Great! This is why I went back to school to get my biochem degree. I was a data scientist working with hadoop + mahout at a startup and I said "this is going to revolutionize medicine." When I get my masters in biomedical engineering in a few years I have a dream to help found a company that applies big data and machine learning to problems with clinical significance. Elizabeth, you may end up hearing from me :P
This is an amazingly hard problem and I wish you all the best.
One thing you won't learn at school is all the complications involved in validating a target. The best resource I know of for this is the "in the pipeline" blog by Derek Lowe. I have spent 100s of hours reading all his posts (they go back 10 years) and especially the comments. It is a hard slog, but the amount you will learn is staggering.
I feel an interesting and charged buzz around this thread, so though to join in.
Things appear hard and severely constrained because we habitually fall into the trap of thinking linearly and incrementally, as opposed to, at least sometimes, multi-linearly and in bounds, by perspective shifting. To do that one has to question the fundamental assumptions of which this linear trajectory were founded on. More often than not, one finds that much of what is considered so is merely based on someone', usually a group's very convincing, albeit not necessarily correct narrative.
Combining the scalibility of software startups with biotechnology is a very interesting approach, indeed. Apart from the different timeframe when it comes to working with software and wetware (e.g. lab work with biomaterials), I wonder how easy it will be to get people and know-how on board. In Europe, young scientists (speak: PhD-slaves) are often bound to their research institutes when it comes to the exploitation of the IP generated by their research. They may end up with publications or patents, but usually it is the partners from industry that bring the product to the market. How does this compare to the U.S.?
True, 'cause one drug for one condition now costs $4-12B, because the success rate is < 0.02%. There needs to be multi-functionality in our innovative efforts when we experiencing such high rates of disruption, say, extinctions.
Generally, the "target" is the naturally existing cellular or molecular structure involved in the pathology of interest that the drug-in-development is meant to act on.
"The thing I never understood is: shouldn’t the biotech community be happy that more people are trying to do this and that more people are funding this work?"
It's a world where lots of great ideas are underfunded (or not at all), because the traditional government sources don't have the money needed to invest in every good idea. Also, it's still generally too expensive to chase every possible business idea in biotech, and there are plenty of ideas with academic promise that don't get anywhere because the traditional sources of investment are too conservative. Scientists get punchy when they see funding going to "unproven" entrepreneurs, while they're slaving away at research that is peer-reviewed, but isn't getting any investor attention.
It's good to see YC bringing in outside expertise for this area, because the likely outcome for amateur angel investment in biotech would be investing in lots of flashy-but-insubstantial entrepreneurs, while legitimate research never gets off the bench. In my experience, it's much harder to evaluate biotech ideas than software.
A funny part of the history of biology: NASA established its Life Sciences office in 1960, and the following decades they funded some pretty influential research in molecular biology and on the origin of life, which was not getting funding from more conventional biology funding sources.
Maybe the highlights would be Lynn Margulis and the endosymbiosis theory, and Carl Woese finding the Archaea in 1977.
References:
JE Strick. 2004. Creating a Cosmic Discipline: The
Crystallization and Consolidation of Exobiology, 1957-1973.
Journal of the History of Biology 37:131-180.
My business straddles both biotech and software and I find it much easier to evaluate good biotech ideas than software ideas. The problem is the money and time required is so much greater with biotech that almost none of the good idea get invested in.
Agreed about the money and time, but I do think biotech ideas are harder to evaluate, if only because the basic knowledge required is far more specialized.
Your average CS undergrad can understand the technical risks of most software startup ideas, but you need at least a graduate degree or equivalent industry experience to be able to evaluate the technical feasibility of lab research. Moreover, life science research depends on sophisticated intuition that can't really be taught in a classroom.
I agree the barrier to entry in biotech is higher than software - in the area of bioinformatics it is usually easier to train someone with a life sciences PhD how to write software (bad software usually) than it is to try to train a CS graduate in the required life sciences. However once you have skills in both area the biotech ideas are easier to evaluate. I think this might be due to there being more good biotech ideas around because far fewer can be exploited because of the cost.
The issue isn't so much YC's investment in projects that may not have scientific rigor, but that of individual investors via crowd-funding platforms. At an extreme, this results in a company like GoBe raising $1.1 million on IndieGogo with PR and a slick video.
YCombinator has the sophistication and time to vet their investments. If they want to back something most of the scientific community thinks won't work, all the best to them. However, raising capital from mom-and-pop investors is another story.
