the same species showed distinct temporal abundance patterns that MAGs collapsed into a single population-level average.
Emerging frontier: functional heterogeneity at the transcript level. Single-cell transcriptomics for microbes is maturing rapidly. A 2025 Science review by Pountain and Yanai cataloged methods that capture antibiotic-tolerant persister subpopulations, prophage induction heterogeneity, and mobile genetic element expression in human gut microbiomes — all functional states that bulk metatranscriptomics averages out of existence.
When Metagenomics Fits Better
For a large class of research questions, metagenomics is the right tool — and the cost difference is not trivial.
Broad community surveys. When the objective is a census — which taxa are present, at what relative abundance, and with what functional gene complement — shotgun metagenomics delivers. The bioinformatics ecosystem is mature, reference databases are extensive, and the cost per sample allows the cohort sizes that statistical power demands.
Abundant population genomics. For dominant community members, metagenomic coverage depth enables high-quality MAG reconstruction that captures pangenome content and metabolic pathway architecture. Single-cell approaches add little marginal value for these abundant populations while multiplying cost.
Hypothesis generation before targeted investigation. A practical workflow for many projects is metagenomic screening followed by single-cell investigation of the specific lineages or mobile elements that the screening identifies as important. Using single-cell genomics for every sample in a large cohort is rarely the best use of resources.
Large cohort and epidemiological studies. Metagenomics scales. Standardized extraction kits, established library preparation protocols, and automated bioinformatics pipelines allow consistent processing across hundreds or thousands of samples — a level of throughput that single-cell methods, despite rapid progress, do not yet match.
Figure 2: A decision framework for choosing between metagenomics and single-cell genomics based on research question, sample type, and required resolution.
Getting the Best of Both
The emerging consensus in the 2024–2025 literature is that integrated workflows outperform either approach alone.
A combined strategy typically follows four stages. First, shotgun metagenomics establishes community composition baselines and identifies knowledge gaps — which lineages are uncharacterized, which mobile elements are prevalent, which functional pathways are differentially abundant. Second, single-cell genomics targets the specific lineages or elements that metagenomics cannot resolve, recovering reference genomes from uncultured taxa and physically linking plasmids and phages to their hosts. Third, metagenomic reads are recruited against SAG-derived scaffolds to improve completeness — the iSAG approach, which can lift genome completeness from the 40–50 percent typical of standalone SAGs toward the 80–90 percent range. Fourth, single-cell transcriptomics or metatranscriptomics adds a functional layer that confirms whether the genomic potential identified in earlier stages is actually expressed.
The Kifushi study's soil data illustrates why integration matters: SAGs and MAGs overlapped by less than one percent, meaning each method recovered genomes the other missed entirely. A project that uses only one approach leaves a substantial fraction of the community uncharacterized.
Figure 3: An integrated workflow showing how metagenomics and single-cell genomics complement each other in a combined analysis strategy.
What to Ask Before You Choose
The decision between single-cell genomics and metagenomics — or the decision to combine them — rests on a small set of project-specific questions.
What is the research question at its most specific? "Characterize the gut microbiome" points toward metagenomics. "Identify which bacterial hosts carry the CTX-M β-lactamase gene in this wastewater sample" points toward single-cell genomics. The more precisely a question can be stated, the clearer the method choice becomes.
How diverse is the sample? Low-diversity samples — human gut, fermented foods, bioreactor communities — are well served by metagenomics. High-diversity samples — soil, sediment, coral holobiont — benefit disproportionately from single-cell approaches because metagenomic assembly fragments in these environments.
Does the question require host-element linkage? If plasmids, phages, or antibiotic resistance genes must be traced to specific host genomes, single-cell genomics is not optional. Metagenomic binning cannot make this link reliably, and no computational workaround changes that fact.
What throughput does the study design require? If statistical power demands hundreds of samples, metagenomics is the scalable choice. Single-cell genomics can be applied to a strategically selected subset — for example, samples from phenotypic extremes or from time points where mobile element transfer is suspected — while metagenomics covers the full cohort.
Is there a functional dimension that matters? Metagenomics predicts functional potential. Metatranscriptomics and metabolomics measure functional output. Single-cell transcriptomics, though still an emerging capability for microbes, can resolve which cells within a population are actually expressing a function of interest — an answer that none of the other approaches provides.
FAQ
How much does single-cell microbial genomics cost compared to metagenomics?
