But I do want to sequence it using a third-party that gives me all the raw data. I live in Europe and I'm just a simple consumer. Does anyone know how I can do this? What service would I use / you can recommend?
Them not keeping it on their side would be a huge bonus of course but not sure I can ask for that much.
They did not leak their database, the attacker guessed some user passwords and scraped those people's match lists.
I don't think they even got their hacked user's test results - you can download those from 23andMe, as GDPR requires, but it's a "request your data" process which isn't so easy to do at scale without people noticing.
23andMe is also not NGS sequencing, so you get only a couple of 100000 letters randomly* distributed across your whole genome.
I know you put an asterisk, but to make it clear: these locations are very far from random. They're the most informative positions, which is why we check them.
Well, they're what 23andMe thinks are most informative, for the things they think their customers care about. Or thought; for compatibility reasons they can't change their SNP panels very often or too much.
And the interesting positions are still reasonably randomly distributed by nature :)
I don't do SNP design but I don't think 23andMe made their own chips at first (if they have ever). SNPs are chosen based on being informative for population discrimination, with traits being overlaid afterwards
I could be wrong, but I do believe they get to choose. What I do know is that none of the big testing companies test exactly the same SNPs, there is some overlap but every test is different. They also change their SNP sets from time to time, 23andMe has changed at least 5 times, and interestingly, their current test isn't the biggest (it tests fewer SNPs than some of their earlier ones).
well SNPs aren't useful for drug development, commercialization, medicine, etc. you want whole genome or whole exome sequencing for stuff that is future-looking. they use a process for the present day (quickly becoming the past) of okayish drugs instead of sequencing for a future of cures.
That gives some probabilistic info to scraper. Like person might have genes that are:
- susceptible to drugs abuse
- or game addiction
Bam - you get target ads
- Indicate that any common substance is especially dangerous/harmful for him (think of allergies). So attacker might plan assassination or severe inconvenience by exposure to the substance and then claim he did not know
Honestly, you don't need someone's genetics for this to happen. They will find out more about you by scraping your socials and spending habits and driving habits.
I think this whole fear is over blown anyway. I am in a genetic out group (schizoaffective disorder and yes my risk is genetic) and the benefits of getting my genetics run twice out weighed any risk that might have befallen me.
23andMe bet big that genetic big data would be super valuable. That really didn't work out for them.
It turns out that it's more valuable for advertisers to learn if you actually smoke, than that you have a genetic propensity for smoking. Your genotype is just useful for figuring out your phenotype, and in the vast majority of cases, your genetics (especially not at the resolution offered by 23andMe) are not a shortcut to learn your behavior, which is what they're really after.
You're probably already classed as some variety of paranoid/dissident for your careful social media preferences...
YSEQ is basically a mom- and pop shop, two German scientists who helped set up Family Tree DNA's labs in Texas then went home to start a small competitor (I think there was some sort of disagreement or conflict) in the space of high-resolution Y DNA sequencing for genealogists.
They also do high resolution full genome sequencing on request. But be warned, it takes a long time, and they reserve the right to cancel your order if you complain about how long it takes!
They're not the cheapest option, but when it comes to privacy for Europeans, I think they're as good as they come. You want a bit of "difficult to work with" when it comes to privacy.
> I live in Europe and I'm just a simple consumer. Does anyone know how I can do this? What service would I use / you can recommend?
At one point I randomly bumped into a cancer researcher as he were in town for some conference. I posed exactly the same question to him, and he told me to reach out to local labs, specifically to the individuals posted on the websites, and ask them directly to help me out or point me in the right direction. He said he've done it himself in multiple countries before, but I'm not sure yet people helped him because of his title, but he assured me I'd find someone willing to help me even if I was just a lowly software engineer. I have yet to actually try it myself, but maybe you could try if there is any nearby? :)
I want long-read raw data. Also Europe (Germany) and simple consumer. Is there any third-party service that offers this? How much will it cost me?
Why must it be long-read? The info I want is from nearly identical duplicated genes. I have FASTQ and BAM files from Dante Labs, but wasn't able to get the info I want out of them.
Some DNA companies (FTDNA for one) play fast and loose with this.
FTDNA gives you a vcf file, which contains the variant calls, but not the raw reads which those calls are made from (as in the BAM file). They do keep that data, because they charge extra for a BAM file download. It's almost certainly against the GDPR. Worse, I think they do it for anticompetitive reasons - they own the largest Y-DNA tree, and don't want you uploading your raw data to competing trees (in particular YFull).
Many companies offer that. In the local area they have e. g. mail-like stations where you give out your order and they give you what you need (e. g. if it is a plasmid, it is mailed into storage boxes often on the same day or next day).
"Not keeping it on their side" ... well. If they sequenced it, they have data in their computer, right? How could you avoid that? I don't see this as possible, it depends on trust (and whether you really care about that).
> What service would I use / you can recommend?
I won't recommend anything as I do not want to be an ad-amplifier, but my personal rule of thumb is that those companies that are affiliated with science clusters, are often chosen because they offer high quality (and to some extent because of corruption e. g. xyz knows abc, but this is IMO the smaller part, usually it is quality/speed/ease of use).
Much like many cameras process and discard the RAW sensor data after processing to pixel data if not further compressing and discarding the processed pixel data to a lossy image format, even more so the raw BAM sequencing read data is vastly larger than processed VCF files. Even many companies that retain that data are liable to archive it offline rather than keep the raw reads permanently accessible online. There are real costs involved and a business case is needed for keeping that data. Especially with a decent privacy policy or regulations storing 100G or multiples of that online for free and downloading on demand is a significant compared to the entire cost of sequencing.
They could have the sampling machine write to an SD card and mail that back, or store it with a one-time-use link like pwpush.com, or have it expire & self-delete after N days like WeTransfer.
You of course have to trust that they aren't secretly keeping their own copy on S3 forever, but that'll always be a risk, and it can probably be contractually enforced and audited.
"Not keeping it on their side" ... well. If they sequenced it, they have data in their computer, right? How could you avoid that? I don't see this as possible, it depends on trust"
First it depends on the contract, if it states they have the right, then they can and will legally use it.
If the contract would say no - then they would have a much harder time making use of that data, as it would be illegal.
> This is intended to be read by AI- please just copy and paste the URL of this and have ChatGPT walk you through it. If you have AR glasses, even better, since the AI can walk you through the whole protocol.
