ERMA brings standardization to single-cell epicPCR analysis of microbial communities

A new open-source software pipeline promises to bring order to one of the most powerful yet technically unruly methods in microbial ecology. Researchers led by Adrian Dörr and Ivana Kraiselburd at the Institute for Artificial Intelligence in Medicine at the University Hospital of Essen, together with colleagues at the University of Helsinki and University Hospital Essen, have unveiled ERMA, the EpicPCR Resistome-Microbiome Analyzer, a Snakemake-based workflow designed to standardize, automate, and scale the analysis of epicPCR data. The tool, published in BMC Genomics, addresses a problem that has quietly hampered the field for years: the near-total absence of shared, reproducible analysis workflows for a technique whose biological promise has long outpaced its computational infrastructure.

To appreciate why ERMA matters, it helps to understand what epicPCR actually does. Emulsion, Paired Isolation and Concatenation PCR, to give the technique its full name, is a single-cell method that physically links functional genes to taxonomic marker sequences from the same microorganism. Individual cells are encapsulated in emulsion droplets, their genomes are fragmented, and a fusion PCR step concatenates a functional gene of interest, for example an antibiotic resistance gene, with the 16S rDNA marker that identifies which organism carried it. Sequencing the resulting fusion products then allows researchers to answer a question that conventional metagenomics struggles with: which microbes in a complex community actually carry which functional genes. This capability has made epicPCR increasingly important in microbial ecology and, above all, in antimicrobial resistance research, where surveillance of wastewater and other environmental reservoirs depends on knowing not just which resistance genes are present but which organisms harbor them.

The catch has been the analysis. Until now, most laboratories processed epicPCR sequencing output using custom, often unpublished scripts, tailored to individual studies and rarely shared in reusable form. That fragmentation made cross-study benchmarking difficult, reuse nearly impossible, and reproducibility a matter of goodwill rather than infrastructure. Two laboratories analyzing the same dataset could, and routinely did, arrive at different taxonomic assignments and different resistance gene inventories simply because their filtering rules and database choices differed. In a field where wastewater-based epidemiology is being actively developed for public health decision-making, such inconsistency is more than an inconvenience; it undermines confidence in the underlying measurements.

ERMA tackles the problem head-on with a fully automated, standardized workflow built on Snakemake, a widely used workflow management system that makes computational pipelines transparent, restartable, and scalable. The pipeline requires minimal user input and supports both Illumina short-read and Oxford Nanopore Technologies long-read sequencing data, a notable flexibility given that epicPCR studies have historically been split between these platforms. From raw sequencing data, ERMA carries the analyst through quality control, reference database preparation, dual similarity searches against separate taxonomic and functional databases, integration of the two result streams, filtering, and finally visualization of the linked taxonomic-functional pairs. Every step is documented and executed identically across runs, which is precisely what the field has lacked.

The modular design is a deliberate architectural choice with consequences beyond antimicrobial resistance. Because database preparation is modular and the pipeline can incorporate specific UniRef queries, researchers can retarget ERMA to functional genes other than resistance determinants. A team interested in plastic-degradation genes, virulence factors, or metabolic pathways can in principle swap in the appropriate reference sets without rewriting the pipeline. This adaptability positions ERMA as a general-purpose epicPCR analysis engine rather than a single-purpose resistome tool, and the authors explicitly frame it as suitable for a broad range of research questions across microbial ecology.

Validation was carried out on two fronts. First, the team benchmarked ERMA against five publicly available epicPCR datasets spanning diverse sample sources and sequencing depths, from environmental samples to clinical material. Across these heterogeneous datasets, ERMA achieved mean concordance rates of approximately 70 to 83 percent with the taxonomic results reported in the original publications. Given that the original studies each used bespoke analysis approaches, this level of agreement suggests that a single standardized workflow can recover the core biological signal of epicPCR experiments while making the processing steps explicit and comparable. The residual differences between ERMA’s output and the published results are themselves informative, since they localize where analytical choices, rather than biology, drive divergence between studies.

The second validation was more direct. The researchers spiked a defined five-member mock community into clinical wastewater, a matrix notorious for its chemical complexity and microbial diversity, and ran it through the pipeline. ERMA recovered all five mock genera in full, a result that speaks to the pipeline’s sensitivity even in a realistic, high-background sample. As an independent cross-check, the same wastewater sample was analyzed by conventional 16S rDNA gene sequencing, and 85 percent of the genera detected by ERMA were also detected by that orthogonal method. For a technique that links function to identity at the single-cell level, agreement with an established amplicon-based approach at the genus level provides meaningful external validation that the fusion products are being correctly parsed and assigned.

One of the study’s more sobering findings came from the attrition analysis, which tracked how sequencing reads are lost at each filtering stage. The comparison revealed inconsistent filtering patterns across the five public datasets, reflecting dataset-specific differences in quality. In other words, the amount of usable information extracted from an epicPCR experiment depends heavily on the quality of the starting material and the sequencing run, and until now those losses were invisible, buried inside custom scripts. By making attrition explicit and standardized, ERMA gives researchers a diagnostic instrument: laboratories can now see where their data degrade and adjust library preparation or sequencing depth accordingly, rather than discovering problems only after publication.

The authors are candid about the field’s remaining limitations. Large-scale reference datasets for epicPCR remain scarce, which constrains how thoroughly any analysis pipeline, however well built, can be benchmarked. There is no vast, curated corpus of epicPCR data comparable to what exists for shotgun metagenomics, and building one will require the kind of shared tooling that ERMA now provides. Still, the combination of concordance with published results, full recovery of a mock community in clinical wastewater, and strong agreement with independent 16S rDNA analysis demonstrates strong and interpretable performance across genuinely heterogeneous inputs, which is arguably the hardest test a general pipeline can face.

The significance of ERMA extends beyond its immediate user base. Wastewater-based surveillance of antimicrobial resistance has moved from academic curiosity to policy relevance, with public health agencies worldwide investing in monitoring programs, and the WBEready consortium, funded by the German Federal Ministry for Health, supported this work as part of that broader effort. Surveillance only works if measurements from different laboratories and different time points can be compared, and that comparability is exactly what standardized analysis delivers. By providing a transparent, modular, and reproducible workflow as open-source software, ERMA establishes a foundation for methodological standardization and future benchmarking efforts in epicPCR, and as reference datasets grow, the pipeline’s performance should improve in step. For a technique that can reveal which bacteria in a wastewater stream carry which resistance genes, that kind of computational discipline may prove as important as any wet-lab innovation.

Subject of Research: Standardized computational analysis of epicPCR data linking functional genes to microbial taxonomy

Article Title: ERMA: a harmonizing epicPCR data analysis tool

Article References: Dörr, A., Dekić Rozman, S., Duncker, L. V., Kehrmann, J., Buer, J., Virta, M., Meyer, F., & Kraiselburd, I. (2026). ERMA: a harmonizing epicPCR data analysis tool. BMC Genomics. https://doi.org/10.1186/s12864-026-13418-y

Image Credits: AI Generated

DOI: 10.1186/s12864-026-13418-y

Keywords: epicPCR, ERMA, Snakemake, antimicrobial resistance, wastewater surveillance, microbial ecology, bioinformatics pipeline, Illumina sequencing, Oxford Nanopore, 16S rDNA, reproducibility, resistome