I just got back from Metabolomics 2026, the 22nd Annual Conference of the Metabolomics Society, held for the first time ever in Latin America, in Buenos Aires. I've been to plenty of analytical chemistry trade shows over the years, but this was my first time sitting in a room full of metabolomics researchers for four straight days — plenary talks in the morning, parallel scientific sessions all afternoon, and poster sessions that ran late into the evening.
I went in knowing my own corner of this world cold. Organomation has spent over six decades building nitrogen blowdown evaporators and concentration equipment that sits at the sample-prep end of the workflow — the step where you take a liquid extract and gently drive off solvent so what's left is concentrated enough to actually analyze. What I didn't fully appreciate, walking in, was just how much sits downstream of that step: the extraction chemistries, the software, the statistics, and entire subfields with their own vocabulary that I'd never had reason to learn. So I want to use this post to walk through the themes that came up again and again, partly because they're genuinely interesting, and partly because a few of them turned out to connect directly back to what we do.
The Disease Story Dominates — But Plants Quietly Won
Disease-focused research was everywhere. Cardiometabolic disease, neurological disorders, cancer, infectious disease and immunometabolism, and nutrimetabolomics each accounted for a double-digit number of abstracts. Jessica Lasky-Su's opening plenary set the tone, describing how metabolomics is increasingly serving as a bridge between molecular discovery and clinical decision-making, with examples ranging from Long COVID to asthma subtyping to biological aging.
But the single largest topic bucket in the whole program wasn't a disease area at all — it was plant metabolomics, by a comfortable margin. Fidele Tugizimana's closing plenary made the case for why: AI-driven metabolomics is unlocking African plant biodiversity for both agricultural resilience and natural-product drug discovery, treating plants as a largely untapped chemical library. For a company like ours that has historically aimed most of its content at pharma, food, and clinical labs, that was a useful nudge — there's a whole agricultural and natural-products audience out there running the same kind of extraction-and-concentration workflows.
mzmine: The Software Everyone Assumes You Already Know
Walk through any session of LC-MS posters and you'll see the same handful of software names cited over and over for turning raw instrument output into something analyzable: mzmine, MS-DIAL, XCMS, GNPS2. mzmine in particular came up constantly — it even had its own pre-conference workshop.
mzmine is free, open-source software that takes raw LC-MS (and now GC-MS, ion mobility, and imaging) data and works through the processing chain: filtering noise, detecting peaks, aligning the same compound's signal across dozens or hundreds of samples, and matching the results against spectral libraries to propose compound identities. The current version was built as a community effort and is now one of the two most widely used tools in the field, alongside XCMS, precisely because metabolomics datasets routinely produce thousands of unresolved features that no one could sort through by hand.
I had only recently learned of mzmine when I interviewed mzio CEO Dr. Ansgar Korf on my Concentration on Chromatography podcast.
What clicked for me is that this is the other bookend of the workflow we sit in the middle of. Our equipment helps get a clean, concentrated sample ready for the instrument. Tools like mzmine are what happens to the data after the instrument finishes — and apparently that step is just as much of a bottleneck as anything on the bench.
Exposomics: A Word I Had to Look Up Twice
I'd genuinely never heard "exposomics" before this conference, and by day two it felt like it was in half the talks. The exposome refers to the entirety of environmental exposures — chemical, dietary, microbial, even social — that a person accumulates over a lifetime, and exposomics is the discipline built around measuring it and tying it to health outcomes. Several vendor talks framed it as a natural extension of existing metabolomics platforms: the same high-resolution LC-MS and NMR instruments built for endogenous metabolites can also pick up low-abundance environmental xenobiotics, contaminants, and biomonitoring targets in the same biofluid sample.
That distinction — internal metabolite versus external exposure chemical — was new to me, but the sample-handling problem looks familiar. Whether you're concentrating a plasma extract to look for disease biomarkers or concentrating it to look for trace contaminants, you're still solving the same problem: get the analyte concentrated without losing it or degrading it before it reaches the instrument.
Dual-Phase Extraction, and Why a Solvent's Boiling Point Suddenly Matters
This is the one that hit closest to home. I kept seeing references to "biphasic" or "dual-phase extraction," and it turns out this describes a family of classic sample-prep methods — Folch, Bligh-Dyer, and Matyash among them — where a biological sample is mixed with a polar and a non-polar solvent so it separates into two layers: an aqueous phase carrying polar metabolites and an organic phase carrying lipids, both pulled from the very same sample. One poster from a group in Thailand directly compared dual-phase extraction against protein precipitation and solid-phase extraction methods for plasma NMR work, finding that each approach pulled out a noticeably different slice of the metabolome.
The detail that really caught my attention, though, was buried in a methods comparison paper a few people referenced: the MTBE solvent commonly used in the Matyash variant evaporates quickly enough on its own that it can throw off the accuracy of the phase-recovery step and hurt reproducibility between runs. That's not an abstract software problem — that's a controlled-evaporation problem, which is exactly the kind of thing nitrogen blowdown systems exist to manage. It was a good reminder that even in the most software- and AI-forward corners of this field, the chemistry of getting solvent off a sample cleanly and consistently still matters.
A Few More Threads Worth Watching
Machine learning and AI showed up as a theme in their own right, not just inside the plant biology talks — there was a dedicated session on machine learning and AI systems, plus a strong cluster of posters on multi-omics integration, where metabolomics is layered with genomics, transcriptomics, or proteomics data rather than analyzed alone. There was also a whole track on data ecosystems and knowledge sharing — open repositories, community spectral libraries, and tools like GNPS2 designed to make other labs' raw data searchable and reusable rather than locked in a single paper's supplementary files.
What I'm Taking Back to Organomation
I went to Buenos Aires expecting to learn a little about where our evaporators fit into modern metabolomics workflows. I came back with a much longer list of vocabulary than I expected, and a clearer sense that the field's energy right now is split between two poles: increasingly sophisticated software and AI on one end, and still-evolving, still-debated sample-prep chemistry on the other. We live on that second end. I'm planning to dig into a few of these threads — exposomics and extraction method validation especially — in upcoming podcast episodes and posts, so if any of this is in your wheelhouse, I'd genuinely like to hear from you.