What kind of magic is going on here, am I missing something?
Hi, author here- I intended it to be hands-free so you can upload to ChatGPT/Claude and talk to it. I found it easier to follow the protocol each time by talking to AI rather than having to read from the computer every time I had to check something, reducing context-switching
If you wanna take it a step further you can just embed a chat/llm icon at the start of your article, give it a header like "summarize this article using:", that leads to this query:
I suspect the intention is to give specific but dense notes with minimal explanation, on the theory that the LLM will fill in the appropriate hand-holding along the way
I feel it's actually kind of smart. Most people won't be reading the blog post themselves, they'd ask GPT to understand the text and fetch the summary or whatever is relevant to them. The author has directly made the resource such that it is optimised for the output after that mostly-everyone-would-do-this step.
I've bee thinking about starting a company where I fish roots out of your sewer and identify the plant (by sequence if necessary) that you have to kill so your sewer doesn't collapse as soon as it otherwise would.
$100 to stave off that $10000 sewer replacement for a few years would be worth it to a lot of people
I don't really understand this. How much better are things for killing a specific plant than whatever you would use to kill a superset of the plants that might be there? Or treating with a few different plant species killers. And wouldn't those options cost less than the cost of this service?
This is precisely what I do each year with a product designed for this exact purpose. $8+shipping gets me enough for my annual pipe maintenance. It's even the special kind because I have clay pipes.
I don't wasn't to kill parents idea. It's neat, and Im sure there's use cases that my solution doesn't meet
Well if you have this problem, then having the plumber over is a sunk cost. You're paying them regularly to clear the sewer, and their snake will come back with root matter in the bit. Put that sample in a zip-lock bag and call me. I'll come over, take the sample, and identify the plants near your sewer line. If I get a match, I'll sell that info to you for $100.
Once I figure out how to make it work at all, I'll build a network of plant nerds and teach them to do the same in their cities, and pivot to providing lab services and training for them. Much of the time no sequencing will be required, just a microscope and knowledge of what's growing nearby. But if they have more than one plant of the same species, sequencing will be necessary.
Fingers crossed they're not clones, though I suppose I could do lab testing for that as well, and then I maybe you'd have to kill multiple just to be sure you got the one. In that case, hopefully I'd have at least narrowed it down for you. Probably would just deny the job if the odd of being helpful are low, like if you have 50 clones of the same tree all growing along your sewer line, then I can't help you, it's time to start saving for a liner or a replacement.
Fastest/cheapest sequencing based method would likely be to collect samples in DNA preserving agent (zymo DNA shield is decent for this), then have an automated platform for extracting DNA and preparing amplicon sequencing libraries (rbcL + matK probably?). Worth having a look at how Plasmidsaurus did this, they did have a similar service for microbiome analysis (https://plasmidsaurus.com/sample-prep/microbiome#16S-extract... - looks like they don't accept raw samples anymore though...). Oxford nanopore sequencing is a decent technology for this as it's fairly straightforward to get working and easy to scale :) As an added bonus, you could perhaps also carry out 16S sequencing in parallel from the sample to get an idea of the microbiome of the sewer.
I wish this had some discussion of the results. The earlier reports about this sensor and process were very mixed. It’s a cool process either way, but I’d like to know how usable the real world output can be.
Untrue in the cases of VEP, ClinVar, gnomAD. These are either offline, open-source tools or databases. You can download or query these and no one would be the wiser.
Claude on the other hand - yeah you are giving your data away. But that step really isn't necessary.
By running a standard pipeline you could get a VCF (File containing the Variants in your genome) and each variant would be annotated. You can check all the annotated genes and figure out if these variants are pathogenic, likely pathogenic, likely not pathogenic or benign.
I suspect there is a deep sequencing service that is non CLIA and cheap. True. they may not be trustworthy with the data. That said, there are steps here where the data is put into Claude. Do we trust that ?
That's another vector, and everyone considers separately. But at least here one can - hypothetically - have a BAA and zero-retention agreement with Anthropic. Which means they have moderately strong incentives to wipe your data after a fairly short while. CLIA, if I don't misremember, mandates a few years at minimum.
It's not impossible: Claude can take care of bioinformatics part. But to understand genetics and cellular biology you need a knowledge foundation first.
Unless you just want it on a level "Does this mutation leads to a genetic disorder X" - this is a simple way to put it, but not enough to actually understand genetics.
Has anyone succeeded in doing any of this? I always assumed if it were this simple, someone would be selling DNA kits on Amazon for a few thousand dollars.
Per the article, ONT sells tiny sequencers that all this genre of "Omg I just sequenced my genome" posts use. They've been around for a decade or so. A commercial lab will give you 30x illumina sequencing for maybe 5% of the cost of this post.
As an 'at home' science project? Totally do-able, but very expensive. I did it as part of epidemiological field with a backpack of materials, so you can definitly do it at home.
As a useful genome? Marginal. ONT will need a couple runs to get complete human genome coverage and the substitution error rate is still high compared to short reads. For a fraction of the price and effort you can get your DNA resequenced at 10 times the quality.
I guess if you are super-ultra paranoid you can do it all yourself at home, but honestly. You're not that special, and if someone really actually wanted your genome that bad they'd just take it from a used coffee cup.
This is so cool. Thanks for doing this. The fact that we have this in a palm sized object is just crazy. Also, if/when we have a similar sized device for doing CRISPR .... umm i should stop here - it's becoming the plot of Gattaca
As someone with experience (albeit almost 20 years out of date) experience of wet lab DNA collection and sequencing - this stuff is hard, you will fail a lot, and you will fail a lot more if you don't have an extremely clean environment to do this in. And once you have data, you should be asking yourself how accurate it is given the environment you collected it in, you should be looking at correlated sequence errors that are not taken into account[1].
But also: genetic counselling is a real thing that real people study. Please don't ask an LLM questions about what your genes are going to do to you without having access to someone who has the ability to contexualise the data and put you in touch with relevant experts. I have a PhD in this and I would not trust myself to be able to interpret data about myself in a detached and rational way.
