Launch HN: Probably Genetic (YC W19): At-home DNA testing for every gene

[astroH]

Hi Hacker News, We’re Harley and Lukas from Probably Genetic (https://www.probablygenetic.com). We built an at-home physician-ordered DNA test that covers all genes and looks for pathogenic variants related to thousands of rare genetic conditions.

Why rare genetic conditions? It may seem like a niche problem but there are ~400 million people worldwide with a rare genetic condition, half of which are currently undiagnosed. To put this into perspective, it’s estimated that 1 in 10 Americans has a rare condition and while each of the individual ~10,000 conditions is rare, the population that suffers such conditions is larger than cancer and HIV combined. Furthermore, for the patients that have been diagnosed, it takes on average 8 years for doctors to identify their conditions. You’ve already heard of rare genetic conditions, you might just not be aware of it. Remember the Ice Bucket Challenge for ALS? Most of these diseases initially look like more common conditions, such as autism, chronic pain, ADHD, or even the flu, before patients get worse. This diagnostic odyssey can be extremely costly for patients and in our experience, some are spending more than $30,000 and seeing more than 10 doctors before they get access to the right specialists and testing.

We have seen this problem first-hand. Lukas is a rare-disease expert and worked on the world’s largest rare disease project (the 100,000 genomes project) as a PhD candidate at Oxford University in the UK. I am trained as a theoretical computational astrophysicist and during my PhD and fellowship at Cambridge University and Oxford University, I spent my spare time working with National Healthcare Service doctors developing and publishing medical diagnostics with machine learning. Our original idea was actually slightly different from what we have now. We spent a lot of effort on developing a symptom checker specifically for rare conditions with the idea to comb through existing medical records and flag patients with potential rare genetic conditions because, unsurprisingly, WebMD and others aren't really great for this purpose. As we were building this, we realized that for the patients we worked with, even if their symptoms were suggestive of a genetic condition, access to clinical-grade genetic testing was extremely difficult for many as it was either too expensive because insurance wouldn’t cover the cost, or they couldn’t find a doctor that would order it. Thus, we decided to use our expertise to both find these patients more efficiently and built up a service to drastically reduce the time and cost to access clinical-grade genetic testing.

About the test:

Just like most DNA tests, you can do this from home and it’s noninvasive, all we need is a little saliva. Unlike most DNA tests, ours is physician-ordered, sequenced in a CLIA-accredited and CAP-certified lab, the results are signed out by a licensed clinical lab director, all users have access to genetic counseling, and we try to incorporate as much phenotypic data as possible into the analysis. Our product is a whole-exome sequencing test with 100x coverage and covers all of the more than 20,000 genes, where 85% of known disease-causing variants occur.

People always ask, how are you different from 23andme? Looking for a rare genetic condition is kind of like trying to find a typo in a novel. Using a 23andme (or similar) test to look for such a condition is like trying to find a typo in the first Harry Potter novel and stopping after 75 words. Those tests are just not meant for this purpose. Most are based on genotyping arrays that look for very specific variants at predetermined locations in the genome. However, the variants that cause rare diseases can occur anywhere. For example, there are over 1,700 different mutations in the CFTR gene that can cause cystic fibrosis. Approximately 85% of the known pathogenic mutations occur in the protein-coding regions of DNA called the exome. Our test is a whole exome sequencing test rather than a genotyping array, which allows us to cover all of the genes in a person’s DNA. We often get the question, why not do whole genome rather than whole exome? Right now it simply comes down to accessibility. For most consumers, whole genome sequencing is still too expensive and the additional gain in terms of coverage of pathogenic mutations doesn’t necessarily warrant the significant price increase. That being said, if you are interested in clinical-grade whole genome sequencing, we can soon offer this as well.

Patient privacy is a huge concern for us and something we think about all the time. Quite simply, we will never share any of our users' data without explicit consent and we are more than happy to both delete our users' data and destroy their sample if requested.

