gilleain · HackerTrans

قبل 4 أيامdiscuss

Well, confusingly, the last picture on the blogpost look (to my untrained eye) to be tetrapods. Those are 'tetrahedral' (ish) shapes.

In contrast, the https://en.wikipedia.org/wiki/Dolos (plural 'Dolosse') are two tapering cylinders at right angles to each other. They have the same symmetry, but a different structure. Does that make any difference to how effective they are? No idea!

edit: When I said the "same symmetry" of course I should have said the dolos has some of the symmetries of the tetrapod (subgroup?).

gilleain

قبل 6 أيامdiscuss

Interesting point - usually people approach the question "who draws the tiger" from the other end. As in, the assumption is (not unreasonably!) the genAI is doing all the work and the AI 'artist' is doing nothing.

Of course, the more control you have - as you say - over the process, the more you are the artist.

I have to say, I'm not that interested in whether the output 'really' is art or not. I'm sympathetic to non-AI artists and I personally prefer art made by humans. However ... it is not impossible to make something that looks good or communicates an interesting idea with AI, but not trivially.

gilleain

قبل 6 أيامdiscuss

Of course, now the question the boy in the article asked - "where does the Tiger come from?" - could be answered with "the prompt".

The arguments today are whether a prompt like "draw a Tiger" are the same somehow to someone using a pencil or digital stylus or whatever to draw a tiger.

"But I do know that there’s no program that will draw one from scratch, any size or pose or color or style, like I want it, without an artist at the helm. And there’s no magic answer for us artists except that we are each unique. We are the only variable"

Not anymore! Now the only variable (for good or bad) is the prompt? Just add "super photorealistic, no bad anatomy" I suppose :)

gilleain

قبل 15 يومًاdiscuss

I see that I was unclear - I did not mean "the law goes along with this, and I don't know why?" I meant "I do not understand why Sov. Cit. would believe that the law would agree to these ideas".

gilleain

قبل 15 يومًاdiscuss

A very clear example of this is the 'Sovereign Citizen', who have bizarre beliefs around how to interact with courts. As far as I understand it, there are some 'cheat codes' people believe (incorrectly) are effective in literally getting you out of jail free.

https://en.wikipedia.org/wiki/Sovereign_citizen_movement

> Another common belief among sovereign citizens is that they can opt out of the purported contract, making themselves immune from the laws they do not wish to follow, by declining to "consent": when confronted by police officers or other officials, sovereign citizens typically attempt to negate their authority by saying, "I do not consent"

Like, why would this be true, and if it was, why would law enforcement and courts go along with it? I find it very odd.

gilleain

قبل 22 يومًاdiscuss

I suspect Sandi Toksvig, one of the hosts of QI. One of the 'success' messages is "quite interestng!".

No offence mean to anyone, but the whole exercise feels very QI : superficial 'understanding' of a large range of things (for example words) without much of a connection between these words.

gilleain

قبل 24 يومًاdiscuss

That's actually a feature that washing machines have had for a while - "generative washing" - it is where the extra odd socks come from.

gilleain

قبل 25 يومًاdiscuss

I recently bought the book 'Watch Repair for Beginners' for reference (a project I slightly unwisely agreed to do).

It has some great diagrams, but obviously nothing on these interactive animations (er, naturally, since it is a book).

However the author (Harold C. Kelley) has descriptions for the diagrams similar to a maths proof - like "Warning lever W is raised in position to engage the pin P ... The unlocking lever U lifts the drop lever D ..." - not easy to follow, but maybe if you have the mechanism in front of you!

gilleain

قبل 27 يومًاdiscuss

I meant metallocenes in general:

https://en.wikipedia.org/wiki/Metallocene

A metal atom sandwiched between two Cp rings. You _can_ model this as 5 single bonds between each atom of a ring (so 10 total C-M bonds), or you have to have some kind of 'edge' (bond) between the ring as a whole and the metal.

The more general issue is that a graph model of a chemical assumes a 'bond' is between exactly two atoms. Three-center hydrogen bonds are another example where this model fails to capture the chemistry very well.

Of course, it's a tradeoff - you can model _most_ compounds with just graphs (plus atom type, charge, chirality) and the relatively few that do not quite fit are special cases.

gilleain

قبل 27 يومًاdiscuss

There is a lot of graph theory in Chemistry - modelling chemicals as (vertex/edge coloured) graphs, reaction networks, etc.

