<![CDATA[Decomposer - Blog]]>Fri, 01 Jan 2016 03:52:26 -0800Weebly<![CDATA[Princess Amygdala's Mitochondrion and the proton-motive Force´╗┐]]>Fri, 01 Jan 2016 07:02:08 GMThttp://www.decomposer.org/blog/princess-amygdalas-mitochondrion-and-the-proton-motive-forceApologies for the sad sad Phantom Menace punning, but I had a hankering to write up my book-reportish thoughts on the pop-biology I've been learning from Nick Lane's writings, and after taking the kids to see the new Star Wars movie, I couldn't help it [note to neuro-interested folk: this post will not talk about brains or amygdalae at all, it's just so close to Amidala that I couldn't resist].

Starting in September or so, I went off on a crazy devouring rampage of pop-biology (through the specific lens of Nick Lane's excellent books), zipping through "Power, Sex, Suicide (Mitochondria and the Meaning of Life)" first, then attempting to be as entranced with "Oxygen, The Molecule that Made the World" (but wasn't able to find it as interesting), before switching to his latest, "The Vital Question (Why is Life the Way it is?)", which was pretty amazing, although it recapitulated some of what was in the other two.

Prof. Lane is a working biochemist and evolutionary biologist who's been circling around a bunch of interesting ideas over the past couple of decades, and his books are written as "popular science", but at the same time trying to tie together the results of a large amount of current research into a cohesive story - one which is not necessarily "the consensus belief" of the scientific community as yet.  It's not "his crazy pet theory" (a la Penrose's quantum consciousness), but it's also not (quite) what you'd find in a biology textbook either (at least, not after a big of amateurish digging in "Molecular Biology of The Cell").  Thankfully, he's pretty good about noting exactly where in his books he's about to depart on speculation (either theory that has not yet been verified by experiment, or interpretations of experimental work which are not yet considered scientific consensus), and when he's stating things which are commonly known (among biologists).

At a high level, the main points which Lane addresses are super fundamental: the origin of life itself, and the origin of eukaryotic (including all multicellular) life.

The extraordinarily dated (and apparently no longer accepted) explanation I always heard for the former was along the lines of the "primodial soup": step 1) gasses in the early early atmosphere could get zapped by lightning and poof! organic molecules.  step 2)  ...  step 3) life!  Lane advocates an origin story in the "microporous labyrinths" of alkaline hydrothermal vents in the ocean floor - originally put forward by geochemist Mike Russell and microbiologist Bill Martin.  For those familar with the towering hot "black smoker" hydrothermal vents, covered in giant tube worms and surrounded by eyeless shrimp: we're not talking about these.  There have been some advocates of life originating in these vents, also, but Lane argues that the likely chemistry of the early ocean (in particular, lacking oxygen, and being heavy on the dissolved carbon dioxide) would preclude this idea (not to mention that the early prebiotic reactions would be reacting on the surface of the vents, and dispersion into the free sea seems pretty easy.

The warm (but not super hot) alkali hydrothermal vent hypothesis which he favors goes something like this: the sea floor above (but not directly) magma chambers near the tectonic spreading centers has cool sea water percolating deep under the sea floor meeting alkaline, hydrogen-rich fluids which are the products of serpintinisation: a reaction where olivine, a rock common in the mantle, reacts with sea water.  The geology of these vents has these combined factors: the carbon dioxide-rich down-seeping sea water (which Lane says is likely to have been mildly acidic in the Hadean era, 4 billion years ago, although he admits the pH of the oceans back then is not currently constrained by any modern observations) and warm and buoyant alkaline fluid venting up, both interfacing in thin porous iron-sulfide bearing holes in the rock.  