Interestingly, we often have the opposite problem where we receive inbound requests of people trying to share their data with us to see if we can help them. Out of the more than 10,000 rare diseases, over 95% do not have an FDA-approved treatment, which is why the rare disease community is so motivated to leverage their personal insights. We have started a waitlist to provide such services and are actively seeking ways to help these people and integrate them into our community, even if they have not had sequencing through us.

Finally, how much does it cost? A single test right now costs $899 on our website, but we try to offer the test in trios where we sequence both the patient and two family members as this often gives a higher diagnostic yield. The latter option is $1,799. We expect that the test price will decrease significantly with time as the cost of sequencing drops and more of the analysis can be automated. We don’t currently accept insurance; however, in our experience, using more traditional channels to access this kind of testing can result in bills of >$10,000, not all of which may be covered. Many insurance providers don’t even cover this kind of testing, except for very specific purposes, despite more and more of the medical literature recommending exome sequencing as a first-tier diagnostic for specific indications. Ideally, we would make the product so affordable that it simply makes sense to use us rather than billing insurance for the test plus the numerous doctor and specialist visits needed before and after it’s ordered. We are currently offering the test at cost, as we aim not to profit off of the patients that need it most, and this is sustainable because, with the consent of the patients, we can leverage our data asset for drug discovery, clinical trial recruitment, and drug repurposing.

Consumer genetic testing is growing rapidly as an industry and nearly doubling every single year. What is missing from this market is accessible physician-ordered testing that can genuinely help those with complex symptoms and undiagnosed genetic conditions. This is what we hope to provide. If you have any questions or feedback, we would love to hear it and please check us out at https://www.probablygenetic.com.

[mattparmett]

Congrats on the launch! I'm a VC that's invested in this space and I think you are hitting on an important distinction between physician-ordered and consumer-directed testing (we often call the latter "medical tourism").

Another very important part of the testing process, particularly in the rare disease space, is the pre- and post-test genetic counseling. I see that all users have access to genetic counseling, which is great - just curious about how you are handling that part of the process?

[astroH]

Thanks so much for the comment. That's a really great way to describe the distinction! In terms of genetic counseling, for post-test purposes, we won't release positive results until the patient has booked a session with one of the genetic counselors where the results are released. In the case of negative results, we give the patient the option to discuss with a counselor but do not make it mandatory. In both cases, the counseling is included in the cost of the service. For pre-test counseling, the way it currently works in our service is that when users request a test, they have to fill out a detailed medical intake questionnaire which is then reviewed by both a genetic counselor and a physician. Right now, it is up to the discretion of the genetic counselor that reviews the information whether they feel like they should reach out to the patient and discuss how this may impact them or whether they have requested the appropriate test. If demand increases significantly for pre-test counseling, this is definitely something we will able to add moving forward.

[all2]

> Patient privacy is a huge concern for us and something we think about all the time. Quite simply, we will never share any of our users' data without explicit consent and we are more than happy to both delete our users' data and destroy their sample if requested.

This should be the default rather than a user-requested option. I would consider auto-deleting user data (ensuring privacy for anyone's data outside some window of time) to be akin to auto-opt-in organ donation: most people don't know or care enough to change the default and will go with whatever is on the form (which is why organ donation is 99ish% of some populations in Europe and drastically lower than that in the United States - we should ignore the moral implications of choosing for people and focus on the outcomes of choosing a 'sane' default action). If you want an iron-clad guarantee sequences will never fall into the hands of bad actors (anyone who has delusions of genocide, for example) deletion/destruction is the way to go; perhaps even two or three passes of bit-by-bit overwrite.

Consider a future where such sequences will allow bad actors to engineer viruses against family units, regions of peoples, and so on.

[rkagerer]

I'm interested in more detail on your privacy policy, and reading any contractual language on this subject in your Terms of Use (or whatever other fine print forms the agreement between you and your customers).

I stay far away from services like 23andMe because of their user-hostile terms (at least that's the impression I got when I last scrutinized them a few years ago).

To me the ideal DNA test would be one I process myself at home, without sending anything to the cloud. I recognize that's not viable since most of us lack home laboratories.