Let me suggest the mitochondrion as the main target to explore:
- In a very simple but accurate analogy, a computer will not work properly - if at all - if the power is not coming. You can have the best hardware; you can put endless hours to develop the best possible software, but nothing matters if the power is not coming.
- Mitochondria produce 90% of the energy - ATP - we need to function.
- Mitochondria are a relatively simple structures. Its DNA is way smaller and simpler than the nuclear - double helix - DNA.
- Many health issues seems to be the result of a primary mitochondrial dysfunction, or
- Many health issues will result in a secondary mitochondrial dysfunction.
I do have a mutation in my mtDNA (mitochondrial DNA), so
- I know what I am talking about and
- Yes, I am very biased; my life depends on this.
Mitochondrial diseases don't seem like an easy target because they are really metabolic and homeostatic disorders, which are very complex. If the main failure of mitochondrial disorders were a lack of energy production, how could cancer cells grow faster than regular cells, yet turn off mitochondrial energy production in favor of anaerobic glycolysis and lactic acid fermentation (which is much less efficient)? Mitochondria have an under appreciated role in ion homeostasis, intermediary metabolism, and biosynthesis that seems to be the cause of many mitochondrial diseases. The mitochondrial genome only codes for a few proteins, but relies heavily on the nuclear genome for proteins. In other words, the mitochondria isn't just an isolated power plant; it's an integrated system that provides power in addition to logic, material processing, signaling, etc. Sorry to hear about your disorder, fortunately there are people researching these diseases.
Well said, I thought HN were only for geeks, but you clearly have a solid biology background.
I agree that faulty mitochondria will break homeostasis and from this point on, the problem is unsolvable.
But homeostasis will break because:
1. Plain vanilla energy deficit. Not enough energy. Babies die within the first days after birth.
or
2. Inefficient - but normal - energy production. This will produce an excess of free radicals, "emitting" all kinds of weird signals to the rest of the cell/tissue/body. Most of mitochondrial diseases and diseases that cause a secondary mitochondrial dysfunction fall in this category. "The threshold effect" - mutant mitochondria outnumber healthy mitochondria as you grow older - will eventually take you to point 1.
Given this, repairing the mitochondria will bring homeostasis back. Currently, there are 3 lines of thought:
1. Repair the faulty mitochondria
2. Allotopic expression (literally, make backup copies of mitochondria DNA in the nuclear DNA and let them "express" from here instead of from mtDNA)
3. Kill faulty mitochondria. Given that mitochondria go through a fusion/fission cycles every few days, killing faulty mitochondria will restore a healthy mitochondria population in a very short period. The underlying genetic defect is still there, but it causes no harm at all.
Thanks for your concert. I didn´t want to be tragic in my original post. My life depends on my mitochondria, but I am doing relatively well considering my disease.
For the curious, I have CPEO + myopathy, common mtDNA deletion, 35% mutation load.
I think emerging research in genome engineering and sculpting evolutionary fitness using genome engineering tools will probably solve these disorders. However, these technologies have--in my opinion--lower hanging and more profitable fruit in other biomedical interventions, such as cancer therapy. They will probably cure mitochondrial disorders as a side-effect, though, and possibly by small start-ups (since there's such little money in these rare, one-shot-cure diseases).
Could you point at a high quality reputable paper saying "as low as 1:1,000"? Did you mean 1 in 1000 people with disease have a mitochondrial cause, or 1 in 1000 people have a mitochondrially caused disease?
The guy who postulated and validated the way that the mitochondrion operates was very much a scientific outsider and had to secure funding and run his lab outside the traditional scientific apparatus.
To be honest I doubt I would have believed him at the time given the context. It's a pretty complicated theory that requires a subtle understanding of biophysics.
Remember the reception to the Avery/McCloud experiment- people still insisted protein was the heritable molecule. People simply couldn't understand how a "simple" polymer like DNA could encode information. it was a real mind-bender because (AFAICT) nobody really appreciated how nonuniform polymers could code for information.
Or the reception to the discovery of enzymatic RNA. How could RNA possibly be enzymatic? Well, now we know that RNA runs the game and protein is just muscle.
The SENS Research Foundation has a network of supporters among portions of the VC crowd in the Bay Area. Some of the halo of attached or associated young researchers there should explore putting one of their lines of research more likely to get somewhere interesting soon into YC, see what happens.
