Hacker News Reader

I worked in the synthetic biology lab (on a different project) while the early stages of the Syn3.0 was being done, and also have paid several visits to chat with them in the meantime. [Proof: first author on http://dx.doi.org/10.3390/ijms16012020] I was a fly on the wall for most of the group meetings and even contributed some unpublished results (there were transposons that were causing problems by shuffling the the DNA in the yeast and I hypothesized the orientation that causes the issue and discovered the yeast genes responsible for this process)

Firstly what constitutes "minimal" depends on what you feed the organism. There's a set of bacteria called phytoplasma which are plant parasites that are missing genes to make nucleotides (even mycoplasmas have those) and they've adapted by sucking nucleotides from their hosts.

Some insider information: Most of the mystery essential genes are vague cell wall proteins. Probably what is going on is that if you knock out too many of these genes, you lose cell wall turgidity and the cell becomes nonviable. So what is important is not so much which of these genes you have, but how many of them you have.

Second insider information, for fun: The "hypothetical minimal genome" (syn2.0 is referred to as HMG in the paper) was actually a backronym because we called it the "hail mary genome" but decided that was inappropriate for publication.

> So what is important is not so much which of these genes you have, but how many of them you have.

That hints that you could just duplicate the same cell wall protein gene instead, right?

presumably. Or just put in fewer of them, behind stronger promoters.

The "hypothetical minimal genome"...was actually a backronym because we called it the "hail mary genome" but decided that was inappropriate for publication.

One in-house ORM I worked with was called TFP, because it was entirely written on a plane ride to Houston, and they named it TFP because the authors liked to say they wouldn't touch it again with a ten foot pole. Their company later thought about trying to sell it, with the name "Technique for Persistence."

>> Firstly what constitutes "minimal" depends on what you feed the organism.

So maybe the environment needs to be changed to support an organism with a smaller genome. The world today is certainly different than when life began. It's probably difficult to evolve the environment, so they'd need to understand what goes wrong with gene deletions and figure out how a different environment could help.

When dnautics says "different environment" he's not talking about the global environment, he's talking about the environment around the cell. A minimal viable cell should be able to survive on water, some dissolved gasses, and an energy source like glucose. Bacteria that have even smaller genomes are doing that by stealing components from other lifeforms.

I'm not exactly sure what went wrong, but I think you're really misunderstanding something dnautics said.

> A minimal viable cell should be able to survive on water, some dissolved gasses, and an energy source like glucose.

M. mycoides certainly does not survive on that. Mycoplasmas need cholesterol and lipids because they lack the HMGCoA reductase pathway and FAS, so all the media you grow them on are enriched for that (started as FBS, but we found horse serum worked great and was cheaper). In general, I think you need a few more things than just that for most basic life (nitrogen, phosphorus, metals)... Is using glucose cheating? Because you could use light for energy and CO2 for carbon, but that adds a ton ton ton of genes...

So this becomes a nitpicky definitional conundrum.

Hell, we found an intranuclear parasite that didn't even have its own ATP synthase, glycolytic pathway or any other obvious way to generate ATP.

>> Bacteria that have even smaller genomes are doing that by stealing components from other lifeforms.

There are no bacteria with smaller genomes today, that was the point of making this one. I was not talking about the global environment either, but the number of options for what substances and concentrations may need to be present in the environment around a cell is certainly very large - making trial and error difficult.

> Most of the mystery essential genes are vague cell wall proteins. Probably what is going on is that if you knock out too many of these genes, you lose cell wall turgidity and the cell becomes nonviable. So what is important is not so much which of these genes you have, but how many of them you have.

Sounds like dependency hell. Gene A only works in the presence of Gene B which only works when there's Gene C which needs Gene A. Knock any one of them out and everything breaks.

> Sounds like dependency hell.

Random mutations are like that. It's almost like no-one was planning out how these things should work, and whatever did work stuck.

Weird, huh?

In other words, God writes spaghetti code.

