--- Log opened Mon Feb 24 00:00:03 2020
01:41 < maaku> diybio suggestion: make a COVID-2019 vaccine
01:41 < maaku> according to what I've seen, this should be doable with diybio tools
01:43 < maaku> https://www.sciencenews.org/article/new-coronavirus-vaccine-development-process-accelerating
01:43 < maaku> .title
01:43 < saxo> Sorry, title took too long
01:43 -!- darsie [~kvirc@84-114-73-160.cable.dynamic.surfer.at] has joined ##hplusroadmap
01:44 < maaku> "So Inovio and other companies have developed ways to make vaccines much more quickly. For their SARS-CoV-2 vaccine, Inovio scientists convert the virus’s RNA into DNA and select pieces of the virus that computer simulations have suggested will prod the immune system into making antibodies. Those selected bits of DNA are then inserted into bacteria, which produce large quantities of protein snippets to be used in
01:44 < maaku> the vaccine... For Inovio’s product, it took three hours to design and about a month to manufacture."
02:25 -!- HumanG33k [~HumanG33k@62.147.242.8] has quit [Ping timeout: 240 seconds]
02:41 < nsh> heh
02:46 < fenn> it's the making the vaccine part that's hard. getting rid of pyrogens is not a backyard task
02:47 < fenn> that said, i do wonder what's supposedly taking industry so long. estimate of >1 year until first vaccine just seems absurd
03:10 -!- HumanG33k [~HumanG33k@62.147.242.8] has joined ##hplusroadmap
03:12 -!- HumanG33k [~HumanG33k@62.147.242.8] has quit [Remote host closed the connection]
03:13 -!- HumanG33k [~HumanG33k@62.147.242.8] has joined ##hplusroadmap
03:32 -!- Human_G33k [~HumanG33k@62.147.242.8] has joined ##hplusroadmap
03:35 -!- HumanG33k [~HumanG33k@62.147.242.8] has quit [Ping timeout: 260 seconds]
04:04 -!- CRM114 [~urchin@unaffiliated/urchin] has quit [Ping timeout: 255 seconds]
05:25 -!- CryptoDavid [uid14990@gateway/web/irccloud.com/x-cjnripnxkhkembxh] has joined ##hplusroadmap
05:34 < kanzure> maybe it's all the testing required by regulation
05:35 < fenn> they ought to be testing already then
05:36 -!- turona [~quassel@2a01:c22:3457:1800:82ff:a380:8d64:d5c8] has quit [Ping timeout: 272 seconds]
05:52 < lsneff> This is a good example of regulation that will result in loss of life.
06:15 < gnusha> https://secure.diyhpl.us/cgit/diyhpluswiki/commit/?id=fe263d69 Michael Folkson: Add Rusty webinar transcript >> http://diyhpl.us/diyhpluswiki/transcripts/blockstream-webinars/2019-07-31-rusty-russell-getting-started-with-c-lightning/
06:15 < gnusha> https://secure.diyhpl.us/cgit/diyhpluswiki/commit/?id=9966802d Bryan Bishop: Merge pull request #77 from michaelfolkson/add-rusty-webinar >> http://diyhpl.us/diyhpluswiki/
09:22 -!- yashgaroth [~ffffffff@172.58.22.157] has joined ##hplusroadmap
09:43 < darsie> If they don't test they may be liable for injuries.
10:24 < kanzure> "In vivo multi-dimensional information-keeping in Halobacterium salinarum" https://www.biorxiv.org/content/biorxiv/early/2020/02/15/2020.02.14.949925.full.pdf
10:35 < juri_> nmz787: interesting.
10:37 < juri_> nmz787: thanks for the bug reports. :D
10:39 < juri_> nmz787: I'll fix those this weekend. as for 'make it use all of the cores', add '+RTS -N -qg' to the end of the command line.
11:04 -!- CryptoDavid [uid14990@gateway/web/irccloud.com/x-cjnripnxkhkembxh] has quit [Quit: Connection closed for inactivity]
11:11 -!- Urchin [~urchin@unaffiliated/urchin] has joined ##hplusroadmap
11:16 < nmz787> juri_: with "-r 10" it took around 10 hours to render an STL... but it does indeed look pretty good and correct, and is 341MB ! The quality seems a bit rougher than I'd expect for a part this simple
11:16 < nmz787> timing it now with the extra args you provided
11:19 < nmz787> juri_: are there more efficient ways to view a rendering of the model in 3D, without using STL?