(And: why is the link to Molecular Biology of the Cell to the 6th edition, when the 7th came out 4 years ago? Random fact: the first three editions were co-authored by my supervisor during my first PhD attempt, who went on to demonstrate that Roger Penrose's ideas about the importance of microtubules in chemotaxis in E. coli were absolute bullshit. Great guy)
[1] I spent a while analysing very early (by commercial standards) Illumina data in 2007, and being able to align stuff to reference genomes made it possible to identify certain biases. Nanopore technology is likely to have more of those, and if you don't have the ability to take those into account you may have a very bad time
Nanopore data is a lot easier to analyze than short read sequencing data. You just don’t get the same alignment/assembly issues: these things sequence incredibly long reads.
> this stuff is hard, you will fail a lot, and you will fail a lot more if you don't have an extremely clean environment to do this in
The Oxford Nanopore sequencing technology is one of the most robust to use. You need to buy some kit - but defo doable. Nothing compared pouring your own gel and doing radioactively labelled Sanger reactions :-)
Though you could just go to an sequencing company that services labs ( that just does sequencing outsourcing - rather than a personal genome company ).
Totally agree on the dangers around interpretation.
It might make even more sense once we get to the point of a wider use of encoding the data into dna. For now we have these few commercial players in the field that cad do it (eg look up dna microfactory for storage archiving), IIRC genomika was saying they can do an MB for a 100-200eur.
It seems a bit odd to me to have a home lab for the sequencing and then feed it to a cloud LLM. I would have thought the point of the home lab is privacy.
I like the privacy conscious aspects. Apart from the obvious issue of "run it through Claude" how many of those referenced analysis tools are entirely open source or at least run locally? Would have liked to see that in the article.
Bringing everything to your doorstep and everything at your feet and everything near your fingertips is just what all industries are trying to accomplice. The cartoon Animation Wall-E has scenes in it where they show obese humans doing everything through a screen though notice their legs and feet and it's as if they've mutated to a point where they aren't able to walk anymore and all their transport is through a hovering chair cum bed.
I watched wall-e in theater, and when that scene came on, i remember muttering 'what bs'; since then, i recall that scene every time i see a situation of 'convenience at all costs'; metaphorically they were pretty accurate even after discounting ozempic influence;
And you'd need the DNA of all dogs to boot (unless they are already somehow stored when a dog is registered). But generally speaking, I do like the idea. Also for horse poop which is pretty common on roads here in the countryside.
What is the accuracy in this ? Aka if I run the experiment 10 times how many differences will i get? I don’t have a physical sense on what would be a good number.
You would get a lot of differences, but the errors would cancel each other out with enough depth of coverage.
This technology's baseline accuracy is around 95% per base, so 10x reads of every segment in the sample would give >99% accuracy for each base after aligning the reads with each other.
For most DNA it now sits at ~Q20 (1%) which is a big improvement - there was also some hacky stuff like HERRO and duplex sequencing which improved it substantially. For routine genome sequencing it's definitely suitable, and the long-read advantage helps resolve some of those trickier regions
one main marketing leverage of 23andMe, AncestryDNA, etc are fulfilling the curiosity of people who want to know which part of the world their genes are from. I guess that dataset should be preparatory.
Problem with those providers - they only check 700K positions out of 3 billion and there is no mapping quality or allelic depth data in those dataset and this is critical for assessing whether the detected variant is a false positive or real.
It's not suitable for health investigations since most of DNA is not sequenced and genotyping technology is known to produce high rate of false positive for rare mutations.
(I'm the solo-founder of Gene Inspector Pro, mentioned in the blog post). AMA. :)
It's a bit ironic, since FTDNA and MyHeritage (which uses FTDNA's lab) have switched to NGS now, so they presumably could deliver those notorious "health insights", at least better than 23andMe. But they aren't in that market, and 23andMe shows no inclination to switch. They're probably licking their wounds after the user hack and buyout fiasco.
That's unlikely, but I do think that the health benefits of full sequence testing yourself are largely hypothetical at this point anyway. Unless you're a competitive athlete maybe?
You have no power to correct something with gene editing, true, but there may potentially be preventive action you can take. It's all largely hypothetical today anyway, most things you care about (e.g. cancer, Alzheimer's) you can't usually tell very reliably if you're going to get from a DNA test.
Nothing about this is the future. Sequencing at home will not solve any major problems. It's mainly a fun exercise to demonstrate that sequencing has been commodified.
I disagree, I sequenced with nebula genomics years ago.. you can understand risk factors for various problems so that you can start interventions that make sense way in advance.
I don't wasn't someone else to have a copy of my genome, whilst wanting to analyse my genome. The only way to do that is with affordable home sequencing.
True story, I found a god damn tick in my navel yesterday. Sometimes a bit of navel gazing can be healthy, figuratively and literally.
On a reasonable level, navel gazing (the figurative kind) is maybe better called self-reflection. I use DNA for genealogy, and it seems to me from the people I meet, that many get a healthier approach to our identity once we learn more about our genetic background. Identity politics, collective identity building around ancestry - identity building of all kinds really - needs simple stories. And the stories DNA tell are never simple.
When you are taking care of your health, and need to learn more about your built-in limitations is it still narcissistic?
At the bottom of the page there is a link to Sid Sijbrandij's cancer journey. He is one of the cofounders of GitLab. This is one of the coverages of his story: https://centuryofbio.com/p/sid
Is going to your doctor or eating better food "navel gazing"? Predispositions to some diseases can be read from your DNA. Remember Angelina Jolie undergoing preventive mastectomy because she had a high genetic risk for breast cancer? Well, so do many non-celebrities.
Then there is the specific case of people who may suspect that their bio-parents are someone else, and there is nothing weird about wanting to know where you actually come from.
Man, doctors thought they had it bad before. For just a six yards I can play Peter Thiel at home! $6k invested so I can set an AI in YOLO mode to tell me I have some hyper-specific version of kennel cough?
“But that occurs in dogs?”
“You’re right. Let me look into actual gene sequencing instead of just guessing. I think the N is the load bearing letter.”
Hi, author here- you can read it too, though it is dense. I have updated that specific sentence.
I found it easier to upload the protocol to ChatGPT and have audio walk you through it. This allows you to swap between pipettes, measurements, etc without having to look at the screen, reducing context-switching
> The near-term value is turning a static genome into something queryable
Ok. So ... how exactly is this valuable?
If you realise "hey, I gots Huntington disease", this is going to make you feel better? Or any other incurable disease? I am not disputing that knowing the sequence is useless in general, mind you. I am specifically asking WHY it is necessary to know your genome sequence. This seems to be a simplification or just a "having reached a milestone". But then they don't really explain WHY it is useful. None of the bulletin points he listed is really useful:
> Which variants do I have?