Praise his noodly appendage!

God writes code that writes code that writes code ... that writes spaghetii code

In this simplified system, the number of these cell wall genes in the genome is proportional to the quantity of the cell wall proteins produced. It's more of a quantity issue, rather than a dependency issue. For example, having more developers usually results in more lines of code written in a given period of time, especially in enterprise. The quality of code might be questionable but the quantity is usually there.

> They found that not a single gene is shared across all of life.

It is definitely a function vs identity. Many mutations of a single gene can perform the same function -- there is a lot of redundancy in genetic code.

>If he had done the same set of experiments with a different microbe, he points out, he would have ended up with a different set of genes.

This is very interesting. I expect also, that if they proceeded with their knockout in a different order or different sets at one time they would end up with a different set of genes. Maybe they have some massively parallel way to knockout genes, but I doubt they have explored the entire set of knockouts from the original functioning natural organism. They very likely could be at a local minima of genes required (eg no single gene can be knocked out, but no where near the absolute minimum).

Years ago I read an interesting paper comparing the types of subsystems found in the Linux kernel with subsystems found in biology (or what we understand of them currently).

There is a lot of redundancy in biology, which allows for biological systems to be pretty robust, adaptable and fault tolerant. Of course this also means it's really hard to fix that biological process when things go horribly wrong, such as in many forms of cancer.

Paper: “Comparing genomes to computer operating systems in terms of the topology and evolution of their regulatory control networks.” By Koon-Kiu Yan, Gang Fang, Nitin Bhardwaj, Roger Alexander, Mark Gerstein. Proceedings of the National Academy of Sciences, Vol. 107 No. 18, May 4, 2010.

Thanks for that last paper, epic!

Perhaps a good moment to refer to this one:

Cancer tumors as Metazoa 1.0: tapping genes of ancient ancestors http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3148211/

This means that we need first to decide on common environment that minimal synthetics will live on. Then, with this environment fixed, we could try to isolate minimal set of genes for multiple different organisms that are originally viable in this environment.

Such environment should also be 'minimal' - i.e. having lowest number of resources required to construct it and lowest amounts of those resouces.

> They found that not a single gene is shared across all of life

I think this is a pop-science misunderstanding of what the authors were trying to say. They couldn't find a single unique "minimal set" of genes- but there are genes that we know all organisms share, like the rRNA genes and homologous core ribosomal proteins. All organisms will also require some kind tRNAs and their genes from some source, even if the tRNA genes themselves are not highly homologously conserved.

Exactly. The rRNA peptidyl transferase loop is conserved throughout all life, for example. Like, you say the researchers obviously understand this, but author of the article likely did not.

>In fact, there’s no single set of genes that all living things need in order to exist. ... They found that not a single gene is shared across all of life.

Isn't the current theory that all life on Earth has a common ancestor? Wouldn't this point put that into question and introduce the possibility that life has developed multiple times independently of each other or would this lack of shared genes eventually appear through divergent evolution? It seems like the former option would greatly increase the fl component in the Drake Equation meaning alien life is even more likely than previously thought.

I would guess that the earliest life didn't accidentally hit upon optimal or even reasonably good ways to do whatever is needed to reproduce.

If that's the case, it isn't that unlikely that every evolutionary branch managed to improve (or discard) every part of the original machinery in some way.

Only if all life converges on having dna. If there are multiple origins of life that survive there would probably be multiple different self replicating molecules.

"Isn't the current theory that all life on Earth has a common ancestor? Wouldn't this point put that into question"

Not really. For one thing there's too much commonality across life in terms of things like the genetic code, i.e. what a sequence of genetic letters "means". The genetic code is arbitrary but essentially universal with minor variants.

For another thing, the organisms now present can be separated from each other by as much as 8 billion years of evolution (4 billion on each branch) meaning there's no particular reason to expect some shared gene to be present everywhere. New genes arise as mutations from old ones, old genes become obsolete and get removed, etc.

