--- Log opened Wed Nov 02 00:00:05 2022
00:29 -!- fenn [~fenn@user/fenn] has quit [Ping timeout: 260 seconds]
00:29 -!- fenn [~fenn@bryan.fairlystable.org] has joined #hplusroadmap
00:35 < nmz787> FPGAs are horribly power hungry
00:42 < nmz787> muurkha: just got micropython "ported" (in that I copied and mucked around with the G474 micropython files) and able to toggle an LED from the serial port REPL
00:42 < nmz787> on that STM32 Nucleo G491
00:43 < nmz787> turns out that the G474 was already supported by micropython, has more stock in the US (newark shipped the G491 from the UK, at least it was only $10 shipping), AND has basically the same specs from what I can tell (a few more ADC and DACs and Timers)
00:44 < nmz787> but oh well, I guess sitting here coding for a few hours is better than me even trying to return these boards, with the annoyance of customer service, repacking, driving to drop off the package
01:21 < nmz787> well here goes my progress... still needs the pins/pin-functions cleaned up at the least  https://github.com/micropython/micropython/pull/9818
01:27 -!- darkdarsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has joined #hplusroadmap
01:36 -!- darkdarsie is now known as darsie
01:46 -!- Malvolio [~Malvolio@idlerpg/player/Malvolio] has quit [Ping timeout: 246 seconds]
02:31 -!- livestradamus [~quassel@user/livestradamus] has quit [Read error: Software caused connection abort]
02:36 -!- livestradamus [~quassel@user/livestradamus] has joined #hplusroadmap
02:47 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has left #hplusroadmap []
03:05 -!- Netsplit *.net <-> *.split quits: mirage33552
03:21 -!- Malvolio [~Malvolio@idlerpg/player/Malvolio] has joined #hplusroadmap
03:43 < pasky> somehow tcl is one of the rare languages I really hate because its syntax is entirely incompatible with my brain - i'm not sure how it's possible but that's how it is, i have dreadful memories of hacking eggdrop bots and some tcl/tk scripts
04:55 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0::a324] has joined #hplusroadmap
05:00 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
05:35 -!- catern [~sbaugh@2604:2000:8fc0:b:a9c7:866a:bf36:3407] has quit [Remote host closed the connection]
06:28 -!- catern [~sbaugh@2604:2000:8fc0:b:a9c7:866a:bf36:3407] has joined #hplusroadmap
07:14 < docl> I wonder if anyone has written a better command language?
07:19 < docl> https://yosefk.com/blog/i-cant-believe-im-praising-tcl.html
07:48 < muurkha> nmz787: well, FPGAs do use more power than the same circuit fabbed in the same process node as an ASIC, but this ICE40UP5K I have here is fabbed in 40nm
07:49 < muurkha> glad to hear about micropython!  how much work was it?
09:02 < docl> trying to wrap my head around what makes a language "scripting" vs "command"
09:07 < docl> I guess stuff like whether it's function(arg,arg) vs proc arg arg. the latter being more command line esque, the former being more algebraic
09:08 < docl> e.g. people talk about bash scripting but it's more of a command language
09:23 < maaku> It’s just a windows vs. Unix vocabulary difference, no?
09:23 < muurkha> a programming language is optimized for reading and building levels of abstraction.  a command language is optimized for writing
09:23 < muurkha> a scripting language is a compromise between them
09:26 < maaku> There is definitely a difference between bash and Python, and I would welcome a different categorization.
09:30 < maaku> Also I don’t think command languages get enough love. I’d rather work in a modern command language than Python for scripting, but bash is archaic
09:31 < maaku> I wish there was a UNIX-y Powershell with typed dataflow
09:59 < jrayhawk> a command language uses executables in its primary namespace
10:00 < jrayhawk> hot take: if a shell does not use bytestreams as its default unit of composition, it is not a shell, but an IDE
10:01 < jrayhawk> i have done powershell programming, and it is incompetent at composing executables
10:15 -!- TMM_ [hp@amanda.tmm.cx] has joined #hplusroadmap
10:15 < docl> welcome TMM_ 
10:15 < TMM_> hello
10:16 < jrayhawk> for instance, try executing the following in powershell without buffering an entire file: aws polly synthesize-speech --output-format ogg_vorbis --voice-id 'Justin' --text "Bash is unnecssarily bad at many things, but it can at least compose commands extremely well." /dev/stdout | mpv -
10:18 < TMM_> hello docl! I'm not sure if we met before? I don'r recognize your handle
10:19 < docl> TMM_: I sent you a couple pms inviting you here, not sure if you saw them. I idle in #clasp
10:19 < TMM_> I saw one, that's why I'm here now
10:20 < docl> can you tell us anything about your apm project for godot?