And this is useful ... how exactly?
> Which genes and pathways are affected?
And ... this matters why?
> Which medicines might I metabolize differently?
Ok, so this has a potential use case here, since he
can choose to avoid specific drugs. How useful that
really is in practice is unclear. (Don't confuse
drug companies trying to convince YOU that personalized
medicine is important on THEIR use case.)
> What rare variants should I take seriously?
Seriously ... how? Ok, you avoid some compounds. Now what.
> Where does the model know nothing yet?
Great, so a model that is limited, but now I need to
burden myself with having to know where that limitations
are. So my brain just has extra processing to do, without
getting anything useful in return.
> the “edit yourself with CRISPR” will most likely follow
Except that they have not solved the off-target cleavage
yet. Besides, they milk the prices anyway. DNA manipulation
should be safe, secure, correct and affordable. None of that
is the case right now. They publish papers where CRISPR has
solved everything, but then fail to explain why it isn't
already used by billions. And there are reasons as to why.
> Give your genome to Claude Code
Oh my god ... AI becomes your dependency here.
Note that the step-by-step guide is actually not totally
useless, as it can give a basis for real work. But I highly
doubt that untrained people will easily be able to go through
those steps. Everyone is a master in the lab now? RNA is
easy to handle? Guess then one would have to explain why
RNase A is used (ok ok it's not playing a huge role here
since DNA is the target of isolation, but it is more of an
example of how many things can go wrong, and there is not
really an explanation of why xyz is used; this looks like
an AI step-by-step guide. AI really makes people dumber).
The "non-random errors" point buried a few replies down deserves to be the headline, not a footnote. With Illumina, 10x coverage genuinely washes out errors because they're closer to independent per-read noise. With Nanopore, errors cluster at specific motifs (homopolymers, certain k-mers) due to how the pore physically reads the strand — so the same systematic mistake shows up across most of your reads at that position, and naive majority-vote consensus won't fix it. You need a basecaller/consensus model trained to correct for those specific failure modes (which is exactly what the current-gen Guppy/Dorado models try to do), not just "more depth." That distinction matters a lot for a home setup: coverage is cheap, but knowing where your specific errors are systematic vs. random is what determines whether "buy more reads" actually gets you to clinical-grade accuracy or just gives you a very confident wrong answer.
I don't want to sequence at home.
But I do want to sequence it using a third-party that gives me all the raw data. I live in Europe and I'm just a simple consumer. Does anyone know how I can do this? What service would I use / you can recommend?
Them not keeping it on their side would be a huge bonus of course but not sure I can ask for that much.
> But I do want to sequence it using a third-party
Don't forget to change your DNA when the third party's database will eventually leak. (see the 21andMe's data leak: https://en.wikipedia.org/wiki/23andMe_data_leak)
They did not leak their database, the attacker guessed some user passwords and scraped those people's match lists.
I don't think they even got their hacked user's test results - you can download those from 23andMe, as GDPR requires, but it's a "request your data" process which isn't so easy to do at scale without people noticing.
23andMe is also not NGS sequencing, so you get only a couple of 100000 letters randomly* distributed across your whole genome.
I know you put an asterisk, but to make it clear: these locations are very far from random. They're the most informative positions, which is why we check them.
Well, they're what 23andMe thinks are most informative, for the things they think their customers care about. Or thought; for compatibility reasons they can't change their SNP panels very often or too much.
And the interesting positions are still reasonably randomly distributed by nature :)
I don't do SNP design but I don't think 23andMe made their own chips at first (if they have ever). SNPs are chosen based on being informative for population discrimination, with traits being overlaid afterwards
I could be wrong, but I do believe they get to choose. What I do know is that none of the big testing companies test exactly the same SNPs, there is some overlap but every test is different. They also change their SNP sets from time to time, 23andMe has changed at least 5 times, and interestingly, their current test isn't the biggest (it tests fewer SNPs than some of their earlier ones).
well SNPs aren't useful for drug development, commercialization, medicine, etc. you want whole genome or whole exome sequencing for stuff that is future-looking. they use a process for the present day (quickly becoming the past) of okayish drugs instead of sequencing for a future of cures.
what is the attack vector btw?
That gives some probabilistic info to scraper. Like person might have genes that are: - susceptible to drugs abuse - or game addiction Bam - you get target ads
- Indicate that any common substance is especially dangerous/harmful for him (think of allergies). So attacker might plan assassination or severe inconvenience by exposure to the substance and then claim he did not know
Finding yourself in a genetically-determined out group, and being treated differently / discriminated against because of it.
Honestly, you don't need someone's genetics for this to happen. They will find out more about you by scraping your socials and spending habits and driving habits.
I think this whole fear is over blown anyway. I am in a genetic out group (schizoaffective disorder and yes my risk is genetic) and the benefits of getting my genetics run twice out weighed any risk that might have befallen me.
You have control over your habits, and what you reveal on your "socials". I post very little personal.
You have no control over your genes.
23andMe bet big that genetic big data would be super valuable. That really didn't work out for them.
It turns out that it's more valuable for advertisers to learn if you actually smoke, than that you have a genetic propensity for smoking. Your genotype is just useful for figuring out your phenotype, and in the vast majority of cases, your genetics (especially not at the resolution offered by 23andMe) are not a shortcut to learn your behavior, which is what they're really after.
You're probably already classed as some variety of paranoid/dissident for your careful social media preferences...
> You have control over your habits
If you have to change your habits because a secret system is watching over you it means they are already controlling you.
You cannot hide your bank account, where you drive, what you buy, and on and on and on.
And all I need to get your DNA is pick up a cup you drank from or go through your garbage.
YSEQ is basically a mom- and pop shop, two German scientists who helped set up Family Tree DNA's labs in Texas then went home to start a small competitor (I think there was some sort of disagreement or conflict) in the space of high-resolution Y DNA sequencing for genealogists.
They also do high resolution full genome sequencing on request. But be warned, it takes a long time, and they reserve the right to cancel your order if you complain about how long it takes!
They're not the cheapest option, but when it comes to privacy for Europeans, I think they're as good as they come. You want a bit of "difficult to work with" when it comes to privacy.