It's vaguely like how, if you fork a coding project, and the two projects are allowed to continue indefinitely, eventually there might be no code in common, even without a complete rewrite ever happening.

Deep down, somewhere in the genetic code, it still says "Mozilla (compatible)"

Not when your browser fork is re-purposed for something else almost entirely, like an inline image or documentation viewer. As I understand it, organisms often have novel uses for old tools.

That explains why a human being is like a gecko.

Depends on what kind of life you consider. It's true for mammals. It's true for animals. It's probably even true further up our ancestry. But apparently it doesn't mean that "all things that have genes" share a subset of genes from the same common ancestor. As I understand it that's because they share a common ancestor that didn't have genes (but at that point I'm not sure whether "ancestor" is even the right word).

There is a great book written 20 years ago by Steve Grand called "Creation: Life and how to make it"[1] it basically uses the game Creatures[2] as the testing environment for creating Artificial Life. There are some great thoughts in that book also about how science needs more cross disciplinary experts to connect the dot's between insights in one field and apply it to others.

[1] https://www.amazon.com/Creation-Life-Make-Steve-Grand/dp/067...

[2] http://creatures.wikia.com/wiki/Creatures

> I deeply believed that organisms share the same basic set of genes

They do. What you're responding to says there's no one gene shared by all organisms. The genetic intersection of any two organisms is large; the genetic intersection over all organisms is empty.

> I deeply believed that organisms share the same basic set of genes.

Many groups of organisms do. All organisms is a tall order, if you think about it.

The more I think about that the more makes me believe that the Genome is not the smallest order which defines "life".

Maybe all is hidden behind getting blurry because of Heisebergs uncertainty principle.

If you consider the genome to be information, life is effectively the process that information supports by encoding a method to accomplish the process. Replicating and using energy in a coordinated manner are two processes of life, but their information-based encoding can differ significantly based on the method used.

It's the process of making things work in a cell. Turns out, there are more than one encoding for how to make that happen.

> That's the most interesting point to me, I deeply believed that organisms share the same basic set of genes.

Yes, isn't there a universal gene that encodes the RNA polymerase enzyme? I.e., the machinery that transcribes DNA into RNA?

probably the ribosomes and tRNAs are the most universal-ish genes.

RNA Pol is very diverse across clades.

My understanding is that all known organisms share at least the ribosome RNA genes. Unless you're counting viruses, which is not exactly fair since they hijack the host's ribosomes.

but there is RPL1 up to RPL41, and same for RPS and then MRPL and MRPS , so despite the similarities, one can argue different organisms may have different rRNA genes.

RPL1 and such are genes for ribosomal proteins. I'm talking about the genes that encode the ribosomal RNA subunits. These genes are used to infer the evolutionary relationships of the tree of life, so presumably they are shared across all known organisms, or else it would not be possible to infer a phylogenetic tree.

Is there a single set of instructions needed for Turing Completeness? It's possible to have Turing Completeness in a random access machine with just one instruction. I should think you really need to satisfy certain capabilities, and there would be many ways to satisfy those.

https://news.ycombinator.com/item?id=9751312

>As @sixQuarks has already written, finding the minimal amount of genes when there are 175 unknown...seems to be pretty much impossible.

The point isn't the minimal genome, it's to identify the genes important for a minimal genome.

Genetics is all about filtering the signal from the noise. Now we know that these 175 genes are important, we can start focusing efforts on them.

The problem isn't that we don't know how to figure out what a gene does, it's that there are so many goddamn genes we don't know where to start.

> That's the most interesting point to me, I deeply believed that organisms share the same basic set of genes.

It should seem obvious in retrospect. Turing machines have many incarnations as different instruction sets in modern CPUs. I don't see why cells wouldn't have analogous multiple expressions, unless your deep belief was that there was only one way for protein chains to replicate.

> As @sixQuarks has already written, finding the minimal amount of genes when there are 175 unknown ones and you don't know anything about their dependencies and relationships seems to be pretty much impossible.

