two kinesins on a microtubule

Hello Marius. How is this different than the work that has been done on simulated metabolic networks using graph grammars?

http://www.tbi.univie.ac.at/software/GGL/

Wut.

Mail me pls? GGL is great, I don't want to steal your thunder, I'm always open to discuss, are you? Chemlambda is a graph rewrite system (like GGL) with a random algorithm for reductions. Differences: (1) not any rule, but a small specific set of rules (why: because making rules is cheap and the explosion of consequences is huge), I think that by accepting any rule there are very small chances to fall over really interesting ones, have you seen what chemlambda does? (2) the accent is on individual molecules which eventually do the job in a random environment, instead of CRNs which explode (3) no need for the nodes to be atoms and for the bonds to be real chemical bonds (4) can GGL do lambda calculus? perhaps, by chance, see (1), (5) different graphs (open ribbon graphs) which changes the notion of context, but that's not very serious because there surely is a way to formulate everything so that contexts are half-edges instead of being atoms. (6) And hey, is just awk/sh and javascript, so it works on any computer (7) Has it's own definition of life.

>>7710953

Do you have one of a regular kinesin?

bump for megainterest

Yes https://github.com/chorasimilarity/chemlambda-gui/blob/gh-pages/dynamic/kinesin.mol
but this is an artificial chemistry, I only speculate that it can be done with real chemistry (because the rewrites are so simple then there are real actors for the nodes, ports and bonds).

In this case of an "artificial" kinesin I just built it by looking at what happens with an artificial molecule which reduces as a permutation which is applied to itself
https://plus.google.com/+MariusBuliga/posts/8pD7gitcwwq

The point is that chermlambda is an artificial chemistry which:

- comes with an algorithm for graph rewrites application (basically a random rewrite algorithm).Comparison with GGL, they only produce all possible graphs which can be obtained from an initial one in a step, which gives an explosion of molecules, then they arrange it into a chemical reaction network. Opposed to that, I start with one molecule, carefully designed and I let it evolve randomly, being sure that (with some margins) it achieves the goal for which is has been designed, eventually.

- is Turing complete because it can be used for untyped lambda beta calculus, something GGL can't do AFAIK.

Now, this mix gives surprising results. For example look at the predecessor lambda term, as it appears in chemlambda
https://youtu.be/6LZvIkM6JqY
you notice that after some time it is a process of propagation of a certain configuration of nodes along a sort of a track. This is not visible in lambda calculus, but it gives the idea of a chemical quine, which is a molecule with the property that in a deterministic reduction algorithm has a periodic behaviour. Such a chemlambda quine, this time evolving in a random reduction algorithm, is a good proposal for artificial life. It has a metabolism, it can reproduce and it may even die.

Same thing here with the "kinesin". See for more the chemlambda collection https://plus.google.com/u/0/collection/UjgbX or better go to the github repo and use it.

>>7711807
>> works on any computer
It caused mine to crash

There is a problem of firefox with d3.js which may freeze this browser. I use safari for best results, chromium not bad. All animations have been done by screencasting the results from the scripts. Alternatively if you look in the quiner_....awk script then you see at approx line 30 a parameter "time_val=10;" change it to 100 and the firefox should be able to execute the html output. The downside is that it is going to be slow, like the demos from the page http://chorasimilarity.github.io/chemlambda-gui/dynamic/demos.html which all work with firefox.

If anybody can help with writing a more intelligent js part than what I did, please please do it.

My dumb solution is that I have a firstpart.txt which is the header of a html file and a lastpart.txt, the end. One writes, say, "bash quiner_experia.sh" and sees a list of file.mol. Then you have to type file.mol (chosen from that list) and the sh script calls quiner_experia.awk with file.mol as input. It produces a temporary file essy.txt. Then the sh script puts together firstpart.txt, essy.txt and lastpart.txt and names this file.html. The file.html can be seen with a browser.

The essy.txt is the middle part of a d3.js thing which basically is a list of addNode, addLink, rmNode, rmLink.

The problem (aside the one of firefox with d3.js) is that it may be a long list of calls.

Another problem is that the number of steps (parameter "cycounter" in the awk script) has to be taken smaller or larger, according to the molecule (i.e. the file.mol). For some of those, like the mentioned "quines", a big cycounter (like 1000) may create big file.html. As a rule of thumb, if the sh script is not executed in less than 1 min, better kill -9 the awk and you'll still have the result up to the moment you kill it.

On safari I have no problem with time_val=1.

Whad do you want, that's a patch of scripts which work for research, done by a geometer.

Anybody can do better, I will be publicly and eternally thankful!

>>7710953
today /sci/ was actually cool

>>7710953
Could you just give the rest of us an idea of what this is? I want to understand this, I really do, like are these proteins?

These are like chemical molecules. The rewrites (say a cascade of removeLink, removeNode, addNode,addLink triggered by small configurations of nodes and links) are like chemical reactions, where an invizible "enzyme" bonds to the pattern and rewrites it into a new pattern. The bonding of the enzyme to the pattern is random. The patterns are very small wrt the whole molecule and everything is happening locally and asynchronously.

OK, so just like in real chemistry, but for unknown "nodes" and "bonds". The patterns are basically only two, with different colored nodes.

Fort a chemist the problem would be to find real molecular compounds and real chemical reactions which can implement the "nodes", "bonds" and "rewrites".