10:20 < docl> right on :)
10:21 < TMM_> I'm not familiar with the abbreviation "APM"
10:21 < docl> atomically precise manufacturing, so basically drexlerian nanotech
10:21 < TMM_> Ahhh
10:22 < TMM_> Well, we're working with Eric on that project 
10:22 < TMM_> What in particular would you like to know?
10:23 < docl> is there any code yet?
10:26 -!- mrdata [~mrdata@user/mrdata] has quit [Read error: Software caused connection abort]
10:27 < TMM_> Prototype code, but yes
10:27 < TMM_> We're starting the production code tomorrow
10:27 < docl> nice!
10:28 < TMM_> We currently have rudimentary atomic editing, fairly complete rendering, very rudimentary OpenMM integration 
10:28 < TMM_> We have a fairly complete UI design and feature set document
10:28 -!- mrdata [~mrdata@135-23-182-185.cpe.pppoe.ca] has joined #hplusroadmap
10:28 < TMM_> We have funding, and engineers 
10:28 < TMM_> Also Eric Drexler :)
10:29  * docl high-fives TMM_ 
10:29 < TMM_> For the production phase we'll be removing our own PDB renderer and atomic model and move that into Cando 
10:29 < docl> maaku is launching a nanotech company also, might be good to coordinate
10:30 < TMM_> Once clasp can be linked as a library 
10:31 < docl> https://openmm.org/
10:32 < TMM_> Yeah, we are aware of OpenMM, we're using it now
10:32 < TMM_> We can relax structures made in our editor with it already
10:32 < docl> yeah, was just giving the link for future reference
10:33 < TMM_> Anyway, that's the current state of the project. We're well underway, we have a tentative schedule for September 2023 for our 1.0 release
10:34 < TMM_> Which will at least at an API level be fairly complete 
10:46 < docl> for reference, PDB = Protein Data Bank
10:50 < maaku> jrayhawk: I agree re: powershell. I just wish there was a more UNIX like command language that adopted the idea of typed binary data streams instead of text flow
10:51 < maaku> TMM_ is your code open source? Is it being developed in the open?
10:53 < TMM_> maaku: it will be open source, we will move the code to a public repository and develop in the open from then on when we have something that someone could download and run, even if it is in very early stages of development
10:54 < TMM_> maaku: this is a "marketing" decision in the sense that you don't get contributors to a project that doesn't exist yet. We want there to be something for a community of users and developers to form around. You generally don't get a second chance with telling the world about a project :)
10:54 < TMM_> We will develop with git from the start though, so the full development history will be available once we do go public
10:55 < TMM_> This will happen well before 1.0
11:05 < maaku> I've seen that line too many times to believe it. It would *highly* recommend developing it in the open from day one. 
11:06 < maaku> Otherwise you end up wasting a bunch of time later getting things ready for first public release. Stuff that ends up not being important and sometimes locks you into design corners.
11:06 < kanzure> TMM_: have you seen https://github.com/kanzure/nanoengineer
11:06 < maaku> The "you have to make a splash to get traction" thing is a myth. You can develop on github and develop momentum, having a PR event when you reach a milestone. Accomplishes the same thing, but more reliably
11:07 < maaku> TMM_: Also based on just the Drexler talk tha was posted a few days ago, I hope you're not being too constrained by his high level architectural engineering.
11:08 < maaku> Thinking you can just put PyMOL into a game engine as the renderer, or swap out Python as the scripting language shows he's never actually worked with a game engine before 
11:08 < maaku> I just mean that as constructive feedback if you have any control over the architecture.
11:11 < maaku> Of course I should mention I am somewhat biased as I would love for some other organization to take on development and maintenance of a nanocad environment instead of me
11:11 < maaku> But if it's not done with public development and open source, I'll just have to go my own way as I was planning on doing anyway
11:17 < docl> feel free to ignore this, of course, but I'm currently more a fan of fossil-scm (used by sqlite and tcl) than git. it's at least arguably better for cathedral style than bazaar style, which is the vibe I'm getting here. (plus I'm still a bit of a microsoft-phobe so I kind of have to be anti-github)
11:21 < maaku> docl: are there good web frontends to the fossil issue tracking system?
11:22 < docl> it has one built in
11:22 < maaku> I'm more at home at the CLI of course, but my first concern would be presenting a user-friendly bug reporting tool
11:22 < maaku> I see
11:22 < maaku> I should try it out.
11:23 < docl> basically ships with a web ui that you can use for issue tracking, wiki, timeline
11:23 < docl> doesn't have a good way to edit code in the browser, although it's possible
11:24 < maaku> but is there a simple way to host a site people can access from anywhere to create bug reports?