11:20 < juri_> nope. .stl is the best format it speaks, at the moment.
11:20 < juri_> I'm open to suggestions, however.
11:20 < nmz787> BRLCAD for example, on this model, finishes the CSG stuff in around 15-30 seconds, and I'm then able to use their viewer to interact with it effectively... STL generation from BRLCAD has been running for 8+ hours and still not done though
11:24 < juri_> can you tell whether their STL rendering is single, or multi-core?
11:32 < nmz787> top shows two processes with the converter name... and roughly equal runtime... so I guess it somehow chose 2 ?
11:32 < nmz787> though I thought when I first started it last night, it was only using 1
11:34 < juri_> yeah, i think implicitcad will level out at around 80 CPUs.
11:35 < nmz787> seems it's using 14 now
11:35 < juri_> but it's quite likely i'm wrong, and it will handle quite a bit more.
11:58 < lsneff> Should be possible to compile the csg to glsl and then real-time sphere trace it.
12:04 < kanzure> what resolution for STL generation are you using that makes it run for 8 hours?
12:06 < kanzure> not that i find it implausible.
12:06 < kanzure> because it's not.
12:07 < maaku> fenn: this appears to be a combination of testing schedules, and scaling up production
12:08 < maaku> if there's someone willing to put together a covid-19 vaccine, there may be some money available for it
12:08 < maaku> *for paying the cost of developing it, not buying it
12:09 < maaku> no idea what that would cost though
12:13 < yashgaroth> I feel like there are already several large biotech companies and governments working on a vaccine
12:15 < yashgaroth> most of the effort is in testing in monkeys or whatever animal we can actually infect with it, removing pyrogens isn't especially hard and for a diy effort you might as well leave some of them in
12:16 < yashgaroth> the other problem is that you eventually need to grow live, unattenuated covid-19 to test the monkeys' immunity, and I don't trust anyone in the diybio community with what would be BSL-3 containment
12:21 < yashgaroth> there doesn't seem to be an effective vaccine against e.g. SARS, 17 years after; and with how quickly covid is spreading and (presumably) mutating, I doubt whatever potential vaccine would still be effective in a few months
12:28 < yashgaroth> scale-up is much easier in E. coli of course since they divide every 20 minutes instead of every ~24 hours like animal cells, but antigen fragments like inovio's doing probably won't be effective; for viruses you really want to stimulate the intracellular response, e.g. T-cells attacking infected cells, rather than free-floating virus
12:45 < yashgaroth> oh nm they're not producing an antigen fragment in E. coli, they're doing a plasmid vaccine? if you find out the plasmid sequence I can crank out grams of it pretty easy, or short of that I could guess and design one expressing the spike protein
12:55 < kanzure> yashgaroth: what's the cost for cranking that out?
12:59 < yashgaroth> uhh probably ~$5k for the plasmid, maybe a bit more since the cheaper DNA synthesis houses are in China, I could fix up my bioreactor with another 5k, and probably another 5k for resins/chemicals/filters/columns/QC materials, and I've got access to an FPLC so that's free; inovio uses their fancy skin electroporator for delivery, but someone in here could make a knockoff pretty easy
12:59 < yashgaroth> so 15-20k plus labor
13:05 < yashgaroth> I've manufactured many a plasmid in my time, including DNA vaccines...the problem lies partially in design, since I'm just guessing what inovio is doing, but mostly in effectiveness: DNA vaccines usually suck at getting the body to generate a useful immune response
13:06 -!- Malvolio is now known as Guest52181
13:06 -!- Guest52181 [~Malvolio@unaffiliated/malvolio] has quit [Killed (card.freenode.net (Nickname regained by services))]
13:06 -!- Malvolio [~Malvolio@unaffiliated/malvolio] has joined ##hplusroadmap
13:08 < yashgaroth> inovio seems to have good preliminary data with their MERS vaccine with the same approach, but no one knows how effective it is at hampering live virus in humans
13:17 < kanzure> maaku: there you go.
13:32 < nmz787> kanzure: resolution that wouldn't be good enough for MEMS or nano/micro stuff
13:35 < kanzure> so it wasn't super low resolution and it's still taking forever?
13:35 < kanzure> wait.. super high.