Christoph Flamm (see the link to the GGL) takes the opposite approach: he has real chemistry molecules and reactions and expresses them as graphs and graph rewrites. This helps for simulating known real cascades of reactions.

I can do anything (computable) without the need to have any control over the order of reactions, simply by designing a good molecule.

Abstractly, it works close to functional programming languages (although I had discussions with people from the other side of that fence, who are not sensible to chemistry, and they tend to see ONLY that side), or close to what does a compiler to optimize a program.

But the funny thing is that it does the programming like part without any use of variables (like in chemistry) and it does chemistry like stuff like it's programming.

There are two applications possible: in chemistry, is a potential solution for building molecular computers (but not as they tried, via 0/1 logic gates); on the net, it is a model of asynchronous, purely local (there is no boss, no global space, no global time) computation which has no evidence left, because it does not have any variables and the graph-molecules are not flow diagrams.

File: Kinesin_walking.gif (109KB, 380x286px) Image search: [Google]

109KB, 380x286px

>>7712243

A Kinesin is a "motor molecule" that transports proteins and other molecules around within the cell.

The kinesin "walks" along microtubule molecules with chemical reactions that alternately attach and detach its "feet" from the pathway.

Picture is a cartoon representation of a kinesin walk.

OP's animation is more of an artificial life model, not a simulation of the actual process.

>>7712524

ON
MY
WAY
2
FUK
UR
BITCH

Yes, see this https://youtu.be/YAva4g3Pk6k for a real kinesin. I did one, which has the same functionality, in artificial chemistry. Turns out that it has two legs, the attachment of one triggers the movement of the other. And it has a site for carrying a load. You can take the mol file description https://github.com/chorasimilarity/chemlambda-gui/blob/gh-pages/dynamic/kinesin.mol and apply the rewrites described here http://chorasimilarity.github.io/chemlambda-gui/dynamic/moves.html and with patience you can start to see how it walks and why it may even detach from the "microtubule".

The design is nontrivial, so how to find it without using real chemistry? Magic :) No, it comes from experiments with lambda calculus, starting with the Y combinator which appears as a gun which shoots pair of nodes https://plus.google.com/+MariusBuliga/posts/YGUHjmdjSPJ
then seeing the same configuration which appears in permutations, as seen in this chemistry
https://plus.google.com/+MariusBuliga/posts/E6t6bqKgvVf
but the problem is that it has one leg which hops and does not have any place to hung a load on it, then finally the solution is found.

Now, it's very troubling that one can get a kinesin design by thinking at the Y combinator and permutations.

I like this way of talking. 4chan is the best for the best

>>7712524
https://www.youtube.com/watch?v=SjGjFNX6OlQ

(yes I know it doesn't really move quite like that)

STOP
IT'S A VIRUS
AND SPAM
RUN

File: Kinesin transport.gif (2MB, 256x256px) Image search: [Google]

2MB, 256x256px

>>7712731

Gif version

>(yes I know it doesn't really move quite like that)

That's why I'm intrigued by OP's animation. Kinesin motion is undoubtedly more chaotic than depicted here.

>>7713511
Where does it get the energy for locomotion?

>>7712125
>>doesn't work with firefox
could you fix this?

>>7712359
could you explain what you mean by locally? Are you carrying out graph transformations based on distance? Where's a publication explaining these animations?

Graphic lambda calculus seems interesting. I might have to implement this in the graph transformation system I use.

>>7713535
also you should take a look at REAX. It's a classical molecular dynamics approach for simulating chemical reactions.

>>7713515
Free floating ATP.

>>7713543
ATP can't free float. It's unstable and quickly degrades, which is why it can't be stored.

>>7713511
here's a more realistc version:
https://www.youtube.com/watch?v=Vlb9F_cj6As

>>7713545
but it uses free floating ATP anon.
https://en.wikipedia.org/wiki/Kinesin

This animation with 2 kinesins used the mol file https://github.com/chorasimilarity/chemlambda-gui/blob/gh-pages/dynamic/kinesinX2.mol

It's random, a different run gives different results.

>>7713535
I can explain, but why? You do have a net connection, use it.

>>7713515

There is no energy balance check, BUT do you see all these dots coming and going? It's an open system, not a close one.

In this artificial chemistry, it takes the energy say, from this ensemble (copy paste from the mol file)

A w13d w19 w17
A w14d w17 w18
FOE w18 w16d w19

This is a 3 nodes cousin of the Y combinator gun I mentioned and it shoots triples of two FOE and one A nodes (by reactions which increase the number of nodes, so here is the energy coming from). These triples enter in other reactions with the "feet" and make them "move".

That's how the kinesin may detach from the microtubule. If the Y gun cousin does not provide, because of randomness, then the feet eat it (react with it) and the result is that the kinesin detaches.

gone from google search "chemlambda 4chan" but they still have an old search result which leads nowhere. whatever :)

>>7713562
Christ, that seems slow. This must be a simulation at some very, very slow rate. What is the usual speed of Kinesin?

>>7713946
same person

Looks like it takes 40 steps a second, given a 16nm step size. gives it a speed of 640 nm/s

File: tape_long_5364_bi_1.gif (1MB, 480x360px) Image search: [Google]

1MB, 480x360px

same chemistry, different example.

Chemistry is fucking gay

File: Microtubule_structure.png (635KB, 1200x848px) Image search: [Google]

635KB, 1200x848px

This stuff is fascinating

Microtubules are created from alternating blocks of tubulin protein