Biology needs a Grothendieck (or at least a Hilbert) | Hacker News Reader

Biology needs a Grothendieck (or at least a Hilbert)

thebench.mit.edu

38 pointsthanatosmin36 comments

I came to biology from physics over a decade ago thinking that theory would matter. But modern biology is not waiting for a Grothendieck. We are not wanting for a deep theoretical view to reinterpret our data with. We are wanting for cheap, high quality data about cells and organisms in the first place.

Those of you who have not done biology can't realize how primitive things still are. A CS analogy is that we're trying to reverse engineer a non-deterministic alien architecture from the future via a remote debugger over a noisy line that has no error correction and runs at a fraction of a millibaud, where every peek or poke costs hundreds to thousands of dollars. With the advent of high-throughput sequencing we just figured out how to get noisy ROM/RAM dumps (genotype), but still have precious little ability to quickly interact with the biological systems under study (phenotype). Biology is much more like hacking and reverse engineering than it is like physics or mathematics.

A working exploit is worth much more than general pontification. The transformative events in biology these days are the clever hacks that enable whole new classes of fast experiments to be done: quick homologous recombination induced by CRISPR, optical interrogation of neural systems using light-sensitive ion channels, the development of adeno-associated viruses for rapid introduction of genetic material, etc. In the coming decades the combination of such clever hacks with cheaper automation and sensors will hopefully bring a flood of new data that will make a predictive, quantitative science of biology possible.

With cheap and plentiful data, perhaps clever minds might discover general quantitative principles that we can't see yet. But the data is what's lacking, not cleverness.

While I absolutely agree with the need for cheaper way to obtain data, I don't think this is the major problem we're going to face.

We now have ways of obtaining mountains of data (RNA-chIP comes to mind, but high-speed sequencing is a big one as well), but we don't know how to process it. We end up with thousands of candidate proteins, but then someone has to go through them one by one. Things are incredibly primitive. The huge pain point coming up is our ability to process information, which is very limited. How do you deal with machine generating TBs of data/day ?

I think what's lacking is a culture of automation. Physicists do it, chemists do it, but biologists still do everything "by hand", and not only is that imprecise, is also slow and costly. We also need better ways of obtaining information about other's work. When there are 1000 paper a year published on your subject, how do you filter ? How do you even find the time to read papers in other subjects, which could be interesting in their methods, or just for general intuition and knowledge ?

What we need is good algorithms, able to replace curate our sources of information well, and machines able to do the bulk-processing now needed in biology. Some things are coming up (I'm co-founding a start-up to do exactly this), but it should have happened 10 years ago.

We have ways of obtaining mountains of sequencing data that represents only a single dimension of an unknown system. While, it is a step in the right direction, this data by itself is still lacking. It's like trying to model the trajectory of a bouncing ball with extremely high resolution X coordinates, but horrible Y and Z coordinate data.

I've been working with sequencing data for the past 5 years. The problem really isn't with how to interpret the data, it is that the data itself sucks because it's extremely myopic.

And yes, there are people who are trying to analyze the various *-seq methods holistically. But I haven't really seen anything substantial coming out of it.

I agree with you that its essential to lower the amount of time and cost that experiments take, and find needles to replace our sledgehammers. I work in technology/methods development, so I'm happy that I should a job for the next decade! However, I have to disagree with the claim at the end of your response: once we get the methods right, we can start building biology into a predictive and quantitative science.

I think the debate needs to start at, what does it mean to make biology into a predictive and quantitative science? Sure, once that flood of data comes in (and I'd argue its already here), we can build huge dictionaries of perturbations-responses in every system we study. And sure, this will be extremely useful for drug development, synthetic biology, et cetera. But is it "understanding"? Is it intellectually satisfying?