Next best option would be one where the sample is submitted anonymously and results accessible under a system-generated ID code. That would alleviate some of the risk if you or your partners screw up in terms of privacy and security (it's naive to ignore that possibility - look at how many Fortune 500's have been breached or suffered human error).

A user could "opt in" to share identifying information (ideally only to the extent they're comfortable - e.g. I might be willing to submit coarse data like racial background, age, etc. while withholding my name and address).

In all cases I would still expect to see very strong, binding language in both your agreements with me, and those with the labs you subcontract out to, assuring that my data will be kept confidential and all traces deleted (including from all backups!) when I request.

Regarding re-sequencing and long-term storage for privacy concerned users: give me the option to download the raw data (compressed and encrypted?) and delete it from your platform. I can re-submit to you in a few years for an updated analysis. That avoids the cost of re-sequencing while, for privacy-conscious users like me, minimizes the window in which you storing my data could cause me harm.

[astroH]

I think that's a really good point. In fact, for many lab tests where the results are completely "cut-and-dry", this suggestion totally makes sense. For us, one thing we consider is that the diagnostic yield for these types of sequencing tests is only ~40%-50% in the best case. We know from the literature that if we perform a reanalysis 6 months-1 year, the chances that we find something in patients that don't have a result increases quite substantially as does how we interpret known variants. Thus, continual reanalysis is very important to many of the patients that we work with. Unfortunately, it's too expensive to resequence every time we need to run such a reanalysis algorithm.

[est31]

Also that way the data are safe if the company gets acquired by someone like 23andme or pivots away from this path themselves. The statement that they care about privacy is really great, but it's harder to actually live it.

Anyways, wishing them lots of luck in building a company that actually respects privacy and doesn't just claim it.

[selimthegrim]

So you're monetizing what these guys are doing for free?

https://stanfordcehg.wordpress.com/2017/06/22/feature-interv...

[lukasl]

That's a really interesting post! I hadn't heard of them and just had a look at the website. I think the focus is here is more to first collect survey data on patients and, once sufficient survey data is collected, run a research study on specific diseases, but I might be wrong. That's different from our service. We're providing a fast and easy-to-use genetic testing service, not a research study. I also couldn't find an option on their website to actually get testing, but I'll keep an eye out!

[text70]

Not to mention: https://nebula.org

Whole genome, not just the protein coding regions, which can catch more clandestine promoter, and non-coding SNPs, etc.

[zmmmmm]

How thoroughly have you compared yourselves to existing options like Invitae? They appear to offer pretty much what you are describing for a similar (slightly higher) price [1].

I feel like you may be underestimating both the maturity of the existing market and the complexity of doing what you are proposing at scale. Are you issuing clinical reports for your results? If so, who is doing that as neither of you sounds like you would meet the normal requirements for that.

[1] https://www.invitae.com/en/physician/category/CAT000168/#exo...

[astroH]

Invitae is definitely one of the most impressive companies in the entire genetic testing space. There are many companies that can offer exome testing, but as far as we are aware, in almost all other cases, you still need your own physician to order and you would be surprised how many patients struggle to get their doctors to order genetic testing, especially because in many cases insurances are not reimbursing. Our test is in CLIA-certified and CAP-accredited lab and is physician-ordered. It's important to note that we do not make medical diagnoses we simply provide the report that you can take to your doctor and have them interpret in the context of your other medical history and information. In essence, it's a very similar service but in our case patient-initiated. The report is signed out by a clinical lab director in connection with our sequencing and bioinformatics partners.

[gavinray]

This is incredibly cool, really amazing to see full-genome sequencing versus the very limited number of SNP's that other players in the space are doing.

I am working with a team of people at the moment building a product around genome analysis and insight for people who have existing sequences - sort of like Promethease but more accessible to the average person.

If you are open to it, I would love to get in contact with your team and learn more about what you're doing. Is the email address [email protected] the best place to reach out?

[astroH]

A service like that is definitely needed! Please feel free to get in touch via the contact address.

