

The strain you start with is almost never the strain you produce with. In between sits work that has a technical name — strain development — and that in practice runs on two parallel tracks. It is worth looking at the two ends first, because the distance between them is what tells you how much work is needed.
The starting strain comes from a collection, an environmental isolation or an academic lab. It does the right thing: it makes the molecule you care about. Usually it makes it in a shake flask, under comfortable conditions, at modest titres, with a variability nobody has quantified yet and a generational stability nobody has checked yet.
The finished strain is a different animal. It performs in a fermenter, which means in the presence of gradients, oxygen limitation and mechanical stress; it holds for the number of generations the real process requires; it behaves the same way from batch to batch; and it is documented well enough to be shown to an industrial partner or an investor. Between the two there is a jump that in our experience is often an order of magnitude, and you do not cover it by working in one way only.
Strain development acts on the organism. Screening and selection when the producer still has to be found; adaptive laboratory evolution, or mutagenesis and selection, when there is one but it underperforms. The decision that weighs most is the selection criterion, and it almost never matches the highest titre: in most industrial processes the winning strain is the more stable producer, not the one that shines in a single run.
Process development acts around the organism. Medium, carbon and nitrogen source, temperature, pH, feeding profile, oxygen transfer. This is the underrated half, and very often the one where yield is gained fastest: the same strain, under well designed conditions, can change behaviour without anyone touching its genome.
The point is that the jump from starting titres to industrial ones almost always comes from the combination of the two, and the most expensive mistake is pulling the wrong track. Months spent improving the organism when the limit was oxygen transfer are months you do not get back.
With a few targeted experiments, before committing to a programme. If yield moves when you change feeding, agitation or medium, the bottleneck is process-related and the organism can wait. If the strain stops at the same point whatever you put around it, the limit is biological. It is a cheap diagnosis and it steers everything that follows.
Biological time does not compress: cycles of growth, selection and verification take as long as they take, and a serious adaptive evolution programme is measured in weeks. Anyone promising shortcuts is compressing the verification, not the work.
And there is the risk that the gain does not survive scale. A strain selected under conditions that do not resemble the final process has effectively been selected for the wrong problem.
At the two points you only see if you go and look. The scale transition: a brilliant yield at 50 mL that halves at 100 litres is not bad luck, it is a selection made against a criterion that was too convenient. And generations: productivity that decays after a few dozen generations is not a strain defect, it is a figure nobody measured.
How much does yield weigh on your product cost? Below a certain share, gaining twenty per cent of titre does not change the economics. Is the bottleneck biological or process-related? How many generations does the strain have to hold in the real process? A short batch and continuous production make different demands. And what do you need to be able to demonstrate, and to whom? An industrial partner, a regulator and an investor ask for different data packages.
In Bioindustry projects microbial selection is treated as core R&D, not as a preliminary screening to get out of the way. High-throughput screening is combined with experimental design on the A.L.EX platform, so promising conditions surface early rather than after months of trial and error; adaptive evolution toughens the strain where it needs it; industrial applicability and freedom to operate are checked at the start, not at the end. The two tracks are deliberately kept together, and the final check is always the same: that what works at 2 litres keeps working at 100.
The page on microbial strain development sets out the three project phases, what we measure and what you take home; the one on fermentation process development covers the other track. If you do not yet know which of the two you need, you start from the diagnosis.
Technical content and news from our laboratories, sent when there is something worth saying.