A completely baked form of allotopic expression or toolkits for glucosepane breakdown in tissues could both produce viable and highly useful technology platforms prior to being far enough along to be applied to the reversal of aspects of aging. Neither of those seems too far away to be non-viable for this sort of timescale, and meaningful biotech projects are very cheap these days if you have access to a provisioned lab.
We (longecity members) founded this project out of our pockets a few months ago. I am sure Dr. Matthew 'Oki' O'Connor will love to discuss ideas with you.
One of the side effects of 120k vs 20k opening the model to different sectors/needs at the earliest stages. What will be interesting to see is the developments in expertise and what different requirements there are for managing and mentoring entreprenuers in these different areas. Would be great to hear some thoughts on that, if not immediately perhaps in a years time or so.
Biotech start-ups are very different to most other technology start-ups. I am not sure the ecosystem that YC has is really adaptable to "wet" biotech. I certainly hope I am wrong as biotech could definitely do with new thinking and approaches to funding.
DNA sequencing software is what my company develops (if you still have any ABI machines it might be of interest). If your centre is like my customers the computer based people still need the wet lab people to generate the data.
All Illumina here. Yes, of course we have wet lab people, but excluding admin staff, I think the wet lab staff make up less than half of the workforce.
SV VCs getting burned in cleantech was apparently not lesson enough?
"One of the key misplaced assumptions that Valley VCs made in cleantech boom times is that the rapid progress of Moore’s Law could be created for cleantech with a little bit of VC funding and Valley smarts."
I find the timing of this and the "New Deal" press releases a bit confusing -- is YC starting to take biotech companies now (like YC S14) or in the next batch?
I'm assuming the New Deal terms are for S14, but was also surprised to see the announcement after the applications were all in.
Can anyone clarify / point me to what I've missed?
Mm, I'd also be interested in finding out. The Request for Startups [http://ycombinator.com/rfs.html] doesn't mention anything, but then again they don't look like they've been updated in some time.
I'd be interested in seeing some guidance for nascent biotech companies wanting to apply, given the focus has been on software for so long.
Is $120k going to be enough to even make a dent in a biotech startup?! It's years and years and years of hard work before you may even get denied by the FDA. With all this funny money flying around who has the patience to wait 10 years to maybe see a return?!
Categorically, yes. But, you're only likely to succeed if you for one, question the fundamental assumptions on which things are built. Anything less is incremental, or so far down-track in clinical trialing that it wont get funded or wouldn't be sufficient, respectively). Questioning assumptions is fairly straight-forward because so much out there is high quality but seriously questionable bull. Not seeking to cast derisions, just that biology, by its very organic nature, is volatile and often uncooperative. The problem of course is that now we are talking about obtaining deep-domain knowledge, which can take some time, absolute minimum 5 years to read everything, condense and connect-the missing dots. My project, for example, has taken me literally a dozen years, in part, because I hadn't realized I'd started!
There is a need for biotech startups that help navigate FDA regulations. Think process & documentation for medical solutions, like shovels for gold-miners.
I've been doing consulting work at healthcare manufacturers. The environment is insane -- constraining, slow, unproductive, depressing. And it's used as a huge moat.
Down-voting my comment will not make the FDA regulations go away. You will not bring any biotech products to [edit: medical] market unless you comply with the FDA (or somehow manage to eradicate it).
Health-related products and services are heavily controlled -- in ways completely at odds with the common practices of Internet startups. If you haven't yet had to comply with these regulations, you may want to find out how onerous they are.
Here is one example: Quality System (QS) Regulation/Medical Device Good Manufacturing Practices [1], covering: Quality System, Design Controls, and Human Factors.
They audit your compliance with their regulations. You have to document everything you do, every decision you make, every change you make. That is not how startups commonly operate.
This is THE hardest problem to solve in a biotech start-up.
In the mitochondrial field, companies are going through orphan-drug designation and fast-track approval processes.
This is probably to only advantage researching for niche - rare - diseases: You can save lots of money/time to get the drug in the market. If you can then find some off-label use for you drug, much better.
Who say's biotech products have to go via the FDA? There are all kinds of potential markets with lower or non-existent barriers to market, eg my company which makes Glowing Plants. These are the areas which make more sense initially as you really can take a lean startup approach to them.
how's that going? You guys said you would deliver in april, but it looks like it's been pushed back to june. There are also very few lab-related blog posts, despite a promise by Mr. Drory in the reddit thread that all intermediate results, both good and bad would be disclosed. Also, he said a budget would be disclosed. I haven't seen one on the site yet.