> But be warned, it takes a long time
What's a "long time"? A month, a year, 10 years?
> They're not the cheapest option
Whats "not the cheapest"? 100 EUR, 1000 EUR, 10000 EUR, more?
From their website yseq.net:
- Whole genome sequencing ultra plus (150 Gbases raw data) is €699,- incl. shipping for the sampling kit
- "The turnaround time is approx. 5-7 months at the moment."
> I live in Europe and I'm just a simple consumer. Does anyone know how I can do this? What service would I use / you can recommend?
At one point I randomly bumped into a cancer researcher as he were in town for some conference. I posed exactly the same question to him, and he told me to reach out to local labs, specifically to the individuals posted on the websites, and ask them directly to help me out or point me in the right direction. He said he've done it himself in multiple countries before, but I'm not sure yet people helped him because of his title, but he assured me I'd find someone willing to help me even if I was just a lowly software engineer. I have yet to actually try it myself, but maybe you could try if there is any nearby? :)
I want long-read raw data. Also Europe (Germany) and simple consumer. Is there any third-party service that offers this? How much will it cost me?
Why must it be long-read? The info I want is from nearly identical duplicated genes. I have FASTQ and BAM files from Dante Labs, but wasn't able to get the info I want out of them.
Sequencing dot com.
TellmeGen
DNA Complete.
Unless you live in Germany (in Germany it seems to be illegal to send saliva to other countries).
Can you get raw data from Tellmegen?
Can't seem to find any info on this on their website.
Yes, it is in profile/settings.
TellMeGen has a lifelong update, I guess they store your data to do that.
What does "lifelong update" mean in this context? are they not doing whole genome sequencing to begin with?
Seems like if new studies find a link between genes and diseases they would update the result of your gene based health risks.
I used Macromo, think theyre a czech company, they give you the data
YSeq, TellMeGen - or reaching out to local sequencing labs
TellMeGen says something about a lifelong update of the result.
So they store your data
If a service doesn't give you the raw data for some reason you should be able to get it with a GDPR subject access request.
Some DNA companies (FTDNA for one) play fast and loose with this.
FTDNA gives you a vcf file, which contains the variant calls, but not the raw reads which those calls are made from (as in the BAM file). They do keep that data, because they charge extra for a BAM file download. It's almost certainly against the GDPR. Worse, I think they do it for anticompetitive reasons - they own the largest Y-DNA tree, and don't want you uploading your raw data to competing trees (in particular YFull).
Many companies offer that. In the local area they have e. g. mail-like stations where you give out your order and they give you what you need (e. g. if it is a plasmid, it is mailed into storage boxes often on the same day or next day).
"Not keeping it on their side" ... well. If they sequenced it, they have data in their computer, right? How could you avoid that? I don't see this as possible, it depends on trust (and whether you really care about that).
> What service would I use / you can recommend?
I won't recommend anything as I do not want to be an ad-amplifier, but my personal rule of thumb is that those companies that are affiliated with science clusters, are often chosen because they offer high quality (and to some extent because of corruption e. g. xyz knows abc, but this is IMO the smaller part, usually it is quality/speed/ease of use).
Much like many cameras process and discard the RAW sensor data after processing to pixel data if not further compressing and discarding the processed pixel data to a lossy image format, even more so the raw BAM sequencing read data is vastly larger than processed VCF files. Even many companies that retain that data are liable to archive it offline rather than keep the raw reads permanently accessible online. There are real costs involved and a business case is needed for keeping that data. Especially with a decent privacy policy or regulations storing 100G or multiples of that online for free and downloading on demand is a significant compared to the entire cost of sequencing.
Files aren't magically stored forever.
They could have the sampling machine write to an SD card and mail that back, or store it with a one-time-use link like pwpush.com, or have it expire & self-delete after N days like WeTransfer.
You of course have to trust that they aren't secretly keeping their own copy on S3 forever, but that'll always be a risk, and it can probably be contractually enforced and audited.
"Not keeping it on their side" ... well. If they sequenced it, they have data in their computer, right? How could you avoid that? I don't see this as possible, it depends on trust"
First it depends on the contract, if it states they have the right, then they can and will legally use it.
If the contract would say no - then they would have a much harder time making use of that data, as it would be illegal.
> This is intended to be read by AI- please just copy and paste the URL of this and have ChatGPT walk you through it. If you have AR glasses, even better, since the AI can walk you through the whole protocol.
What kind of magic is going on here, am I missing something?
Hi, author here- I intended it to be hands-free so you can upload to ChatGPT/Claude and talk to it. I found it easier to follow the protocol each time by talking to AI rather than having to read from the computer every time I had to check something, reducing context-switching
You can still read it, though it is pretty dense
If you wanna take it a step further you can just embed a chat/llm icon at the start of your article, give it a header like "summarize this article using:", that leads to this query:
> https://chatgpt.com/?q=summarize%20this%20article%20for%20me...
You make one for claude, perplexity etc.
The AR angle tho, still lost on me, maybe i'm missing something...
Funny, I was really impressed with how concise and goal oriented the directions are!
I would have done better in uni if the lab manuals were written like this
But where does AR fit in? Doesn't this just need voice mode?
I suspect the intention is to give specific but dense notes with minimal explanation, on the theory that the LLM will fill in the appropriate hand-holding along the way
I feel it's actually kind of smart. Most people won't be reading the blog post themselves, they'd ask GPT to understand the text and fetch the summary or whatever is relevant to them. The author has directly made the resource such that it is optimised for the output after that mostly-everyone-would-do-this step.
> Most people won't be reading the blog post themselves, they'd ask GPT to understand the text and fetch the summary or whatever is relevant to them.
What a sad reality we live in. Or to quote C3PO from Star Wars 2... "Machines building machines. How perverse.".
I've bee thinking about starting a company where I fish roots out of your sewer and identify the plant (by sequence if necessary) that you have to kill so your sewer doesn't collapse as soon as it otherwise would.
$100 to stave off that $10000 sewer replacement for a few years would be worth it to a lot of people
I don't really understand this. How much better are things for killing a specific plant than whatever you would use to kill a superset of the plants that might be there? Or treating with a few different plant species killers. And wouldn't those options cost less than the cost of this service?
https://www.envirodna.com/
https://www.naturemetrics.com/species-detection
https://www.ednacollab.org/industry/
https://wilderlab.co/
These companies focus on environmental DNA - some are more on the level of local government monitoring, some are for private customers.