11:24 < maaku> that's more what I mean
11:25 < docl> yeah you can let them log in anonymously or enable registration via email
11:26 < docl> I'm planning to make my own fork/overlay system called fit, will post the code here probably this weekend
11:39 < docl> this might be worth reading to avoid gotchas: https://fossil-scm.org/home/doc/trunk/www/fossil-v-git.wiki
11:42 -!- spaceangel [~spaceange@ip-94-113-214-149.bb.vodafone.cz] has joined #hplusroadmap
11:46 < TMM_> maaku: no, Eric doesn't do the technology design. That's firmly in my grubby paws. 
11:46 < maaku> good
11:47 < TMM_> maaku: the software will absolutely with 100% certainty be open source. It's a requirement for the grants we got, and also my own ethical necessity 
11:48 < TMM_> there is truly absolutely no chance that that won't happen, there is no business model, there will be none other than getting grants 
11:48 < maaku> I'm just suggesting dog fooding open source processes instead of developing it in a proprietary way first. It just causes headaches for everybody otherwise (I'm speaking from experience)
11:50 < TMM_> I'm on the PLC of the Godot project, I'm really very well aware of how to do an open source project and how to make sure we don't fall into those pitfalls
11:50 < TMM_> I understand your concern, but this really isn't my first rodeo 
11:50 < TMM_> :)
11:51 < TMM_> We need to make some kind of splash in order to be able to go around with the hat for our next leg of development
11:52 < TMM_> kanzure: I'm aware of nanoengineer yes! I think Eric was involved with that some time in the distant past as well?
11:53 < kanzure> oh, probably.
11:54 < docl> it was used in this animation, right? https://www.youtube.com/watch?v=zqyZ9bFl_qg
11:54 < Muaddib> [zqyZ9bFl_qg] Nanofactory Animation (4:56)
11:54 < TMM_> It is part of our internal docs on good/bad ideas of exiting tools
11:54 < TMM_> I think one of our engineers spent quite a bit of time in it
11:58 < kanzure> docl: don't know, actually, it may not have been.
12:01 < docl> the end credits mention nanoengineer-1
12:06 < maaku> TMM_: how is the project currently funded?
12:06 < maaku> grants from which entities I mean
12:08 < maaku> but alright I'll drop the open source complaints. just be aware you are losing potential contributors
12:09 < kanzure> maaku in particular is one of those potential contributors
12:15 < TMM_> maaku: if you want you're welcome to get included earlier. This isn't about being secretive exactly 
12:16 < TMM_> maaku: as for current funding sources I'm not sure if I'm allowed to give specifics, if you really want to know I can ask whether that can be made public or not
12:17 < TMM_> It's a mostly no strings attached business though, no assignment of copyright or anything shady
12:18 < maaku> as long as there's no NDAs I would certainly like to be involved
12:18 -!- mrdata [~mrdata@135-23-182-185.cpe.pppoe.ca] has quit [Changing host]
12:18 -!- mrdata [~mrdata@user/mrdata] has joined #hplusroadmap
12:19 < docl> same here
12:19 < muurkha> jrayhawk: your definition of "command language" excludes Tcl, so it's an unrelated definition that isn't what we're talking about
12:19 < TMM_> Okay, I'll make sure that is possible 
12:21 < TMM_> And yeah, there's no need for NDAs. I think it's important for the project that we don't publicize the project until we have something of some substance to show. I hope to get at least some agreement on that point
12:21 < muurkha> TMM_: congratulations!
12:21 < kanzure> what feature set are you targeting?
12:24 < TMM_> We have a design document for our V1.0 feature set. When I've setup the environment to make it easy for outside contributors you can just read that and comment on it
12:25 < TMM_> It's a bit much to go into detail on on IRC. But the basic gist is a cad-like program for APM with heavy focus on user friendliness and to be useable by lay people 
12:26 < TMM_> A large push is also a game mode were we can teach people about the nano scale world to hopefully get a community of citizen engineers in the nano space
12:27 < kanzure> have you ever used any or all of: leocad, freecad, opencascade, solidworks, catia, openscad, autodesk 360 or autodesk mechanical?
12:27 < kanzure> (or nanoengineer for that matter)
12:27 < kanzure> oh, or cadnano.org
12:28 < TMM_> Yeah, we have looked at some of those, and the person who wrote cadnano is involved
12:28 < TMM_> Well, the original one, I think the current one isn't his doing 
12:30 < TMM_> I think there was a version in actionscript
12:31 < TMM_> That's the one Adam did
12:31 < TMM_> I'm not fully up to date on my nano engineering lore. :)
12:31 < kanzure> have you used CAM software before? like slicers or routers or gcode generator things.