Single-cell genomics typically costs several times more per sample than shotgun metagenomics at equivalent depth. The cost premium reflects cell sorting or encapsulation, whole-genome amplification, and more complex library preparation. However, per-genome costs can be competitive when the metric is high-quality genomes recovered rather than samples processed — particularly in complex environments where metagenomic binning produces few usable MAGs. A growing number of providers offer microbial single-cell genome sequencing with flexible throughput options.
Can I apply single-cell genomics to frozen or archived samples?
It depends on the preservation method. Cells must be intact for isolation and lysis, so glycerol stocks, DMSO-preserved samples, or freshly thawed material that retains cell viability are generally suitable. Fixed samples, samples stored in DNA extraction buffer, or samples that have undergone multiple freeze-thaw cycles are poor candidates. Consultation with the sequencing provider about sample compatibility before starting collection is advisable.
Which method recovers more biosynthetic gene clusters?
Single-cell genomics generally recovers larger and more complete biosynthetic gene clusters from complex communities, because it avoids the assembly fragmentation that breaks BGCs across multiple short contigs in metagenomic data. If natural product discovery from uncultured microbes is the goal, single-cell approaches offer a structural advantage that computational binning does not fully compensate for.
Do I need to choose one method, or can I use both on the same samples?
Using both is not only possible — it is the direction the field is moving. The same sample can be split for parallel metagenomic DNA extraction and single-cell sorting. The combined data supports co-assembly workflows that produce higher-quality genomes than either method alone. Several of the microbiome sequencing services available today support multi-approach study designs.
Is single-cell transcriptomics ready for routine microbiome studies?
Microbial single-cell transcriptomics has demonstrated proof-of-concept across gut, rumen, and environmental samples, and commercial platforms now exist. As of 2025, it remains more expensive and lower-throughput than bulk metatranscriptomics and requires specialized expertise for data analysis. For projects where functional heterogeneity — antibiotic persistence, phage induction variability, strain-specific metabolic activity — is a central question, it can provide information unavailable through any other method. The 2025 Science review by Pountain and Yanai provides a thorough assessment of method maturity.
How do I know whether my sample is too complex for metagenomics-only analysis?
If pilot metagenomic sequencing produces assemblies with low N50, high proportions of disconnected short contigs, and few bins meeting medium-quality thresholds (≥50% completeness, ≤10% contamination), the sample is a candidate for single-cell supplementation. Soil, sediment, and highly diverse host-associated communities — coral, sponge, rhizosphere — frequently fall into this category.
Related CD Genomics Services
- Microbial Single-Cell Sequencing — Single-cell isolation, whole-genome amplification, and sequencing for uncultured microbe genome recovery, strain resolution, and mobile element host-linking.
- Microbiome Sequencing Services — End-to-end sequencing across amplicon, shotgun metagenomics, and long-read platforms for microbial community analysis.
References
- Chang T, Gavelis GS, Brown JM, Stepanauskas R. Genomic representativeness and chimerism in large collections of SAGs and MAGs of marine prokaryoplankton. Microbiome. 2024;12:126. doi:10.1186/s40168-024-01848-3
- Hosokawa M, Nishikawa Y. Tools for microbial single-cell genomics for obtaining uncultured microbial genomes. Biophysical Reviews. 2024;16:69-77. doi:10.1007/s12551-023-01124-y
- Kifushi M, Nishikawa Y, Hosokawa M, Anai T, Takeyama H. Strain-level dissection of complex rhizoplane and soil bacterial communities using single-cell genomics and metagenomics. DNA Research. 2025;32(6):dsaf032. doi:10.1093/dnares/dsaf032
- Pountain AW, Yanai I. Dissecting microbial communities with single-cell transcriptome analysis. Science. 2025;389(6764):eadp6252. doi:10.1126/science.adp6252
- Ling M, Szarvas J, Kurmauskaitė V, Kiseliovas V, Žilionis R, Avot B, Munk P, Aarestrup FM. High throughput single cell metagenomic sequencing with semi-permeable capsules: unraveling microbial diversity at the single-cell level in sewage and fecal microbiomes. Frontiers in Microbiology. 2025;15:1516656. doi:10.3389/fmicb.2024.1516656
- Mirzayi C, Renson A, Zohra F, et al. Reporting guidelines for human microbiome research: the STORMS checklist. Nature Medicine. 2021;27(11):1885-1892. doi:10.1038/s41591-021-01552-x
For Research Use Only. Not for use in diagnostic or therapeutic procedures.