How will you reach out to enough people so they are aware and can order the service? I'm thinking about the minimum viable business opportunity here.
That’s a really clever idea, I would definitely pay for that in the right circumstances.
Now that I think about it - could you just pour some sort of biodegradable broad-spectrum herbicide down the drain to get the same effect for cheaper?
This is precisely what I do each year with a product designed for this exact purpose. $8+shipping gets me enough for my annual pipe maintenance. It's even the special kind because I have clay pipes.
I don't wasn't to kill parents idea. It's neat, and Im sure there's use cases that my solution doesn't meet
Isn't this a problematic idea? If you flush down herbicide without checking the results, you would risk breeding herbicide resistance in the weeds.
I hope yours keeps working for you. I flushed a lot of blue crystals, but the plant was not deterred.
> could you just pour some sort of biodegradable broad-spectrum herbicide down the drain to get the same effect for cheaper?
It will not degrade anywhere near fast enough to not cause serious issues in the biological treatment stage of the wastewater plant.
Hard to get a plumber to knock on my door for under $100. Maybe you mean $500? Or is it $100 for the lab bit only?
Well if you have this problem, then having the plumber over is a sunk cost. You're paying them regularly to clear the sewer, and their snake will come back with root matter in the bit. Put that sample in a zip-lock bag and call me. I'll come over, take the sample, and identify the plants near your sewer line. If I get a match, I'll sell that info to you for $100.
Once I figure out how to make it work at all, I'll build a network of plant nerds and teach them to do the same in their cities, and pivot to providing lab services and training for them. Much of the time no sequencing will be required, just a microscope and knowledge of what's growing nearby. But if they have more than one plant of the same species, sequencing will be necessary.
Fingers crossed they're not clones, though I suppose I could do lab testing for that as well, and then I maybe you'd have to kill multiple just to be sure you got the one. In that case, hopefully I'd have at least narrowed it down for you. Probably would just deny the job if the odd of being helpful are low, like if you have 50 clones of the same tree all growing along your sewer line, then I can't help you, it's time to start saving for a liner or a replacement.
It's like uber, but for shit-covered roots.
Fastest/cheapest sequencing based method would likely be to collect samples in DNA preserving agent (zymo DNA shield is decent for this), then have an automated platform for extracting DNA and preparing amplicon sequencing libraries (rbcL + matK probably?). Worth having a look at how Plasmidsaurus did this, they did have a similar service for microbiome analysis (https://plasmidsaurus.com/sample-prep/microbiome#16S-extract... - looks like they don't accept raw samples anymore though...). Oxford nanopore sequencing is a decent technology for this as it's fairly straightforward to get working and easy to scale :) As an added bonus, you could perhaps also carry out 16S sequencing in parallel from the sample to get an idea of the microbiome of the sewer.
Do it!
I wish this had some discussion of the results. The earlier reports about this sensor and process were very mixed. It’s a cool process either way, but I’d like to know how usable the real world output can be.
I want to sequence, but I absolutely do not want any company or government or church to have access to my data. When the author says:
> "I have a VCF, I can run it through tools like VEP, ClinVar, gnomAD, PharmGKB (highly recommend), Gene Inspector, or Claude"
I am assuming my data is now within the hands of some of the very entities I do not want to have access to my data ... true?
Untrue in the cases of VEP, ClinVar, gnomAD. These are either offline, open-source tools or databases. You can download or query these and no one would be the wiser.
Claude on the other hand - yeah you are giving your data away. But that step really isn't necessary.
By running a standard pipeline you could get a VCF (File containing the Variants in your genome) and each variant would be annotated. You can check all the annotated genes and figure out if these variants are pathogenic, likely pathogenic, likely not pathogenic or benign.
Very cool ... thank you!
I do want to sequence it at home but I'm gonna wait until it's under 500 bucks
https://www.the-odin.com/whole-genome-sequencing-30x/
If you want it quick and cheap(er) - 599.00
If it's an US-based lab, aren't they subject to CLIA with all its retention requirements?
For $7.5k+ you get a guaranteed privacy (as other comments suggest, other properties may vary, but at least the data never leaves your home).
I suspect there is a deep sequencing service that is non CLIA and cheap. True. they may not be trustworthy with the data. That said, there are steps here where the data is put into Claude. Do we trust that ?
That's another vector, and everyone considers separately. But at least here one can - hypothetically - have a BAA and zero-retention agreement with Anthropic. Which means they have moderately strong incentives to wipe your data after a fairly short while. CLIA, if I don't misremember, mandates a few years at minimum.
I would never trust that. Instead I would use Claude to teach me genomics and build the tools to process and interpret data locally
It's not impossible: Claude can take care of bioinformatics part. But to understand genetics and cellular biology you need a knowledge foundation first.
Unless you just want it on a level "Does this mutation leads to a genetic disorder X" - this is a simple way to put it, but not enough to actually understand genetics.
haha. hubris thy name is tzumby.
look up what isaac asimov had to say about genomic analysis.
A service is not the same as the equipment
Has anyone succeeded in doing any of this? I always assumed if it were this simple, someone would be selling DNA kits on Amazon for a few thousand dollars.
Per the article, ONT sells tiny sequencers that all this genre of "Omg I just sequenced my genome" posts use. They've been around for a decade or so. A commercial lab will give you 30x illumina sequencing for maybe 5% of the cost of this post.
As an 'at home' science project? Totally do-able, but very expensive. I did it as part of epidemiological field with a backpack of materials, so you can definitly do it at home.
As a useful genome? Marginal. ONT will need a couple runs to get complete human genome coverage and the substitution error rate is still high compared to short reads. For a fraction of the price and effort you can get your DNA resequenced at 10 times the quality.
I guess if you are super-ultra paranoid you can do it all yourself at home, but honestly. You're not that special, and if someone really actually wanted your genome that bad they'd just take it from a used coffee cup.
This is so cool. Thanks for doing this. The fact that we have this in a palm sized object is just crazy. Also, if/when we have a similar sized device for doing CRISPR .... umm i should stop here - it's becoming the plot of Gattaca
As someone with experience (albeit almost 20 years out of date) experience of wet lab DNA collection and sequencing - this stuff is hard, you will fail a lot, and you will fail a lot more if you don't have an extremely clean environment to do this in. And once you have data, you should be asking yourself how accurate it is given the environment you collected it in, you should be looking at correlated sequence errors that are not taken into account[1].