12:31 < TMM_> No
12:32 < kanzure> it pertains to manufacturability of models and the automatic generation of instructions for its fabrication
12:32 < TMM_> I should probably get myself a 3D printer to get a sense of that
12:32 < TMM_> Yeah, I know roughly what CAM does, I've just never touched it
12:32 < docl> I linked this on here the other day BTW https://www.youtube.com/watch?v=52ViQDjDx2Y
12:33 < Muaddib> [52ViQDjDx2Y] Hein-Pieter van Braam, JJ, Ben-Joseph, Eric Drexler | MSEP (11:07)
12:33 < TMM_> Oh yeah, that's me
12:33 < TMM_> That was slightly terrifying 
12:33 < docl> yeah, and drmeister in the comments
12:34 < docl> you're famous now :)
12:34 < TMM_> At least I wore a good shirt 
12:35 < docl> yeah that's a cool shirt
12:35 < TMM_> I'm pretty used to public speaking but it's usually about stuff I'm actually a domain expert on
12:36 < docl> have you seen drmeister's old talks? the spiroligomer stuff is really interesting
12:37 < TMM_> I have not
12:37 < docl> https://www.youtube.com/watch?v=8X69_42Mj-g
12:37 < Muaddib> [8X69_42Mj-g] Clasp: Common Lisp using LLVM and C++ for Molecular Metaprogramming (57:38)
12:37 < TMM_> We spent several hours talking though
12:37 < kanzure> CAD is often the glamorous part, but eventually these machines need to be physically manufactured and there needs to be some sort of conversion from a 3d model to manufacturing or assembly instructions
12:37 < kanzure> part of the difficulty in molecular nanotechnology is that nobody knows, in fact, how to build any of it anyway
12:37 < kanzure> but there are some things that we do know how to build
12:37 < kanzure> for some very small subsets
12:37 < TMM_> Yeah, but since we have no real way of doing that yet I don't think it should be a serious use of resources 
12:38 < kanzure> protein design is pretty close and we do have lots of methods for protein design and fabrication
12:38 < docl> I kind of think he's found an approach that means we don't need DNA/protein as part of the path to apm. but it's protein like enough that the same tools should be relevant
12:38 < TMM_> Yeah, for the small subset of stuff we do know how to build I would like to have a single pipeline from screen up to production at some point
12:38 < TMM_> It is somewhat on the roadmap 
12:38 < maaku> kanzure: I would add solvespace and shapr3d to your list
12:38 < TMM_> But not for 1.0 I guess, at least not from our current funding sources 
12:39 < kanzure> docl: elaborate?
12:39 < TMM_> Eric's initial focus is on dense covailant machines
12:40 < TMM_> And our 1.0 roadmap is focused on that
12:40 < docl> spiroligomers are synthetic oligomers based on amino acids covalently bonded with double bonds
12:40 < maaku> TMM_: I actually really strongly disagree re: importance of CAM. It is very important at this stage.
12:40 < kanzure> yes but how's the 3d design tooling? is it easier to construct mechanical gears and rods and other structures than with protein engineering?
12:41 < TMM_> It is far easier
12:41 < maaku> NanoEngineer-1 worked as a CAD environment at the time. But being able to make designs is kinda pointless if you don't have at least some argument that they are manufacturable
12:41 < kanzure> and it will actually result in physical objects that have the planned structure?
12:41 < maaku> even assuming future manufacturing processes
12:41 < kanzure> i don't think i have seen a spiroligomer gear reported yet
12:41 < TMM_> Eric has a very specific vision on this 
12:41 < maaku> And a lot of the stuff made with NanoEngineer-1 was criticized for that reason--stressed bonds that require very advanced APM to manufacture, for example
12:43 < maaku> Being able to lay out a manufacturing process, including tooltip movements and workpiece geometry during bond formation is critically important.
12:43 < maaku> Especially if this is going to be creating IP libraries of reusable parts, as envisioned.
12:43 < muurkha> yeah, strained bonds are a problem
12:43 < muurkha> but also very useful!
12:43 < TMM_> We want to support CAD/CAM for structures we know how to build as well, but that's not the initial focus and it's not what the grant money was for. We WILL however make sure that in 1.0 this is possible to build such tools with the APIs is important 
12:44 < kanzure> docl: show me the spiroligomer version of baker's axle-rotor protein assemblies (or equivalent) https://www.bakerlab.org/wp-content/uploads/2022/04/science.abm1183.pdf
12:44 < docl> so like muurkha and I were discussing a few days ago, it's probably feasible to use spiroligomers to e.g. make apm diamond parts via a subtractive milling process. not clear on whether chemical etching would work, but could e.g. combine with ion beams (a spiroligomer mechanism would form a mask to block ions from spots you don't want milled)
12:44 < TMM_> If this is something you want to work on from very early on we can make sure we get the APIs you need
12:47 < TMM_> We want the project to be widely useful for practical science and engineering as well, and we will build the foundation to make that easy as part of our initial push. Without either extra funding or people like yourselves it will not be a planned feature set
12:47 < TMM_> Initially, I mean
12:48 < docl> I've tried downloading the cando stack, but I couldn't seem to get anywhere useful with it. I'd love something a bit more gamelike
12:50 < TMM_> That's the plan!