[blue_eyes1978]

is the actual basepair sequence shared with the patient? How much of the chromosomal sequence is actually mapped? I'm interested in how this may be incorporated into a patient's chart to fully flush out future diagnoses as sub-clinical symptoms progress down the pathophysiologic maturation chain.

[lukasl]

Great question! In addition to the result report, we share three different types of files with the patient:

(1) The FASTQ, which contains the read data from the sequencer (see https://support.illumina.com/bulletins/2016/04/fastq-files-e...) (2) The BAM, which contains sequence alignment data (you can learn more here: https://software.broadinstitute.org/software/igv/BAM) (3) The VCF, which contains the variants present in the patient as compared to the human reference genome

We map all >20,000 genes in the human genome, which covers 85% of all known disease-causing mutations. Genes make up about 2% of your DNA.

Regarding integrating this into ongoing healthcare for patients: continuous re-analysis of a patient’s genetic data is important, for two reasons: (1) we discover 100-150 new disease-causing genes every year, which means we can return results today that we couldn’t a year ago, and (2) a patient’s symptoms change over time, which matters because we analyze the patient’s DNA for what’s causing their symptoms. Plus, a lot of diseases have a late-onset, which means you might be perfectly fine until your thirties, and all of a sudden you start developing a condition that you didn’t know about. Adult-onset ALD is an example of such a condition (https://www.mayoclinic.org/diseases-conditions/adrenoleukody...).

[kian]

Where are you located? I was under the impression that many states in the US mandate that you hold on to the genetic data for them, and thus that you can never truly delete or remove it.

[cflyingdutchman]

SF

[ablekh]

Congratulations on your launch and good luck! I'm curious about your participation in YC accelerator and would very much appreciate if you could reflect on it (interview, pros, cons, impressions, etc.). I'm especially interested in this due to my current (passive) work on a deep tech (science-based) startup, albeit in a totally different domain, and it would be great if you could reflect on your YC journey from a deep tech startup's perspective.

[astroH]

YC was immensely helpful for us in a lot of ways. Definitely happy to go into more detail offline but two things I really liked were that there are so many companies in the healthcare space now that have come out of YC and regulation is a hugely scary aspect of the field and getting advice from them on how to navigate many things saved me so much time and headache. YC also very much redefined how I personally think about an MVP and it's usually a lot more minimal than I initially realise. As the community grows bigger, a lot of this will become even better. I almost wish we were slightly further along when we started as a lot of the prolems that companies in our batch were facing are starting to affect us now that we're not in set-up mode. The key thing for me is that we have to make sure we do right by patients above all and I have to integrate that with all of the other advice that sometimes conflicted in my head.

[ampdepolymerase]

What about federal lab requirements to store genetic samples?

[astroH]

Great question. We use a CLIA-certified and CAP-accredited lab, and we must keep the specimen through the completion of the testing and result reporting. If there is any sample remaining after testing, we will hold onto it longer in case we need to run a validation or other tests as recommended, but we will, of course, have it destroyed at the request of the patient. It's recommended that we keep them for as long as they are stable.

[quantguy11959]

Thanks for the comparison against 23 and me but I’d be interested in how you plan to differ from Invitae/Fulgent/Myraid aside from the D2C aspects.

[astroH]

Great question. Right now the major difference is the D2C (patient-initiated, physician-ordered) aspect. Invitae is starting to offer this to some extent but they mainly seem focused on cardiovascular and cancer. Unlike those companies, we don't consider ourselves a testing company. For most patients in the rare disease space, understanding the underlying genetic cause is really only the first step in the process and then most need tons of help navigating the healthcare space afterwards. We started with the test, mainly because of the access and affordability issue patients were having. Many of our patients have requested tools for longitudinal phenotypic data collection, connections with all of the relevant advocacy groups, further genetic counselling, efficient ways to test family members, and detailed, patient-readable information content about their variant. We try to provide everything these patients need (aside from medical advice of course) as the healthcare system isn't currently set up to help rare disease patients.

[aloukissas]

"we will never share your data without your consent" - bravo to you guys for not going the route of 23andme etc.