13:40 < nmz787> yeah, STL apparently sucks as an output format for more than one reason
13:41 < nmz787> the implicitCAD re-run to STL took 45 minutes with 14 cores... trying to increase the resolution by 10X now and see how the STL looks
13:41 < nmz787> but even viewing the previous STL, which was 340MB, was painfully slow
13:42 < nmz787> every twist or zoom operation took many seconds
13:42 < nmz787> BRLCAD's mged viewer though, is smooth as butter
13:42 < nmz787> but Sean Morrison, one of the main devs over there, said STL generation is only single threaded... but might take a look at my output files to see if I'm costructing things inefficiently or something
13:47 < nmz787> however in BRLCAD, raytracing a single layer of the model and emitting a bitmap file of that slice, is pretty immediate
13:47 < nmz787> which is basically all I need for lithography purposes
13:53 -!- CryptoDavid [uid14990@gateway/web/irccloud.com/x-nmkkmjkqwqhsjpfh] has joined ##hplusroadmap
15:14 -!- hehelleshin [~talinck@98.29.27.253] has joined ##hplusroadmap
15:18 -!- Guest43421 [~talinck@98.29.27.253] has quit [Ping timeout: 260 seconds]
15:23 -!- Malvolio [~Malvolio@unaffiliated/malvolio] has quit [Quit: LEGALIZE SUBWOOFERS MAN]
16:19 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has quit [Ping timeout: 252 seconds]
16:43 -!- learnday20 [~learnday2@2601:190:c400:3550::1] has joined ##hplusroadmap
16:44 -!- N-time [~Mark@212.225.172.60] has joined ##hplusroadmap
17:18 -!- learnday20 [~learnday2@2601:190:c400:3550::1] has left ##hplusroadmap ["Leaving"]
17:54 -!- CryptoDavid [uid14990@gateway/web/irccloud.com/x-nmkkmjkqwqhsjpfh] has quit [Quit: Connection closed for inactivity]
18:06 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap
18:06 < maaku> yashgaroth: how stupid would it be to just pick out some viral genes related to the protein casing, produce those and inject with minimal testing?
18:07 < maaku> oh i should have kept reading. i assume that is what you mean by: <yashgaroth> oh nm they're not producing an antigen fragment in E. coli, they're doing a plasmid vaccine? if you find out the plasmid sequence I can crank out grams of it pretty easy, or short of that I could guess and design one expressing the spike protein
18:08 < maaku> yashgaroth: what would be the lead time on this?
18:12 < yashgaroth> yeah I figured out it was a DNA vaccine eventually, by talking it out with myself in the channel; lead time two months for the plasmid since it'll probably be total synthesis to get exactly what we need, and I expect it'll be roughly 10 kilobases...during that time I can acquire all the reagents etc, and after that, 2-4 weeks until I've got some high quality plasmid
18:12 < yashgaroth> so say 3 months, +/- a month
18:12 < maaku> aligns with their own reported timelines too
18:13 < yashgaroth> inovio already have their proprietary backbone to stick the spike gene into, so they have that advantage, but they'd be doing much more QC so maybe it balances out
18:14 < yashgaroth> I can check it for endotoxin and supercoiling, and it'll be clean enough (especially for a vaccine)
18:19 < yashgaroth> honestly the plasmid and its design are the most important part, I could ship the plasmid-bearing cells to anyone who has lab experience and they could get good enough quality if they have access to an incubator-shaker, freezer, centrifuge, and endofree gigaprep kits
18:20 < yashgaroth> but anyone receiving the vaccination would need access to the skin electroporator device, otherwise the delivery efficiency will be trash
18:29 < yashgaroth> re: how stupid would it be to inject with minimal testing; aside from injection-site reactions, your main risk is the vaccine being ineffective, but that's life
18:37 < gnusha> https://secure.diyhpl.us/cgit/diyhpluswiki/commit/?id=3e7e4823 Bryan Bishop: transcript: socratic seminar 6 >> http://diyhpl.us/diyhpluswiki/transcripts/austin-bitcoin-developers/2020-02-24-socratic-seminar-6/
18:51 -!- N-time [~Mark@212.225.172.60] has quit [Ping timeout: 255 seconds]
18:52 < maaku> yashgaroth: yes, that's my naïve thinking: worst likely case is that I'm out a bit of cash and still vulnerable
18:54 < yashgaroth> I could set up an ELISA for blood samples to check antibody titers against the spike protein, but even inovio's data from their MERS vaccine seems to imply it takes months before you see useful titers in a majority of test subjects
18:56 < maaku> yashgaroth: on the last day of a vacation presently, so I'll be afk for a while. but I'd like to follow up on this.
18:57 < yashgaroth> yeah let me know and I can write something up
19:21 < lsneff> Has anyone tried convergent assembly of 3d printed parts?