12:50 < docl> I'm excited :)
12:53 -!- Bike [~bike@pool-98-114-141-229.phlapa.fios.verizon.net] has joined #hplusroadmap
12:53 < TMM_> I'm going AFK for a while, my IRC client is basically always connected. I'll try to respond to future questions here as well
12:55 < docl> hi Bike 
12:56 < Bike> hello.
12:59 < muurkha> at least some APM diamond parts, but maybe not all of them
12:59 < muurkha> maybe not very many!
13:16 < docl> what's the argument that the diversity would be restricted?
13:18 < docl> fusing diamond nanocomponents to make bigger structures could be important, and also seems plausible
13:28 < docl> hmm. strained bonds might be a big limitation on what you can make by milling. if the teeth of your gears tend to just collapse into graphite you might not easily get anywhere
13:28 -!- drmeister [sid45147@id-45147.ilkley.irccloud.com] has joined #hplusroadmap
13:28 < docl> welcome, drmeister 
13:28 < drmeister> Howdy.
13:30 < docl> this chat was started way back by kanzure and fenn, anything apm related is definitely on topic
13:32 < drmeister> Yeah - I'm looking at the log.
13:32 < docl> right on
13:32 < kanzure> battle not with logs, lest ye become a log, and if you gaze into the log, the log gazes also into you.
13:34 < drmeister> Caught up.
13:34  * fenn sings the log song
13:35 < fenn> o/~ It's log, it's log, It's big, it's heavy, it's wood. It's log, it's log, it's better than bad, it's good. " Everyone wants a log You're gonna love it, log Come on and get your log Everyone needs a log Log log log o/~
13:35 < kanzure> are there any spiroligomers that have a validated physical structure known to be similar to gears, rods, shafts, or any other basic mechanical devices?
13:36 < drmeister> Spiroligomers will be used to create structures that are maybe 20-50% the size of proteins with much more rigid structure and enormous chemical functional diversity, meaning more ways to link them together through covalent bonds to create larger structures.
13:37 < drmeister> There are no spiroligomers that yet form gears.  Every spiroligomer is a rod/shaft.
13:37 < drmeister> Ah wait - I mis-spoke.
13:38 < drmeister> We've made structures like this - is that gear-like enough?
13:38 < drmeister> https://usercontent.irccloud-cdn.com/file/lm1gIotB/image.png
13:39 < drmeister> We are using them to make filters.
13:39 < kanzure> hmm, maybe, i suppose if it was sufficiently rigid
13:39 < kanzure> do you happen to know what is the stepwise yield of spiroligomer synthesis in the liquid or solid phase synthesis approach?
13:39 < drmeister> https://usercontent.irccloud-cdn.com/file/omeTG3OI/image.png
13:39 < drmeister> Ultimately to do things like separate metals from sea water.
13:40 < drmeister> Are you familiar with peptide or DNA synthesis?
13:40 < kanzure> more than i'd like to be
13:41 < drmeister> As good as peptides, better than DNA synthesis.
13:41 < kanzure> in phosphoramidite chemistry the stepwise yield is at 99 to 99.5% these days
13:41 < kanzure> and for some strange reason the terminal transferase oligonucleotide synthesis reactions are lower (DNAScript has only reported 60-80 nt molecules??)
13:41 < drmeister> No, you throw 10-20 equivalents of phosphoramidite at each coupling - so the yield is a fraction of a fraction of a percent.
13:42 < kanzure> i misspoke then; i meant the error rate. after 100 or 200 nucleobases you have the majority of the molecules with an incorrect sequence due to yield/error rate.
13:42 < yashgaroth> I think people mostly care about the yield of extended vs nonextended polymers per step
13:42 < drmeister> The coupling yield of a single coupling approaches 100% - we have the same thing going on.
13:43 < drmeister> We use peptide synthesis chemistry. So our coupling yields match peptide synthesis coupling yields.