Apparently hydrogen and carbon dioxide, under these conditions (in particular in the absence of oxygen, which was certainly true in the prebiotic times) reacts spontaneously to form amino acids, fatty acids, carbohydrates, and nucleotides.  I need to do some more chemistry reading before I can buy that without a [citation needed] tacked on!  Especially because while these reactions are both energetically and entropically favorable, there is still a significant kinetic barrier, when thinking about forming organic compounds from carbon dioxide.  Something I need to read about more is that life apparently would like to have the kinetic barrier between hydrogen and carbon dioxide and formaldehyde lowered, but the barrier between formaldehyde and methane (via methanol) retained (for later use in releasing energy).  The idea goes that if you have tiny bubble-like pockets in an iron-rich (i.e. happy electron acceptor) rocks with the chemical environment described above, you can trap organic products which are produced in this far-from-equilibrium environment, and keep them in some place where self-reproducing molecules can propagate, and eventually learn to build some kind of lipid-based wall inside of the crust of the geological "membrane" they have containing them (side note: Lane and his collaborators have done experiments to try and verify that these reactions actually do take place, and have shown that depending on hydrogen concentration, pressure, and pH, you can actually produce formate, formaldahyde, as well as ribose and deoxyribose, all from hydrogen-infused de-oxygenated water and carbon dioxide).

This is where one of the other interesting clues which supports this idea come from: the tree of life.  Phylogenic analysis comparing achaean, bacterial, and eukaryotic genomes shows that while all three domains of life share some commonalities: they all have (basically) the same genetic code, translating DNA into RNA into proteins, they all have ribosomes to do the protein building, and they all drive ATP synthesis via chemiosmosis: a proton gradient across a membrane (and from what I can tell, ATP-ase [molecular animation of this crazy machine in action] itself is apparently conserved between archaea and bacteria, even though some details in its structure differ].  Some other citric acid cycle components are likewise the same, but not much else.  In particular: the cell wall is comprised of completely different things!  

In some ways, if you think about it, the fact that all bacteria share similar components of their outer cell walls and plasma membrane, and all archaea share similarly among that domain of life, that fact almost implies that "life" did arise twice, if you consider an enclosing membrane to be a key requirement of life [note: Lane isn't saying this, I am just rambling here, bear with me].  But the point being that if self-replicating energy consuming structures existed without a cell wall or membrane they built themselves around them, then you really definitely needed an environment where geology provided something to contain the reactions, or else they'd dissipate away.  And life is nothing if it's not a far-from-equilibrium scenario, so I've certainly warmed up to the hypothesis that you want an interface between different pH level fluids, at a temperature and pressure amenable to creating biomolecules.

There's a lot more to it then this, of course, but if you want to get a feel for it, just start digging into Nick Lane's books over the past decade or so (in particular the ones mentioned way up at the top of this screed).  But it's almost midnight, so sadly (or luckily, I suppose) I don't have time to either go further on the bacterial / archeal split, the magic of the proton-motive force, or the other Big Question I alluded to at the beginning: the origin of eukaryotic life (and hence all multicellular complexity).  That will have to wait for another blog post, I suppose.

<![CDATA[ML Talk for Non-CS Folk]]>Wed, 18 Jun 2014 04:36:46 GMThttp://www.decomposer.org/blog/ml-talk-for-non-cs-folkAbout 24 hours to go before I give my (relatively short) talk at Re.Work Berlin (https://www.re-work.co/tech-berlin ).  Pretty nervous, not the least of which is because I'm in such great company (the list of presenters is pretty amazing), but also because I'm trying to present work which is not *specifically* my own, but which is highly mathematically technical, to an audience of non-CS/non-math folk.  

So it's a little intimidating: how do you present a telescoping view of machine learning for personalization to people who won't even really know what machine learning *is*?  In 15 minutes or less?

I guess we'll see.  I'll post a link to the slides up here when I'm done.]]>
<![CDATA[Model building vs. decomposition]]>Fri, 13 Jun 2014 16:42:00 GMThttp://www.decomposer.org/blog/model-building-vs-decompositionI've always found that building complex things out of simple components is easier than trying to directly decompose complex things themselves.

I think that's the difference between people who are more like mathematicians, and those who are more like scientists.  The real world is messy and complex, and you don't know what the building blocks are going to be, so reductionist scientists have to decompose down, rather than build up.  Which isn't to say that e.g. theoretical physicists don't often start with simple building blocks and build models.

What about you?  Do you prefer to work in "clean-room" environments with models built out of very simple basic components, and see what they do together?  Or dig into big messy complex datasets and see what you can extract out of them?]]>