19:22 < lsneff> It'd be complicated, would require automated part removal, automated welding of parts together, etc, but it'd speed up big prints a lot
19:22 < lsneff> It have to grab parts and put groups of them together in mid-air with robotic arms.
19:28 < fenn> there are very few people even doing automated assembly of 3d printed parts
19:29 < fenn> one of the purported perks is that with 3d printing you can design a part such that no assembly is required
19:33 < lsneff> Theoretically, you could still design a part with just one or a few pieces, and software running on the printer would split it up into blocks that can be attached together automatically.
19:33 < lsneff> But yeah, more complexity, and also it's extremely difficult.
19:34 < lsneff> It's practice for nanofactories :P
19:39 < fenn> people do glue things together because their printer is not big enough
19:39 < fenn> someone capable of doing automated assembly would just buy a bigger printer instead
19:42 < lsneff> The point would be to speed up printing.
19:42 < fenn> oh, right
19:43 < lsneff> Essentially divide the printing time by the number of miniature printers supplying blocks into the convergent assembly process.
19:45 < lsneff> A machine with 30 mini printers would be able to print in an hour what it would take a single printer a day.
19:46 < lsneff> A thousand printers would take a print that would take a year and compress it into less than a day.
19:49 < lsneff> With really big arms, you could print buildings in days.
19:49 -!- HumanGeek [~HumanG33k@62.147.242.8] has joined ##hplusroadmap
19:49 < fenn> i think errors will accumulate
19:50 < lsneff> I agree. That's a big killer for the idea.
19:50 < fenn> maybe after lots of experimentation you could calibrate out systematic errors
19:50 -!- dongcarl8 [~dongcarl@unaffiliated/dongcarl] has joined ##hplusroadmap
19:51 < lsneff> Perhaps.
19:51 -!- Hooloo42 [Hooloovoo@sorunome.de] has joined ##hplusroadmap
19:51 < lsneff> At the very least, it appears to be a relatively unexplored space in manufacturing.
19:52 -!- Human_G33k [~HumanG33k@62.147.242.8] has quit [Ping timeout: 260 seconds]
19:52 -!- pasky [~pasky@nikam.ms.mff.cuni.cz] has quit [Ping timeout: 260 seconds]
19:52 -!- dongcarl [~dongcarl@unaffiliated/dongcarl] has quit [Ping timeout: 260 seconds]
19:52 -!- dongcarl8 is now known as dongcarl
19:52 -!- Hooloovo0 [Hooloovoo@sorunome.de] has quit [Ping timeout: 260 seconds]
19:53 < fenn> a large precision fixture table could also rather easily straighten out some errors
19:53 -!- pasky [~pasky@nikam.ms.mff.cuni.cz] has joined ##hplusroadmap
19:53 < fenn> you have small gaps between the parts full of glue, then the fixture table aligns the relevant mechanical interfaces
19:54 < fenn> when the glue sets the interfaces are accurately aligned
19:54 < fenn> probably more accurate than printing a big part
19:54 -!- Hooloo42 [Hooloovoo@sorunome.de] has quit [Client Quit]
19:55 < lsneff> You could have printed pegs and slots built into the blocks as well to align them automatically.
19:57 < lsneff> Or even snap fit joints to hold them together while adhesive or solvent cures.
19:58 -!- Hooloovo0 [Hooloovoo@sorunome.de] has joined ##hplusroadmap
20:03 -!- Malvolio [~Malvolio@unaffiliated/malvolio] has joined ##hplusroadmap
20:11 < fenn> no, printed pegs would contain errors
20:23 < lsneff> The intermediaries will be irregular shapes that will be difficult to line up with external surfaces.
20:25 < fenn> you'll just have to be quick and assemble before the glue dries
20:55 -!- darsie [~kvirc@84-114-73-160.cable.dynamic.surfer.at] has quit [Ping timeout: 252 seconds]
21:46 -!- yashgaroth [~ffffffff@172.58.22.157] has quit [Quit: Leaving]
22:10 -!- shawwwn [uid6132@gateway/web/irccloud.com/x-lbrzlyintabnqnbl] has joined ##hplusroadmap
22:59 -!- CRM114 [~urchin@unaffiliated/urchin] has joined ##hplusroadmap
23:03 -!- Urchin [~urchin@unaffiliated/urchin] has quit [Ping timeout: 272 seconds]
23:50 -!- N-time [~Mark@212.225.172.60] has joined ##hplusroadmap
--- Log closed Tue Feb 25 00:00:04 2020