[aroch]

In fairness to 23andme, they also don't share your data without your consent ( Individual Data Sharing Consent : https://www.23andme.com/about/individual-data-consent/ and population-based Research Consent: https://www.23andme.com/about/consent/).

It just so happens that when they do share your data it may be with a Pharmaceutical company

[astroH]

Cheers! :)

[the_watcher]

Do these tests qualify for FSA/HSA/MSA reimbursement? If so, you should highlight that, particularly in November/December when people are making their elections for the next year (or in the case of an FSA, trying to figure out how to make sure they use their entire balance).

[astroH]

Really great suggestion. We haven't tried accepting this yet but naively, I would imagine they should be. I wish I could give a more concrete answer but I will further look into the details and ask our in-house GC.

[the_watcher]

As an anecdote, the first year I had an FSA, I didn't fully realize that it was use it or lose it and ended up spending $1500 on things like sunscreen, contact solution, and first aid kits.

[neil_s]

So glad to see this service exist. While Shkreli is out jacking up prices for rare disease drugs, you guys are making help more accessible by providing doctors trained to order these tests, running the tests at certified labs, and being super transparent with data sharing.

Cheering for you to succeed!

[astroH]

Thanks so much for the kind words and support!

[aaauaucuggaa]

Super exciting launch! Always good to see more bio companies in YC. I applied for an engineering position and would love to know more about what challenges are being tackled on the software/bioinformatics front.

[colsandurz]

Would you ever be able to test two potential parents to see if they're both carriers for genetic diseases? Or, is that a completely different type of test?

[astroH]

Yes, we could test two potential parents to see if they are carriers. Depending on what you're looking for though, if it's a specific mutation, there might be a slightly cheaper option on the market as our test is very broad spectrum.

[elektor]

Harley and Lukas, this looks like a very promising start-up and especially timely for Rare Disease Week. Congrats on the launch!

My question to you: You've made a comparison of your service to 23andMe in your post, can you provide some similar insight as to why your service would be preferable to Dante Labs? They claim to provide a "Whole Genome Sequencing Full DNA Test (30X)" [1]

Thank you!

[1]https://us.dantelabs.com/collections/advanced-dna-tests/prod...

[astroH]

Thanks so much for the comment! In order to offer this testing and give back a result report that can really be used by a doctor, it should be physician-ordered, processed in a CLIA-certified and ideally, a CAP-accredited lab, have the lab report signed out by a clinical lab director, and ideally (for ethical reasons) offer genetic counseling to patients. From the bioinformatics perspective, the bioinformatician should be trying to take in all of the medical information we know about the patient and apply this to the analysis. This is what we offer. In contrast, with Dante labs, you basically send them your sample and get back raw data and some very basic reports. If you bring these to a doctor, the recommendation might be to just get another genetic test via a lab that follows all of the procedures as stated above so a person may end up doing it twice if they go that route. Ours offers the complete end-to-end solution that you would get in the healthcare system while Dante labs is really just a way to access your raw DNA.

[rootusrootus]

Actually delivering results would make them extremely preferable. I've been waiting on Dante Labs since May of 2019. After complaining to them later in the year about the lengthy delay, my status magically changed to 'sequencing started' and has been there for a few months now.

I have given up on ever getting results from Dante Labs, and I'm unsure if I would even trust the sequencing results either way. I recommend anybody else considering them do their own due diligence before making a purchase.

[CodiePetersen]

Great explanations. Hope you guys do well. I do like the fact you delete and destroy data and samples.

[lukasl]

Thank you!

[cflyingdutchman]

I read your FAQ but still couldn't tell if you test for Scheuermann's Disease Orpha:3135.

[astroH]

So it's important to note we don't test for diseases but rather variants which may be the etiology. I'm not an expert in this condition by any means but a quick search says we still don't know the candidate genes for this and many have been excluded which would be problematic for our test.

[p9s]

Congratulations on your lunch, we do hope you will take data privacy seriously.

[lukasl]

Thank you!

[spease]

Can this test detect repeated sequences, eg as in Huntington’s disease?

[astroH]

Our test won't detect Huntington's unfortunately.