>This will be totally transparent project - we will introduce a novel concept called "constant peer review" - In academia no one publish things that didn't work - we will publish everything.
>We take being a good Steward of our backers funds as #1 priority. Our budget will also be publish online and you can see where the money went.
I don't believe it's possible to tackle the gatekeeper or incumbents and win. You have to bypass them altogether before revisiting them. You have to, in Bill Gurley's words, have an orthogonal solution to the problem. My words, you have to be 100% disruptive. Or, you seek fast-track, accelerated approval, breakthrough therapy, and/or priority review, but then you will likely come across and have to compete against the incumbents, who've by the way woken-up on the orphan disease problem too. Truth is, we have bigger problems than orphan diseases in the world. Biotech startups should be tackling major diseases and focusing on transformative innovation, by questions the fundamentals.
Interesting move. I've often heard people complain that the best minds are writing HTML/Javascript when they should be solving real problems (e.g. medicine/health)... obviously because Tech pays big money. I'm not undermining the utility of FB & Whatsapp but I feel a larger percentage of the World population needs better healthcare & medicine. I'd be happy to see the balance of crazy paying startups shift towards life sciences. What could be a better incentive to innovate than this. Have a BS/MS/PhD? Good, go hack the body & decode it. 'x' years from now, there'll be open plugins to debug & fine tune your body.
The barrier to entry for writing an app has been lowered because of money+innovation in tech. Hoping for same in medicine. Human simulators would make it so easy to test medicines.
I don't think investing should be dumbed down to betting. The public already has high aversion towards "throwing $120K at [random] companies". YC (and with them a lot of accelerators/investors) analyze their prospects.
Sure, sometimes there is a chance and an underbelly-feeling involved in whether a startup should be invested in, but to degrade investments as "betting" is probably the highest insult an investment/accelerator-firm can get.
The parent's suggestion is that to refer to a YC investment as a "bet" is a little bit of a misnomer. It's a targeted, reasoned investment, not "120K on 'Little Devil' to win."
Science has become sclurotic. Teams that are funded outside of state science system are going to be able to take risks and try things that would be unthinkable in academia. Each of these teams in an experiment in a different way to do experiments. All of science will benefit from the results of these meta-experiments.
My experiences with industry research to date do not particularly back that up. For every "academia is too conservative", there's a pushback of "Yeah, but will this be profitable?"
False dichotomy, there are avenues of research that are outside of the traditional academy (and the 'new tradition' of state-funded science) and outside of industry. For example, Janelia Farms, run by HHMI. As I alluded to above, historically, Peter D. Mitchell's work (which garnered him a nobel prize), as well.
The story of how state-run science funding is broken is exemplified by the story of Douglas Prasher.
It's not a false dichotomy in the context of the discussion here. The 'bets' alluded to in the title are financial bets. And the first screen of any bet is "how big is the market" etc. In Academia science, the "market" is simply prestige...in "VC" backed systems, the market is simply "money". Basically everything in academic <as a system> is geared to polishing a reputation/presitge of the sponsoring institution. That prestige is the currency that allows the various non-profit entities to raise more money (ie, make money) so even this "prestige" is just another "invisible hand" of the market for donation or tax dollars. The marketing and propoganda around fundraising for universities, foundations, or tax-based government funding sources is "achieving our mission"...and the measurement of the execution along that front ("success"!) is the <term of art> called "prestige". As your story points out, "prestige", and the money that accrues to it, is only loosely correlated with "the truth" or "good science"...{etc}.
I wouldn't describe Janelia Farms as outside the traditional academy. It's just a foundation with great facilities and an extremely high bar to entry, but a low bar for continued funding.
I think that what is different now is the cost of trying out a new idea. A biotech team is able to try something out without buy-in from academia or the big bio-tech firms. They need some money, but much less than they would have needed 10 years ago.
And that is very cool. But frankly, is as true in academia - a new assistant prof. has the startup funding to try something big and splashy without bankrupting the lab.
"Biotech is getting cheaper" applies to both groups.
Hell, the same can be true of postdocs. Or grad students. I know I've gotten at least one paper thanks to cheap computational resources for unfunded musings.
