One place with a lot of low-hanging fruit is just making predictors that perform better for east asians. There are groups working on this at the moment but none of them have released high quality East Asian predictors to the public yet. Another would be offering screening for traits besides just disease. There may be some companies doing the latter in the US but they are not public at the moment.
Another really big differentiator would be if you can create better predictors for non-disease traits. There are many important non-disease traits that parents would likely want to select for, but the predictors for them are not good simply because biobanks don’t collect the needed data. You COULD start your own biobank, or perhaps if you have really good connections, get one of the Chinese biobanks to collect the data for you.
Among the traits I’m thinking of would be things like:
personality
mental energy
tolerance to sleep deprivation
pro-social tendencies
interpersonal skills
Our predictors for these traits are terrible right now.
It’s worth noting though that there is already at least one Chinese company offering polygenic embryo screening for diabetes.
If you wanted to make your own biobank it would probably cost on the order of $20-$50 mil. So pretty expensive for a startup. If you could get access to genotype and phenotype data from Chinese biobanks (or even those from Taiwan or Japan), you could probably train some pretty good predictors for maybe $100k-$200k of labor from a reasonably skilled data scientist.
But you’d then run into the problem Orchid is having right now: convincing clinicians to offer your test, or convincing patients to request it. That’s actually one of the most underrated challenges. By default, the answer from most clinics will be no. And if you’re offering a secondary service, you will need to do a lot of educational outreach because most patients and clinicians have never heard of PGT-P. And those that have are often misinformed by the cadre of academics who publish misleading studies on the subject trying to show it doesn’t work. There have been less than 500 children born via PGT-P worldwide at this point.
If you JUST wanted to offer data analysis services for western clients, your main challenge would be data access. You’ll need access to data from biobanks or from 23&Me and they are very conservative when it comes to uses for things like embryo screening.
One place with a lot of low-hanging fruit is just making predictors that perform better for east asians. There are groups working on this at the moment but none of them have released high quality East Asian predictors to the public yet. Another would be offering screening for traits besides just disease. There may be some companies doing the latter in the US but they are not public at the moment.
Another really big differentiator would be if you can create better predictors for non-disease traits. There are many important non-disease traits that parents would likely want to select for, but the predictors for them are not good simply because biobanks don’t collect the needed data. You COULD start your own biobank, or perhaps if you have really good connections, get one of the Chinese biobanks to collect the data for you.
Among the traits I’m thinking of would be things like:
personality
mental energy
tolerance to sleep deprivation
pro-social tendencies
interpersonal skills
Our predictors for these traits are terrible right now.
It’s worth noting though that there is already at least one Chinese company offering polygenic embryo screening for diabetes.
If you wanted to make your own biobank it would probably cost on the order of $20-$50 mil. So pretty expensive for a startup. If you could get access to genotype and phenotype data from Chinese biobanks (or even those from Taiwan or Japan), you could probably train some pretty good predictors for maybe $100k-$200k of labor from a reasonably skilled data scientist.
But you’d then run into the problem Orchid is having right now: convincing clinicians to offer your test, or convincing patients to request it. That’s actually one of the most underrated challenges. By default, the answer from most clinics will be no. And if you’re offering a secondary service, you will need to do a lot of educational outreach because most patients and clinicians have never heard of PGT-P. And those that have are often misinformed by the cadre of academics who publish misleading studies on the subject trying to show it doesn’t work. There have been less than 500 children born via PGT-P worldwide at this point.
If you JUST wanted to offer data analysis services for western clients, your main challenge would be data access. You’ll need access to data from biobanks or from 23&Me and they are very conservative when it comes to uses for things like embryo screening.