I think the biggest risk is actually competition from other sequencing service providers (such as Azenta/Genewiz, who have been doing Sanger sequencing as a service for years) who are now launching competing services. One of the main reasons that Plasmidsaurus and Primordium have been so successful is that existing Sanger service providers got complacent and didn't keep with the latest technology.
Fantastic read, huge plasmidsaurus fan. Bit of an old-man pedantic quibble with the idea “plasmids work in mammalian cells”. I think most geneticists would discern between plasmids and episomes or bacterial artificial chromosomes, which contain eukaryotic DNA replication elements and other sequences necessary for mammalian expression. Yeast plasmids kind of muddy the water here. The main difference is copy number and size, with most bacterial ‘plasmids’ classically being ~<10kb and many copies, and the BAC class being bigger, up to 100kb+ and one or few copies per cell. The two classes tend to have different molecular cloning workflows as well due to toxicity, duplication, etc. Nobody calls a human BAC for injecting a mouse embryo to create humanized mice a ‘plasmid’, although they do indeed have rhyme in certain characteristics and workflow. Just an old man yelling at clouds here, ignore me, lovely post.
Very interesting analysis! Perhaps a comparison could be made with TSMC's semiconductor foundry business. One might suspect that TSMC is just a service that makes ASML's lithography machines accessible to chip designers, and that Plasmidsaurus makes Nanopore sequencing available to scientists working with plasmids. (And this suspicion perhaps helped each company face less intense competition early on.) But not only is it challenging to make complex equipment reliable for highly-demanding, time-constrained users, there's arguably a technological moat due to economies of scale. If I had to guess, one sees a failure mode of the equipment, investigates it, and then deploys a fix that benefits other users. Thus, one is able to amortize the cost of investigating and fixing failures across many customers, and one is able to build a moat comprised of all these fixes.
Something you might have missed on why labs don't just get a nanopore sequencer and do it themselves - the chips are highly parallel! This means you can't get the price per plasmid down to around $15 unless you have 96 samples at a time to sequence. Most molecular biology labs (like ours) will most frequently have single digit samples to test at a time. So the choice becomes: i) pay for next day sequencing for an extra dollar or two per sample ii) wait for potentially weeks until you have enough samples to run the sequencing yourself. For what it's worth, we have switched from plasmidsaurus to Eurofins, essentially because we can ship together with our traditional sequencing samples.
I do a lot of plasmid sequencing. I wonder if Plasmidsaurus multiplexes deeper than 96 samples within 1 flow cell (using the Flongle). Such a sequencing prep kit (>96 samples) is not available today off the shelf, and they have probably solved this problem? Publications like CircuitSeq show it is possible.
>Competition from other sequencing companies
I think the biggest risk is actually competition from other sequencing service providers (such as Azenta/Genewiz, who have been doing Sanger sequencing as a service for years) who are now launching competing services. One of the main reasons that Plasmidsaurus and Primordium have been so successful is that existing Sanger service providers got complacent and didn't keep with the latest technology.
Fantastic read, huge plasmidsaurus fan. Bit of an old-man pedantic quibble with the idea “plasmids work in mammalian cells”. I think most geneticists would discern between plasmids and episomes or bacterial artificial chromosomes, which contain eukaryotic DNA replication elements and other sequences necessary for mammalian expression. Yeast plasmids kind of muddy the water here. The main difference is copy number and size, with most bacterial ‘plasmids’ classically being ~<10kb and many copies, and the BAC class being bigger, up to 100kb+ and one or few copies per cell. The two classes tend to have different molecular cloning workflows as well due to toxicity, duplication, etc. Nobody calls a human BAC for injecting a mouse embryo to create humanized mice a ‘plasmid’, although they do indeed have rhyme in certain characteristics and workflow. Just an old man yelling at clouds here, ignore me, lovely post.
Very interesting analysis! Perhaps a comparison could be made with TSMC's semiconductor foundry business. One might suspect that TSMC is just a service that makes ASML's lithography machines accessible to chip designers, and that Plasmidsaurus makes Nanopore sequencing available to scientists working with plasmids. (And this suspicion perhaps helped each company face less intense competition early on.) But not only is it challenging to make complex equipment reliable for highly-demanding, time-constrained users, there's arguably a technological moat due to economies of scale. If I had to guess, one sees a failure mode of the equipment, investigates it, and then deploys a fix that benefits other users. Thus, one is able to amortize the cost of investigating and fixing failures across many customers, and one is able to build a moat comprised of all these fixes.
Something you might have missed on why labs don't just get a nanopore sequencer and do it themselves - the chips are highly parallel! This means you can't get the price per plasmid down to around $15 unless you have 96 samples at a time to sequence. Most molecular biology labs (like ours) will most frequently have single digit samples to test at a time. So the choice becomes: i) pay for next day sequencing for an extra dollar or two per sample ii) wait for potentially weeks until you have enough samples to run the sequencing yourself. For what it's worth, we have switched from plasmidsaurus to Eurofins, essentially because we can ship together with our traditional sequencing samples.
I do a lot of plasmid sequencing. I wonder if Plasmidsaurus multiplexes deeper than 96 samples within 1 flow cell (using the Flongle). Such a sequencing prep kit (>96 samples) is not available today off the shelf, and they have probably solved this problem? Publications like CircuitSeq show it is possible.