13:43 < kanzure> there has been a lot of recent work in oligonucleotide chemistry to optimize the error rate down closer to zero
13:44 < kanzure> but i'm less familiar with recent work on peptide synthesis chemistry, not sure if that's as active any more
13:44 -!- spaceangel_ [~spaceange@ip-94-113-214-149.bb.vodafone.cz] has joined #hplusroadmap
13:44 < drmeister> The thing is we only need to make 4-mers to make a stiff strut 2 nanometers long.  Then we purify that and link it to other struts to create three-dimensional structures.
13:44 < drmeister> We've made almost 700 of them in the last year. 
13:45 < kanzure> in the land of DNA and proteins, to make a bigger protein of a specific shape you need like a ~1000mer
13:45 < fenn> drmeister: was the spiroligomer concept inspired by steve fowkes' nanopolymer systems? a colleague of his (tom nufert) mentioned that you had done some simulation work for them a couple years prior. http://www.nanopolymersystems.com/technology-summary
13:45 < drmeister> oligonucleotide chemistry needs to throw a lot of oligonucleotide at each coupling to achieve those high coupling yields because they need to make looong sequences.
13:46 < drmeister> fenn: No.  I don't know Steve Fowkes work.
13:46 -!- spaceangel [~spaceange@ip-94-113-214-149.bb.vodafone.cz] has quit [Ping timeout: 252 seconds]
13:46 < fenn> apparently nanopolymer systems could never get the chemistry to work out; they would just get random garbage instead of the sequence they were after. i thought it needed a capping/deprotection step like in DNA synthesis in order to add the monomers in sequence
13:47 < drmeister> I see - those polymers are achiral and connected by single bonds.  Flat and floppy.
13:49 < drmeister> kanzure: That's DNA and proteins - I don't think you need that for bootstrapping nanotech.   Proteins and DNA are very soft and require folding.  Protein folding is a grand challenge of science.   
13:50 < drmeister> What I think we want is rigid, brick-like structures with complementary chemical surfaces that link up through a combination of self-assembly and directed assembly to create larger rigid structures.
13:51 < kanzure> are there any software tools where i can put in a three-dimensional model or description and out pops the spiroligomer steps? or at least an analysis that spiroligomers aren't known to be able to construct this object etc?
13:51 < docl> are the protein folding AI announcements we've been seeing overhyped?
13:51 < docl> .t https://www.newscientist.com/article/2330866-deepminds-protein-folding-ai-cracks-biologys-biggest-problem/
13:51 < saxo> Protein folding: DeepMind AI cracks biology's biggest problem | New Scientist
13:52 < kanzure> i think for certain kinds of protein structures we can make those reliably, although not arbitrary structures
13:52 < kanzure> not yet anyway.
13:52 < drmeister> kanzure: Cando is the software tool for that.   We are still working on the applications that do what you are asking for.
13:53 < drmeister> I'm very impressed by what the protein folding community has come up with - but in a way it's still "looking under the lamp post". They use X-ray protein structures of proteins to predict structures of proteins.
13:54 < kanzure> ah, this? https://github.com/cando-developers/cando
13:55 < drmeister> The big problem that few people are aware of is that protein design hasn't cracked the catalyst design problem.   Catalysts are molecules that make other molecules.   It's the most important nanotech capability to develop - because it would let us create feedstocks from inexpensive materials.
13:56 < fenn> "CanDo’s online server runs on the commercial finite element program Abaqus" huh ok
13:56 < drmeister> fenn: Not that Cando.
13:57 < fenn> plz link then
13:57 < drmeister> https://github.com/cando-developers/cando
13:57 < drmeister> It's an unfortunate coincidence of naming.  I did publish with it first.
13:57 < drmeister> I'm pretty sure I did... oh well.
13:58 < kanzure> in molecular biology the common refrain is "structure is function"; just having a set of mechanical protein components would be quite helpful for nanotech, although i agree catalysis and functionalization would be ideal.
13:58 < kanzure> is there something other than x-ray crystallography that protein people should be doing for structure validation?
13:58 < drmeister> "Structure is function" is the central principle of nature.
14:00 < drmeister> Catalysis would let us create solar fuels from CO2 and sunlight; convert inexpensive alkanes into valuable feedstocks; cleave cross-links that form wrinkles and lead to aging and create everything we depend on from natures starting materials: CO2, N2, H2O, and O2.
14:02 < drmeister> X-ray crystallography is awesome.  Cryo-EM and microcrystalline electron diffraction are new approaches that are coming online.
14:02 < drmeister> But really, what we need is more people working on making molecules.
14:02 < kanzure> .wik Microcrystal electron diffraction
14:02 < saxo> "Microcrystal electron diffraction, or MicroED, is a CryoEM method that was developed by the Gonen laboratory in late 2013 at the Janelia Research Campus of the Howard Hughes Medical Institute." - https://en.wikipedia.org/wiki/Microcrystal_electron_diffraction
14:06 < maaku> drmeister: I thought metal recovery from seawater was not economically viable on energy grounds
14:07 < maaku> docl: alphafold is being overhyped, but it is an important tool nevertheless
14:08 < maaku> the article you linked to is their announcement that they simulated folding of all PDB proteins with alphafold. 