Right now it's unclear to me how to even evaluate where are our current understanding of biological systems are incomplete <b> in ways that matter </b>. Sure, in any system there are remaining things to be discovered, but maybe they are 2nd order, 3rd order, 10th order effects on the behavior of the system. The way people are incentivized though, they will do all they can to convince you that the 3rd order effect they study is really just as important as the 1st order, and it often takes more effort than is worth it to see through the smoke screens.

Developing more methods and technology further fractionates the field more than it serves to unify it (with the exception numbering what I can count on my hands). Until we have any idea where we're going, it's futile to just wait for more data to come in.

This is the best analogy of modern biology that I've ever heard.

Just to add one bit of color commentary...

> but still have precious little ability to quickly interact with the biological systems under study (phenotype)

In case it's not obvious, part of the inability to interact 'quickly' is that organisms take their damned time to grow. Even E. coli will take about 20 minutes to double (under the best circumstances). Fruit flies take a few weeks. Mice? even longer. I suppose the field is ripe for some breakthrough to speed that up, but I don't know what it is.

In what sense is Charles Darwin not that person for biology? (I looked up the author of the short essay post submitted here, and it appears that his career has been devoted to nonbiological technology. I wonder what analogies from mathematics and electronics he is using to think about biology, perhaps without having looked at what the theoreticians of biology have had to explain in biological data?)

Thanks for the comment! Haha, I am definitely aware of Charles Darwin. I agree with you that much theoretical work has been done, and early on biology was largely a theoretical subject (we had an accurate quantitative theory of genetics before we knew what genes were!). (see my comment in the root thread)

I was hoping in this blog post to point to a more disturbing trend in our modern culture, where the need to solve problems in health have moved attention away from generalization and more towards individual instances of human diseases.

Darwin is that person since he discovered the only universal principle in biology. There is no more universal biological concepts to discover since evolution by natural selection makes universal properties impossible.

It doesn't make universal properties 'impossible' and it certainly doesn't make shared properties impossible. Here's an example of a theory which would extend the basics set forth by Darwin: https://www.quantamagazine.org/20140122-a-new-physics-theory...

And anyway, Darwin's theory has already been expanded on with some discoveries relating to inter-species genetic transfer such as through viral infection and also some interesting insights about mitochondrial ancestry. There is still a lot of mysteries about evolution and there are certain to be unexpected surprises.

Of course not, the fact that all known living things have DNA or RNA (and that this translates to proteins) is another universal biological concept for example.

Another is that the live is carbon based

Concepts like ecosystem or the nature of predation are also universal.

These things are not universals. Let's take the fact that all living things have DNA or RNA. We know that the first living thing did not have DNA and RNA. The only reason that it is believed that all living things have DNA/RNA is that this is how we look for new life - we say that something is alive if it has RNA/DNA. For example, if some of the early life that did not use RNA/DNA was still around we would not know since we would not recognise it as life.

A chemical reaction is not a life form, by definition. You can found sodium or gold atoms in a lot of life forms playing essential roles to keep this things alive for some more time; but if you dissolve salt in water you are not killing some salt people.

There is not a single exception known until today by the human race so, do not lie to yourself, those concepts are absolutely universal (right now).

Moreover if any slime or rock was really a early life form we should be able to recognised as that, since many years ago, for sure. A gold necklace that will create more gold or gain mass with time would not have gone unnoticed, neither a limestone rock growing and taking the form of the drawer that contains it.

Finally, an early form of life that did not use RNA/DNA in earth will be contaminated with RNA/DNA of its parasites or symbionts after a few thousands of years. The ecology predict this.

That's an interesting conjecture. Maybe the point isn't to find properties that are completely universal to all life, but more, do we even have a good definition of what life is? Could we always recognize life if we saw it? I think these are the sort of theoretical questions that are getting left out of the conversation.

Don't forget http://en.wikipedia.org/wiki/Alfred_Russel_Wallace

There were quite a few people who discovered evolution by natural selection before Darwin published the origin of the species, but the reason Darwin is given credit is because his explanation was so complete. It is in someways similar to Einstein and special relativity.