As for @danieltillett's comment, "What you need to do for tenure" does indeed promote some conservative and safe thinking, but:
1. Post-tenure you have a great deal more freedom.
2. I'm not convinced it's more conservative research generating than "We need to keep the VCs/Shareholders/etc." happy.
3. A big, splashy, innovative paper that makes it into Science/Nature/Cell/etc. is a high risk but high reward play.
selected based on how many nature/science (Cell/PNAS/PLOSBio) papers you have. You could have 5 papers in lower journals where you've invented a miracle drug and never get a faculty position.
Of the members of my committee, and the chair of my former department, there are all of zero publications in Science/Nature/Cell or PLOS Biology. While often a shorthand for what you need, it's hardly a generalizable statement. These are all very good researchers, tenured, with positions at R1 universities. And all of them in fields appropriate to this particular discussion.
Great news, especially after the "biotech is dead" chants from Wall St. for the last couple of weeks.
I hope to see some posts like "XYZ Company (YC ##) hiring bioinformatics developer" sometime soon!
Actually the "might not be compatible with medical research" is the great part about it. Life sciences research often tends to be short-sighted, plagued by the egos of scientists, the politics of scientific publishing and the fight for government funding. I hope YC chooses the "incompatible" route.
Responding to criticism of the Immunity Project's lack of peer review:
"If the traditional science community wants to say these things aren’t going to work, they can keep doing that. That’s what used to happen with software companies. The Immunity Project may fail, and the next 30 things we fund may fail — and the whole model is such that that’s OK. We can have 31 failures and one success, and the model still works great. That’s how we’re designed."
> How many Nature papers are not reproducible? To call out a new bunch of people who are trying something different because people are afraid they’ll fail is sort of hypocritical given that we don’t really check whether anybody else’s results are correct, either.
Peer review is more about politics then reproducibility. Reproducibility is really important, but Nature doesn't seem to agree.
I have a theory with peer review, if it works, it is important and it is reproducible, then people will reproduce it and build on it. If any of the three parts are not there, then no one will want / need / be able to reproduce it, and it will fade into scientific obscurity, while the rest of the scientific community keep moving forwards.
Or people spend a lot of time and money trying to chase down something that a ground "found" and published.
It's very common to read a paper, think it sounds good, then try build on it only to find that you can't.
Peer review is completely flawed. Until it's made more transparent, like at F1000, it's going to continue that way.
Not being a negative Nancy here, but I've been on both sides of the fence (reviewed and had grants/papers reviewed) and it isn't an efficient system. Much like many other things in science, it's an antiquated process that needs to be overhauled to get back to it's "roots".
"Or people spend a lot of time and money trying to chase down something that a ground "found" and published."
Then again, you have the problem that negative results are rarely published. Many groups may repeat the same thing over and over, not realizing that many other groups have confirmed a negative result.
(Actually I think I saw a post here about a journal that aims to adress this problem).
That is 100% true, but that isn't a fair comparison.
Peer review should be open, scientists shouldn't be allowed to choose who reviews/doesn't review their papers, and the process by which selection is done needs to be more rigorous.
My suggestion, open it up to everyone in the field and make sure that the reviewers are identified. Anonymous peer review has the same effect that anonymous posting on internet forums has, people become e-thugs or just say things that they shouldn't ever say because they know the reader doesn't know who they are.
This sounds like it doesn't happen, but you would be surprised that the stuff that happens behind closed doors. Scientists are people, people are flawed (which is fine), but that shouldn't hold back scientific discoveries.
This comes up a lot, and there is truth to it. But my flippant response is usually "the ones that aren't that important".
In just the past two months we've seen examples of the system working well for both reproduction (CRISPR) and invalidation (acid bath-induced pluripotency) of spectacular results.
Those are victories, but I wouldn't credit them to peer review. I think of "peer review" as what happens between a manuscript being submitted and it being published.
The point is they don't need to defend the Immunity Project. They are giving them the resources to (ideally) allow them to defend themselves. If they fail, YC is fine with that, but the Immunity Project at least has the chance to try something.
I think a lot of the pushback, and absolutely my source for concern, was not so much YC funding the Immunity Project, but the hype spike that comes from a high profile investment around something with very, very little actual scientific evidence behind it.
The lack of any peer-reviewed papers backing up their approach is not a problem at all if professionals invest in that company. They should be able to get the data from the company and form their own opinion on it.
It's a bit different if you crowdsource the whole thing. The individual backers don't have the ability and knowledge to dig as deep to actually judge the validity of the approach. So I can certainly understand that this draws some criticism.