14:09 < maaku> not that they simulated it correctly, or got the right fold, but that they ran alphfold on all those inputs. it's a nothingburger story
14:14 < drmeister> Most metal recovery from seawater is not economical using current technologies.   Uranium is just at the cusp of being viable.
14:14 < maaku> let me rephrase. i thought the viability was based on fundamental energy expenditures, not current technology levels. is that not so?
14:15 < maaku> also I would presume this is much more viable when working on the brine waste water of a desalination plant
14:16 < maaku> one more question, since this is something I've thought about: if you're at the cusp of viability, wouldn't it make sense to use existing sea salt deposits?
14:21 < drmeister> I don't think so.  Imagine you had a beachball covered in a membrane full of atomically precise pores that only allow passage of one metal ion.  Inside the beach ball you have a electrochemical half cell that reduces the metal ion to metal or you have an anion that complexes with the metal ion and forms an insoluble salt.
14:21 < drmeister> You toss it in the ocean and pick it up six months later.
14:22 < drmeister> It's full of uranium or gold.
14:24 < drmeister> It's like bioconcentration of metals - it happens because of selective ion channels and some sequestration mechanism.
14:24 < docl> does the energy that pushes the ion through come from the heat of the seawater in that example?
14:24 < drmeister> The devil is in the details of the ion channels.
14:24 < drmeister> No, it's just a selective filter and a sink on the inside.
14:25 < drmeister> Yes, I guess you can say the energy that pushes the ion through comes from the heat of the seawater.   
14:25 < drmeister> It's selective diffusion.  It's not an impossible thing like a Maxwell's demon.
14:26 < drmeister> If you can create atomically precise channels that selectively pass ions - then you could create this.
14:26 < maaku> drmeister: ok yes, that makes sense. thanks!
14:26 < drmeister> That's going to be a nanostructure on the scale of maybe 3x2x2 nanometers.  About 5-10 kilodaltons.
14:27 < drmeister> Nature does this all the time.  You can't swing a cat without hitting a selective ion channel.
14:27 < yashgaroth> probably easier to start with simple metal coordination, coat a large enough membrane with a compound that binds uranium. Recover metal off the surface with a pH change or something and then dunk it back in the sea
14:27 < drmeister> It just has no interest in uranium or lithium or rare earth metal ions.
14:28 < drmeister> Right - but if you do it with a channel - that acts like a catalyst.  One channel passes many, many, many ions.
14:29 < drmeister> You get an inflection point in price/performance.
14:29 < drmeister> cost/benefit.
14:29 < drmeister> One of those.
14:30 < yashgaroth> I've no experience with spiroligomers, but for proteins at least the "floppiness" is critical in eg forming a selective ion channel. Though they'd get immediately destroyed by proteases in the ocean, so there is that
14:30 < drmeister> I put it out there as an example of an insanely valuable nanoscale device that is useful all by itself.
14:31 < maaku> that and structural diamond, and diamond-void quantum computers
14:31 < drmeister> There is "floppiness" and there is "FLOPPINESS".
14:31 < yashgaroth> oh definitely, even eliminating the horrific extraction practices in mining of rare earths etc would be a good use case
14:31 < docl> yeah, and this kind of apm filtration could also be used in artificial kidneys
14:31 < yashgaroth> if you've got enough disulfides they stay pretty rigid
14:32 -!- sivoais [~zaki@199.19.225.239] has quit [Read error: Software caused connection abort]
14:32 < drmeister> Everything at the atomic scale at room temperature flops around - it's a matter of degree.   Diamond to hydrocarbon chains.
14:32 < drmeister> The floppiness I was criticizing is the big-time floppiness of folding polymers.
14:32 < yashgaroth> yeah I see those single bonds in the spiro R-groups
14:33 < drmeister> I can criticize them because I worked with proteins for 5 years.   I created the first large, unnatural, designed protein.  Designed, expressed, purified, crystallized 4HB1.
14:33 < fenn> docl: the energy that pushes the ion through is osmosis caused by the difference in concentration on the inside vs outside. it's really just that there are more ions randomly going in than randomly going out, because they get stuck there once inside (by electroplating or whatever)
14:33 < drmeister> https://www.rcsb.org/structure/4hb1
14:35 < drmeister> I hates them, nasty proteins - but I'm made out of them.  Go figure.
14:35 < drmeister> That's why I invented spiroligomers.
