The more useful answer took longer, and it was about everything around the machine.
There is a particular way a hospital talks itself out of a working sequencing service, and it almost always begins with the question the director asked me. Which sequencer should we buy. It feels like the right place to start. It is where the brochures start, where the demos start, where the budget line starts. In my experience it is also where a surprising number of programmes quietly stall.
The instrument is maybe a fifth of what decides whether in-house sequencing works. I have watched enough labs stand one up, and watched a few fail to, to be fairly sure of the ratio. The machine is the part everyone can see. It has a price, a footprint, a throughput figure that fits neatly on a slide. It is reassuring precisely because it is legible. The other four fifths are not, and that is where the work lives.
Start with the menu, not the platform
Start with what the lab is for, which means the test menu, not the platform. Which assays, for which specimens, answering which clinical questions, at what volume this year and three years from now. Get that wrong and you have bought throughput you cannot fill or chemistry that does not suit the work. The platform is a consequence of the menu. It is not the starting point, however much the sales process would prefer it to be.
The economics, done honestly
Then the economics, done with some honesty, which is rarer than it should be. Everyone models cost per sample against volume and finds the point where in-house beats send-out. Fewer people model the number that keeps a finance director awake, which is the shape of the cash over time. There is a trough. For months, often six to twelve of them, the capital is committed, the salaries are running, the accreditation is not yet finished, and not one billable in-house result has left the building. You are paying for the new lab and the old send-out arrangement at the same time. Breakeven volume tells you the programme can work. It does not tell you how deep the hole gets before it does, or when you climb out of it. Both numbers matter. Only one of them tends to reach the slide.
Accreditation is the project
Accreditation is the part people file under paperwork and then discover is the project. Budget for it from the first day, and budget for the whole of it, the bioinformatics pipeline and the reporting standards and the external quality assessment, not only the wet lab everyone thinks of first. It is the long pole, and it does not get shorter because you looked away from it.
The people are the real constraint
Then the people, who are the real constraint, not the chemistry. A sequencer with no one to build and maintain the pipeline, and no one qualified to sign out a variant, is not a service. It is a backlog with a warranty. The dry lab deserves the same seriousness as the wet lab, staffed and trained before go live, not after the instrument has arrived and the samples are already waiting.
There is one more thing, and it is the one that separates a lab that works from a box that merely functions. Utilisation. Cost per sample is a promise the spreadsheet makes and only a full flow cell keeps. Run half empty, run when someone remembers to, and the figure you sold to the board simply evaporates. The labs that make in-house sequencing pay are the ones running real samples every week, on a cadence, whatever the weather. That regularity is the entire game. It is also the least impressive sentence in any business case, which is probably why it goes missing from so many of them.
Why the sequencer is the easy part
So why is the person who sells sequencers telling you the sequencer is the easy part. Because a machine sold into a lab with no programme around it does not stay sold. It becomes an expensive object under a dust cover, a line item someone comes to regret, a reference site I can never quote. That is a poor outcome for the hospital, and, less obviously, a poor outcome for me. The instruments I would rather have in the field are the ones running every week, in labs that planned for the other four fifths. Those are the accounts that grow, that add assays, that are still there in two years. My interest and the director's point in the same direction, which is not always true in this trade and is worth saying plainly when it is.
The practical version of all this is close to boring. Pick a platform you trust, one you will not spend three years second-guessing, and the hardware stops being the risk. We back Element for that reason, because it lets a lab stop worrying about the box and put that worry where it belongs, on the programme. The choice of instrument should be the calmest decision in the whole exercise. If it is the most anxious one, something upstream has already gone wrong.
The director bought the AVITI24™, for what it is worth. That was the short conversation. The long one, about the menu and the money and the people and the weekly cadence, is the one that decides whether he has a sequencing service in two years or a costly paperweight. It does not fit in a quote, and it rarely fits in a brochure. It is the real build, and it is buildable, which is the whole point. The box was never the hard part.