But also: genetic counselling is a real thing that real people study. Please don't ask an LLM questions about what your genes are going to do to you without having access to someone who has the ability to contexualise the data and put you in touch with relevant experts. I have a PhD in this and I would not trust myself to be able to interpret data about myself in a detached and rational way.
(And: why is the link to Molecular Biology of the Cell to the 6th edition, when the 7th came out 4 years ago? Random fact: the first three editions were co-authored by my supervisor during my first PhD attempt, who went on to demonstrate that Roger Penrose's ideas about the importance of microtubules in chemotaxis in E. coli were absolute bullshit. Great guy)
[1] I spent a while analysing very early (by commercial standards) Illumina data in 2007, and being able to align stuff to reference genomes made it possible to identify certain biases. Nanopore technology is likely to have more of those, and if you don't have the ability to take those into account you may have a very bad time
Nanopore data is a lot easier to analyze than short read sequencing data. You just don’t get the same alignment/assembly issues: these things sequence incredibly long reads.
(also a biochemist, MSc)
> this stuff is hard, you will fail a lot, and you will fail a lot more if you don't have an extremely clean environment to do this in
The Oxford Nanopore sequencing technology is one of the most robust to use. You need to buy some kit - but defo doable. Nothing compared pouring your own gel and doing radioactively labelled Sanger reactions :-)
Though you could just go to an sequencing company that services labs ( that just does sequencing outsourcing - rather than a personal genome company ).
Totally agree on the dangers around interpretation.
It might make even more sense once we get to the point of a wider use of encoding the data into dna. For now we have these few commercial players in the field that cad do it (eg look up dna microfactory for storage archiving), IIRC genomika was saying they can do an MB for a 100-200eur.
Is this the same guy behind https://iwantosequencemygenomeathome.com/ ?
[Edit: Apparently no. That link is by Seth Howes, who also shared the OP post though: https://x.com/SethSHowes/status/2074231119730430203 ]
It seems a bit odd to me to have a home lab for the sequencing and then feed it to a cloud LLM. I would have thought the point of the home lab is privacy.
Not really. It is to learn.
How can I know if the the results I get are real or just some garbage?
via comparison to other nucleotide sequences. This is called sequence alignment: https://en.wikipedia.org/wiki/Sequence_alignment
Run it multiple times, and then get professionals in on it as well, and then compare.
> Oxford Nanopore Technologies MinION ($7.5k)
Hmm, I'll come back in a few years when things have become cheaper.
It's been around for a decade and was $1k in early access, so don't hold your breath.
I like the privacy conscious aspects. Apart from the obvious issue of "run it through Claude" how many of those referenced analysis tools are entirely open source or at least run locally? Would have liked to see that in the article.
At a quick glance, they all seem to have published their source code and they do run locally.
If the sequencer was 10x cheaper then I might do it.
I'm waiting for someone to make an open source gene sharing platform to rival ancestry etc in the future.
There must be some cool way to share enough structure with some cryptography to share parts of your dna to find relatives etc
Bringing everything to your doorstep and everything at your feet and everything near your fingertips is just what all industries are trying to accomplice. The cartoon Animation Wall-E has scenes in it where they show obese humans doing everything through a screen though notice their legs and feet and it's as if they've mutated to a point where they aren't able to walk anymore and all their transport is through a hovering chair cum bed.
I watched wall-e in theater, and when that scene came on, i remember muttering 'what bs'; since then, i recall that scene every time i see a situation of 'convenience at all costs'; metaphorically they were pretty accurate even after discounting ozempic influence;
With such cheap costs every city should be sequencing dog turds and sending out big fines. The pay back would be very quick.
I think cameras or even just a reporting hotline would be easier. The problem is lack of motivation I think rather than a money thing.
It's quite a bit more difficult with poop. A rather large proportion of the DNA in poop isn't from the creature that pooped it. (mostly bacteria)
And you'd need the DNA of all dogs to boot (unless they are already somehow stored when a dog is registered). But generally speaking, I do like the idea. Also for horse poop which is pretty common on roads here in the countryside.
What is the accuracy in this ? Aka if I run the experiment 10 times how many differences will i get? I don’t have a physical sense on what would be a good number.
You would get a lot of differences, but the errors would cancel each other out with enough depth of coverage.
This technology's baseline accuracy is around 95% per base, so 10x reads of every segment in the sample would give >99% accuracy for each base after aligning the reads with each other.
https://en.wikipedia.org/wiki/Coverage_(genetics)
> so 10x reads of every segment in the sample would give >99% accuracy for each base after aligning the reads with each other
This assumes random errors, which IIRC isn't the case for Oxford Nanopore.
Oxford Nanopore unfortunately has a high error rate (3-5%) compared to other sequencing technologies. And the errors are non-random
This is simply not true, advances in basecalling and using Duplex give you >Q25
For most DNA it now sits at ~Q20 (1%) which is a big improvement - there was also some hacky stuff like HERRO and duplex sequencing which improved it substantially. For routine genome sequencing it's definitely suitable, and the long-read advantage helps resolve some of those trickier regions
one main marketing leverage of 23andMe, AncestryDNA, etc are fulfilling the curiosity of people who want to know which part of the world their genes are from. I guess that dataset should be preparatory.
Problem with those providers - they only check 700K positions out of 3 billion and there is no mapping quality or allelic depth data in those dataset and this is critical for assessing whether the detected variant is a false positive or real.
It's not suitable for health investigations since most of DNA is not sequenced and genotyping technology is known to produce high rate of false positive for rare mutations.
(I'm the solo-founder of Gene Inspector Pro, mentioned in the blog post). AMA. :)
It's a bit ironic, since FTDNA and MyHeritage (which uses FTDNA's lab) have switched to NGS now, so they presumably could deliver those notorious "health insights", at least better than 23andMe. But they aren't in that market, and 23andMe shows no inclination to switch. They're probably licking their wounds after the user hack and buyout fiasco.
I'm too afraid I would learn something awful about myself.
That's unlikely, but I do think that the health benefits of full sequence testing yourself are largely hypothetical at this point anyway. Unless you're a competitive athlete maybe?
It's genealogy it's useful for. But genealogy, it's really useful for.