Of course some molecules (eg aromatic systems, like ferrocene) are not naturally representable as graphs. I wonder if it is the same with synthesis - are there reactions hard to model as a graph (or petri net or whatever). One simple example I know is that you have to be careful with including a node for 'water' as it gets connected to everything else! Or at least in biochemistry it does.

gilleain

الشهر الماضيdiscuss

You got me. Usually I read them.

edit: Huh. Actually not a bad read. It even mentions ' On Growth and Form' which is interesting, if outdated. There are more modern texts like 'Shapes', 'Flow', and 'Branches' by Philip J Ball.

gilleain

الشهر الماضيdiscuss

Surface area to volume ratio?

gilleain

الشهر الماضيdiscuss

A very reasonable question, that I don't have an immediate concrete answer to!

Apparently I upvoted this question in the past (found it by searching for an answer - no AI, like the good old days)

https://biology.stackexchange.com/questions/2507/are-there-a...

One answer mentions actin and hexokinase. I'm not familiar with the actin fold, but looks like a bab sandwich of some kind.

Another commenter on this page mentioned the 'Rossman fold', another classic, and TIM barrels also occur to me. One caution is that some of these I would consider higher-level patterns - the 'Topology' level of CATH hierarchy.

Naturally, the more high-level (abstract) the fold pattern, the larger the sequence space it covers. It is less interesting to say that a helical bundle (for example) covers a lot of diverse sequences.

gilleain

الشهر الماضيdiscuss

That would be difficult - a metabolic map is a diagram showing the known reactions. At any point in time, only a subset of these will be active. Like a road map - at midnight, only some roads will have traffic.

I think what you are looking for is more like a model of the metabolome, showing the flow through the network under certain conditions (steady-state, growth, cell stress, etc). Not sure if there is a readily available database of such models, or how easy it would be just to run them and get meaningful results.

gilleain

الشهر الماضيdiscuss

> add too numbers

Did you do this on purpose to anger both Mathematicians and keen spellers?

gilleain

الشهر الماضيdiscuss

Heh, on my watchlist - ""The Unreasonable Effectiveness of Group Theory" - https://www.youtube.com/watch?v=1XsXRUsNEC4

gilleain

الشهر الماضيdiscuss

Yes, I was watching a video about that the other day - the 'dark proteome' or the 'ghost proteome' or similar.

gilleain

الشهر الماضيdiscuss

A random sequence may not fold at all! I seem to remember a paper that tried this, creating a bunch of random proteins, and checking how much structure they had - I think they were helical bundles, but don't quote me.

The nice thing about stable folds, is that 'nearby' sequences in sequence space - as in, point mutations - are the same fold. If each sequence had a completely different fold, then mutation would be much more destructive. Surprisingly, however, sequences that are far apart in sequence space can also adopt the same fold (convergent evolution).

gilleain

الشهر الماضيdiscuss

I see. I meant 'energetically accessible', but you mean more like 'affordably accessible' (in the sense that the molecular toolkit of a cell is what can 'afford' certain structures, due to chaperones available and so on).

Who knows what might be possible if you designed a cell from scratch - perhaps you could rework all the machinery to access other parts of fold space. After all, there are some weird and wonderful machines out there like the 'Vault' (https://en.wikipedia.org/wiki/Vault_(organelle)) that can fit whole proteins inside them. Possibly a different cage-like structure could help fold designed proteins into as-before unseen structures.

gilleain

الشهر الماضيdiscuss

This is about folds, not amino acids - even if you used a larger alphabet of residues, I somehow doubt that you would get many more folds.

Thinking more about the question of protein _length_ - I'm also not convinced that longer proteins (more than say 750aa) would produce more novel folds. Larger proteins tend to be multi-domain; that is, a longer chain will fold into multiple compact domains, each one a separate fold.

I suppose there could be 'megafolds' out there in fold space, beyond 1000aa - like a 12-bladed beta propeller, or a beta-helix with alpha helices on the outside or some other wacky thing. Whether that would substantially increase the numbers of total folds, I doubt, but that is of course a guess.

(ref - https://pmc.ncbi.nlm.nih.gov/articles/PMC10251718/ for protein lengths)

comments