Is there any way at all at simulating the tests? I'd like to apply metaheuristics to search for which combinatorics work

This would require a quantum chemistry simulation many orders of magnitude larger and longer than what we currently capable of simulating. Though I suppose you could in principle use heuristics to greatly reduce the conputational complexity.

Not all life has DNA.

Do you know an article about that? Did not find anything about a life form without DNA, except http://news.nationalgeographic.com/news/2010/12/101202-nasa-....

The definition of life itself is not really obvious to me either, some sources count viruses as life other don't.

I'm guessing he means RNA viruses https://en.m.wikipedia.org/wiki/RNA_virus

The definition of life is a tricky question. The most obvious example is red blood cells dump their DNA. So they can't replicate, and virus can't target them. However, insect drones also can't reproduce and are considered alive so it's a little more open to debate than you might think.

As to life without DNA. The methodology is somewhat in question for this but here is a more in depth link: http://science.sciencemag.org/content/332/6034/1163.full Now, this is really 1:1 with DNA just using slightly different chemistry, but it's hard to call something DNA when you swap out one of the building blocks.

Other examples are hard to locate in large part because you can't direct detect this stuff.

Anyway, my point was while a virus is 'stuck' using DNA as it needs to infect things that use DNA. However, they can use RNA internally.

PS: For a really out there example, prions seem really close to life.

The paper you cite about the bacterium that uses arsenic instead of phosphorus in its DNA was conclusively discredited. http://www.nature.com/news/arsenic-life-bacterium-prefers-ph...

This does not mean that arsenate does not get into the bacteria, he points out. “It just shows that this bacterium has evolved to extract phosphate under almost all circumstances.”

Yes, it really wants phosphorus. But, a strong preference does not create phosphorus when there is none to be found.

Not that single cells really want or try etc, but you get the idea.

One definition that I like to include in defining life, is something that uses energy to decrease (or maintain?) local entropy.

This comment is not as absurd as everyone seems to think. It is possible that life started out as RNA[1] on Earth or elsewhere in the galaxy[2].

[1] https://en.wikipedia.org/wiki/RNA_world [2] https://en.wikipedia.org/wiki/Panspermia

I'm not aware of any life which lacks DNA. This is kind of "definitional" - really depends on your definition of life, but I don't include viruses, because they don't have a regulated metabolism.

>That's the most interesting point to me, I deeply believed that organisms share the same basic set of genes.

this seems to support the idea that instead of evolving from a single ancestral life form, all beings were created with care

edit: or at least would seem to support the weaker assertion that singl life form ancestry evolutionary theory is inaccurate.

if this is logically incorrect please feel free to tell my why. downvotes are not for disagreement!

It's obviously incorrect. It's not even really a logic error, it's a category error.

At face value the statement "if not all living things share a common ancestor, that makes them appear more like they were intentionally created" (although that line of reasoning, i.e. the watchmaker argument, has been debunked time and time again) seems reasonable.

However the statement seems far more grandiose than it actually is.

If two very "primitive" organisms (i.e. operating at a much lower complexity with much smaller and fewer moving parts than even, say, an earthworm) don't share their sets of DNA that may be unexpected (because whatever step there was from "not having DNA" to "having DNA" had to happen twice, separately, successfully).

If we found out that, say, all higher lifeforms shared no common DNA with other lifeforms outside their group and those groups closely aligned to the Christian "kinds" (which don't really map to biologically distinct groups in meaningful or consistent ways) THAT would be astounding.

But even that extreme case wouldn't make the "creation hypothesis" (if you even want to call it that) more plausible because it presumes the existence of an unexplained organism that is far more complex than anything it supposedly created. It doesn't explain anything -- it just shuts up questions about the "origin" by giving an answer that is impossible to analyse further.

There may have been a creator. But it's impossible to make a consistent logical argument for it. And it's entirely impossible to make an argument that that creator would in any way resemble something modern mainstream religions are worshipping.