14:35 < yashgaroth> haha I can only defend proteins so much, they are a pain
14:36 < drmeister> Everyone fixates on them and DNA because that's all they know.
14:38 -!- sivoais [~zaki@199.19.225.239] has joined #hplusroadmap
14:38 < yashgaroth> the advantages of ribosomal synthesis and scaleup are pretty strong, it's just a question of whether "AI" folding prediction can beat rational design from a rigid scaffold
14:39 < drmeister> Ribosomes are awesome. 
14:41 < yashgaroth> naturally (pardon the pun). So are you more focused on spiroligomers for catalysis, protein binding, or nanostructure stuff at the moment?
14:42 < docl> fenn: that makes sense to me, but I get the impression people think separation requires energy inherently, in a way that scales up the more dilute the solute is in the solution. it seems like the osmotic pressure pushes in whatever kind of ion it can fit through the holes, though, so it's not quite like the case of 1ppb or whatever in a pure water solution
14:46 < docl> I'm thinking if you just had the beachball full of pure water it would eventually reach equillibrium with the ocean's salt concentration, just with your selected metal inside as the solute instead of everything else (assuming it doesn't precipitate out at that concentration)
14:49 < kanzure> i must be misunderstanding something; how do you get to arbitrary 3d shapes from just ~4 mers?
14:59 < kanzure> can you get arbitrary degrees of freedom from the full (admittedly small) library of all possible 4mer spiroligomers?
15:00 < kanzure> coupling yield would be an issue as you add more spiroligomers to the construction i imagine, unless you're doing something stochastic?
15:03 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has quit [Ping timeout: 268 seconds]
15:05 < muurkha> docl: precisely because fusing diamon nanocomponents could be difficult
15:06 < muurkha> docl: yes, separation does require energy (exergy) inherently, because Maxwell's Demon
15:13 -!- Gooberpatrol66 [~Gooberpat@user/gooberpatrol66] has quit [Quit: Leaving]
15:21 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
15:21 < fenn> mixing the pure water with your selected ion increases the entropy of the beachball system
15:21 < fenn> it would take energy to get the ions back out
15:22 < docl> muurkha: do you suspect drmeister's claim is incredible then? say we're talking gold. it's an incredibly dilute solute in seawater, parts per trillion
15:22 < fenn> kanzure: theyre just doing 4-mers for now; it's not a rigid limit
15:26 < docl> gold would precipitate out above a certain concentration. so if you have 1ppt in saltwater containing 35 grams of salt per liter, that's not going to precipitate out, but if you move it across a membrane into pure water, it's going to precipitate out long before it reaches the equivalent molar concentration vs the salt outside
15:27 < docl> it's not that there's no energy involved, just that it's very small, i.e. what you get by dissolving 35 grams of salt in pure water
15:29 < drmeister> kanzure: You fold them up and lock the ends - like a folding ruler but chunkier.
15:29 < drmeister> I gotta run guys - it's nice to meet you all.
15:29 < docl> have a good one
15:29 < docl> thanks for chatting with us
15:49 < muurkha> docl: I don't understand the proposed mechanism, but I think you're right that the thermodynamically minimal energy required to purify things is orders of magnitude smaller than the energy consumed by currently popular processes
15:50 -!- spaceangel_ [~spaceange@ip-94-113-214-149.bb.vodafone.cz] has quit [Remote host closed the connection]
16:05 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has quit [Ping timeout: 276 seconds]
16:08 < kanzure> "Evolutionary-scale prediction of atomic level protein structure with a language model" https://www.biorxiv.org/content/10.1101/2022.07.20.500902v2
16:50 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
17:25 -!- Hooloovoo [~Hooloovoo@hax0rbana.org] has quit [Quit: ZNC 1.8.2+deb2+b1 - https://znc.in]
17:26 -!- Hooloovoo [~Hooloovoo@hax0rbana.org] has joined #hplusroadmap
17:41 -!- Malvolio is now known as CVDFNSTR
17:47 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has quit [Ping timeout: 272 seconds]
18:44 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0::a324] has quit [Quit: Leaving]
18:46 -!- Gooberpatrol66 [~Gooberpat@user/gooberpatrol66] has joined #hplusroadmap
22:36 -!- superkuh [~superkuh@user/superkuh] has quit [Excess Flood]
22:39 -!- superkuh [~superkuh@user/superkuh] has joined #hplusroadmap
23:24 -!- faceface [~faceface@user/faceface] has quit [Remote host closed the connection]
23:28 -!- fenn [~fenn@bryan.fairlystable.org] has quit [Changing host]
23:28 -!- fenn [~fenn@user/fenn] has joined #hplusroadmap
--- Log closed Thu Nov 03 00:00:06 2022