I'm wondering why the "at least a Hilbert" is thrown in there. His name doesn't appear anywhere else in the article, and the wording suggests that Hilbert is some sort of poor-man's Grothendieck, which is an association I'm failing to make.

Hilbert wanted to establish mathematics on a firm foundation because a number of paradoxes suggested that the current foundations was faulty (Russell's paradox being one, there are others that I can't remember as well).

Basically Hilbert wanted to find a list of axioms from which all of mathematics could be derived and then prove that these axioms were consistent, i.e., that you can't create two contradicting statements from them.

But then Gödel showed with his incompleteness theorems that this wasn't possible.

So I guess the author was suggesting that biology have a firm, axiomatic basis?

The reference to Grothendieck and Hilbert don't seem to be about establishing something like an 'axiomatic basis', but about having a leader who can organize a platform of modernization in the field. They both lead programs. That's about as far as the thinking goes on that one, I suspect.

good point - when I was initially writing this post (I'm the original author) I was planning to include a section discussing "the Hilbert questions" in mathematics. These questions not only guided the field, but just the conversation which preceded it was likely what catalyzed lots of the developments in 20th century math.

I think biology is ripe for such a discussion. We just need to convince people that it's worth having!

Biology is based on complex interactions among proteins and genetic material. There is no way to construct generalities about such systems that bypass careful observation of their inner workings. Just look at how long it took to discover DNA after the discovery of heritable traits. You can't hand wave your way around knowing how these molecules are assembled and transcribed.

Mathematics is an invented tool that can be arbitrarily defined to suit a problem space. There will never be a Grothendieck of biology because the mechanisms of life aren't so neat and ordered that they can reduced to first principles.

> There will never be a Grothendieck of biology because the mechanisms of life aren't so neat and ordered that they can reduced to first principles.

Let us simulate the universe and let that simulation be the ruleset for the first principles.

Not saying it's going to be neat and ordered, but on the other hand, the idea of it surely is. Whether or not it is ever feasible (or even theoretically possible) to do so doesn't take away from the merits of the idea, I think.

Well, Darwin would certainly have to be considered as one such example, so I wouldn't say that it can never happen. But I agree, at the cellular level things get much crazier. Never is a long time though.

Such a person in biology is impossible. Biology does not have any underlying common principles other than evolution by natural selection.

Is it not theoretically interesting to understand what the consequences of this "axiom" are? For example, can our current living systems evolve to catalyze any chemical reaction? Are there things in "task space" that biology can't evolve a solution to?

is it not theoretically interesting to understand what the consequences of this "axiom" are? For example, can our current living systems evolve to catalyze any chemical reaction? Are there things in "task space" that biology can't evolve a solution to?

Actually it does, and that principle is called chemistry.

And feedback loops. Actually feedback loops are arguably more important for living organisms than DNA because life could conceivably store its 'blueprints' in some other fashion but in order to survive in changing environments through adaptations, it's crucial that there be feedback loops.

What gets me about this is that a single misplaced bomb during WW2 would have stopped these discoveries, and he would have just been another casualty statistic. It makes you think about what advances other misplaced (or well placed) bombs have prevented.

Original author here - I just discovered this thread. Thanks so much for all the comments!

Some context behind why I wrote this post: as a researcher and technologist in biology, I've realized that it's unclear to me what the real questions are; that this lack of clarity is not due to my immaturity; and the field seems dangerously aimless.

>99% of the time, work is justified (and finished) with some application to health and medicine, with little regard for theory building or true understanding of a system. Bringing up this point usually solicits two responses: 1) we don't have enough data yet 2) there's little use for theory in biology. I hope I can give more in depth responses where I see these themes in other comments, but I'll generalize my thoughts here.