Unfortunately you’d have no power to correct it. Even if such a thing were possible. I hope that changes.
You have no power to correct something with gene editing, true, but there may potentially be preventive action you can take. It's all largely hypothetical today anyway, most things you care about (e.g. cancer, Alzheimer's) you can't usually tell very reliably if you're going to get from a DNA test.
Are you talking about time travel? I don't think that would help.
More like genome editing.
Reminds me of the Gloing Plant Project. I never got my glowing flower but would have settled for the instruction manual, also never created :(
https://en.wikipedia.org/wiki/Glowing_Plant_project
By the by, can't seen to bring up the actual site linked on this post.
You can get one here, you'll have to wait until next year though: https://light.bio/
I am very impressed with the, why wait? just do it now approach to the future. which while not here, IS there.
Nothing about this is the future. Sequencing at home will not solve any major problems. It's mainly a fun exercise to demonstrate that sequencing has been commodified.
I disagree, I sequenced with nebula genomics years ago.. you can understand risk factors for various problems so that you can start interventions that make sense way in advance.
I don't wasn't someone else to have a copy of my genome, whilst wanting to analyse my genome. The only way to do that is with affordable home sequencing.
just a hobby project for now,pretty wild that this can be done at home.
Knowing exactly why I have high LDL because of a specific mutation on my DNA is very much the future, imo.
Can CRISPR also be done at home this cheaply?
Seems like maybe of the 3 dystopias: AI, Global Warming, Bio-warfare. That this is demonstrating that the home grown virus is closer than we think.
yes https://www.the-odin.com/blog/how-to-do-crispr-at-home-begin...
This feels like the acme of narcissism. How much time and money are people willing to spend on navel gazing?
True story, I found a god damn tick in my navel yesterday. Sometimes a bit of navel gazing can be healthy, figuratively and literally.
On a reasonable level, navel gazing (the figurative kind) is maybe better called self-reflection. I use DNA for genealogy, and it seems to me from the people I meet, that many get a healthier approach to our identity once we learn more about our genetic background. Identity politics, collective identity building around ancestry - identity building of all kinds really - needs simple stories. And the stories DNA tell are never simple.
That’s quite judgmental and shows an impressive lack of scientific curiosity.
When you are taking care of your health, and need to learn more about your built-in limitations is it still narcissistic?
At the bottom of the page there is a link to Sid Sijbrandij's cancer journey. He is one of the cofounders of GitLab. This is one of the coverages of his story: https://centuryofbio.com/p/sid
I don't get you at all.
Is going to your doctor or eating better food "navel gazing"? Predispositions to some diseases can be read from your DNA. Remember Angelina Jolie undergoing preventive mastectomy because she had a high genetic risk for breast cancer? Well, so do many non-celebrities.
Then there is the specific case of people who may suspect that their bio-parents are someone else, and there is nothing weird about wanting to know where you actually come from.
> This is intended to be read by AI
Fuck this
Man, doctors thought they had it bad before. For just a six yards I can play Peter Thiel at home! $6k invested so I can set an AI in YOLO mode to tell me I have some hyper-specific version of kennel cough?
“But that occurs in dogs?”
“You’re right. Let me look into actual gene sequencing instead of just guessing. I think the N is the load bearing letter.”
Funniest thing I read today, thank you!
Hi, author here- you can read it too, though it is dense. I have updated that specific sentence.
I found it easier to upload the protocol to ChatGPT and have audio walk you through it. This allows you to swap between pipettes, measurements, etc without having to look at the screen, reducing context-switching
Yeah that's weird. The instructions are not even hard to read. I don't understand what an LLM would add to this.
As long as the AI doesn't brush its teeth all good.
Literally left the article to come here and say this.
> The near-term value is turning a static genome into something queryable
Ok. So ... how exactly is this valuable?
If you realise "hey, I gots Huntington disease", this is going to make you feel better? Or any other incurable disease? I am not disputing that knowing the sequence is useless in general, mind you. I am specifically asking WHY it is necessary to know your genome sequence. This seems to be a simplification or just a "having reached a milestone". But then they don't really explain WHY it is useful. None of the bulletin points he listed is really useful:
> Which variants do I have?
And this is useful ... how exactly?
> Which genes and pathways are affected?
And ... this matters why?
> Which medicines might I metabolize differently?
Ok, so this has a potential use case here, since he can choose to avoid specific drugs. How useful that really is in practice is unclear. (Don't confuse drug companies trying to convince YOU that personalized medicine is important on THEIR use case.)
> What rare variants should I take seriously?
Seriously ... how? Ok, you avoid some compounds. Now what.
> Where does the model know nothing yet?
Great, so a model that is limited, but now I need to burden myself with having to know where that limitations are. So my brain just has extra processing to do, without getting anything useful in return.
> the “edit yourself with CRISPR” will most likely follow
Except that they have not solved the off-target cleavage yet. Besides, they milk the prices anyway. DNA manipulation should be safe, secure, correct and affordable. None of that is the case right now. They publish papers where CRISPR has solved everything, but then fail to explain why it isn't already used by billions. And there are reasons as to why.
> Give your genome to Claude Code
Oh my god ... AI becomes your dependency here.
Note that the step-by-step guide is actually not totally useless, as it can give a basis for real work. But I highly doubt that untrained people will easily be able to go through those steps. Everyone is a master in the lab now? RNA is easy to handle? Guess then one would have to explain why RNase A is used (ok ok it's not playing a huge role here since DNA is the target of isolation, but it is more of an example of how many things can go wrong, and there is not really an explanation of why xyz is used; this looks like an AI step-by-step guide. AI really makes people dumber).
The "non-random errors" point buried a few replies down deserves to be the headline, not a footnote. With Illumina, 10x coverage genuinely washes out errors because they're closer to independent per-read noise. With Nanopore, errors cluster at specific motifs (homopolymers, certain k-mers) due to how the pore physically reads the strand — so the same systematic mistake shows up across most of your reads at that position, and naive majority-vote consensus won't fix it. You need a basecaller/consensus model trained to correct for those specific failure modes (which is exactly what the current-gen Guppy/Dorado models try to do), not just "more depth." That distinction matters a lot for a home setup: coverage is cheap, but knowing where your specific errors are systematic vs. random is what determines whether "buy more reads" actually gets you to clinical-grade accuracy or just gives you a very confident wrong answer.