First, how do we judge what questions are worth asking and researching, vs trivial details? How do we even know whether we understand whether a question has been adequately answered? The current state of research to me often seems like John Searle's Chinese Room: we just generate a huge dictionary of perturbation-responses and call this "understanding". This might be effective to find drug targets, but is it real understanding? It might be the first step, but it seems to me that we're swimming in plenty of data already and it's worth it to step back and try to figure out what what it is we are actually trying to do. I'm comfortable with the fact that answers to our questions might be a long ways off, but it's scary that no one seems to care that we don't even know what the questions are.

Second, I'd argue biology has historically been very theoretical. We developed a whole quantitative theory of genetics (Fisher, Wright, Haldane, ...) before we had identified what a gene actually was. Hell, it was still debated what matter even looked like at that the microscale (where we knew genes must exist)! In the advent of molecular biology, Watson, Crick, Brenner, Gamow, Delbruck, et al, predicted through (some) experiments but mostly reasoning and conjecture much of what was later found to be true about transcription/translation. There are countless other examples of where theory has driven the study of the origin of life, molecular biology, ecology, et cetera. Only recently, when health science started to dominate biological research, did biology become less theoretical and more focused on individual instances of problems.

tl;dr - What's most scary is that we don't even know what the foundational questions are anymore for modern biology, and no one seems to even care. Rather than focusing on clearly stated foundations, the field is guided by the latest trends in glamour journals, which tend to obfuscate questions rather than answer them.

I nominate Eric Lander.

Gee, Lander was first a mathematician.

His lectures on biology at MIT are on YouTube and are a lot of fun. He's really charging ahead.

Missing the trees for the forest. We already have exactly what the article is calling for in biology, and it is called chemistry.

I sympathize with the sentiment, but biology is not a simple reduction to chemistry (although chemical principles do undergird it, and it is useful to describe biology in terms of chemistry as often as possible)

Nonetheless it is appalling how little chemistry biologists know. We were having a dna shearing issue in our lab, and I once asked, "what is the chemical mechanism of shearing" and no one could provide a convincing mechanism, which meant their solutions to the problem were not much better than shamanism.

It's even worse with the faddish worship of interdisciplinary-ism... Basically chemical biology (versus the more classical field of biochemistry, which is deep) is littered with dilettantes who were neither good enough to do straight chemistry nor deeply knowledgeable about biology.

If there will be a grothendieck or a Hilbert in biology, that person will almost certainly have to have some amount real chemistry training, and also deep chemistry knowledge. Probably training in information theory is important too.

In my view, physics is the study of the universe focused through the lens on energetic state functions, with a reductionist bent, chemistry is the study of the universe focused mostly on the diversity of matter with a passing nod to path dependence, and biology is the study of self-organizing information propagation with a strong acknowldgement of path dependence.

If you see a lot of people that don't deep into a particular discipline in a competitive environment, 90% of the cases is because they really don't need it.

Not all problems of Biology can be solved with chemistry, in fact much of the really puzzling and big problems of the current biology can not be solved using just one science.

For Biochemists chemistry is the only and truly god of course. Geneticists need to know only a little salt of chemistry and a lot of computer skills in fact. Ecologists are much more impressed by math and physics. Taxonomists need basically a good eye for picking subtle differences in 3D structures and an encyclopedic knowledge of the life forms.

Early biologists have few concepts to study, so one single people could grasp the majority. This is not possible today because most human research has done in the last 100 years and one single human brain could not cover all the new ideas and concepts emerging at superfast pace.

Totally agreed - I might steal your last paragraph there!

Chemistry is "the details" of biology, but it tells us next to nothing about what we actually care about, probably what you called "self-organizing information propagation" .

I think good definition for life shouldn't include any dependence on the chemistry used to manifest it. Agreed?

yeah, it is just a broad stroke characterization of the feels of each field. Some types of particle physicists would be probably be reclassified as chemists under my scheme, which they would bristle at, although in all honesty they are inheritors of a long line of tradition that goes through the indisputably chemist lavoisier, berzelius, and curie - more than they are newton or maxwell, planck, or einstein.

Reply on news.ycombinator.com