--- Log opened Tue Sep 30 00:00:54 2025
00:19 < hprmbridge> kanzure> "Connectome-seq: high-throughput mapping of neuronal connectivity at single-synapse resolution via barcode sequencing" https://www.biorxiv.org/content/10.1101/2025.02.13.638129v2.abstract
00:19 < hprmbridge> kanzure> "Connectome-seq relies on the labeling of synaptic connections using engineered protein-RNA complexes to translate a synapse into a pair of unique RNA barcodes. The intended outcome is that one barcode will be associated with the pre-synaptic neuron and the other with the post-synaptic neuron, allowing for later profiling via high-throughput sequencing."
00:20 < hprmbridge> kanzure> "Two protein anchors, termed SynBar (synaptic barcoding), were designed: PreSynBar to target presynaptic terminals and PostSynBar to target postsynaptic sites. PreSynBar consists of a modified neurexin 1β protein fused to the large fragment of split-GFP (GFP1-10) and an RNA-binding domain, λN22 peptide, similar to other reported designs. PostSynBar is based on neuroligin 1, also fused to the
00:20 < hprmbridge> kanzure> complementary small fragment of split-GFP (GFP11) and the same RNA-binding domain."
00:21 < hprmbridge> kanzure> "These trans-membrane protein constructs are designed to interact at synapses, reconstituting GFP and bringing together their associated RNA barcodes to anchor at the synaptic membranes. The protein scaffolds were largely adopted from the SynView constructs, where split-GFP fragments are embedded inside the synaptic proteins rather than at their N-terminals."
00:24 < hprmbridge> kanzure> what:
00:25 < hprmbridge> kanzure> "Nuclei contain both cellular transcriptomes and copies of either PreRNA or PostRNA, allowing identification of cell types and their associated barcodes. During gentle brain tissue homogenization, the cell membranes connecting pre-synaptic terminals to axons and post-synaptic sites to dendrites are physically separated. The exposed membrane surfaces at both the pre- and post-synaptic compartments
00:25 < hprmbridge> kanzure> then reseal, forming structures called synaptosomes. Since PreSynBar and PostSynBar proteins are connected across the synaptic cleft, tightly bound to the associated RNA barcodes, these protein-RNA complexes should remain intact as long as the synaptic interface is preserved. This allows synaptosomes to maintain the paired PreRNA and PostRNA information from connected neurons, providing direct
00:25 < hprmbridge> kanzure> evidence of synaptic connections. By matching barcodes between nuclei and synaptosomes, we can link the identity of connected neurons and map their connectivity patterns."
00:27 < hprmbridge> kanzure> nani! "Workflow for parallel isolation of nuclei and synaptosomes. Following AAV injection into pre- and postsynaptic regions, brain tissue is processed through differential centrifugation. Slow spins yield crude pre- and post-nuclei, while a fast spin of the supernatant of the slow spin produces the synaptosome fraction."
01:57 < hprmbridge> kanzure> https://www.aleksagordic.com/blog/matmul
02:10 < fenn> it would make sense that NoiseStep and other training algorithms that don't rely on matmul are being suppressed by those with interests in maintaining the world order
02:10 < fenn> regardless, it's odd that no large bitnet models exist
02:11 < fenn> (bitnet and friends shouldn't require matmul, only addition)
02:48 < kanzure> okay, let's see.
02:49 < kanzure> first, assume genome contains all pairwise barcodes for synaptic connections and neuronal cell identities from previous connectome-seq sequencing datasets
02:49 < kanzure> each neuron randomly activates one identity from the genomic dataset, or possibly some hierarchical system of identity generality -> specificity over time
02:50 < kanzure> express corresponding synaptic proteins, RNA barcodes, etc to each synapse connected to the cell
02:50 < kanzure> invent some kind of way to measure a "synaptic match score" or "distance metric" with each neighbor or all neighbors and issue some sort of signal from the synapse that is trafficked back to the soma or nucleus
02:51 < kanzure> these signals once in the nucleus would, uh, probably be genomically integrated in a temporary memory location because it won't happen all at once, and be used for calculating a confidence score as to whether the neuron picked the right identity in the network based off of the observed neighboring identities
02:52 < kanzure> neurons with low internal confidence scores then switch their identity by swapping with a neighbor (?? not sure how to do this deterministically and force the other neuron to change their identity), or pick an identity from the genome that is high confidence but not also already claimed by a neighbor (??), magic happens here
02:53 < kanzure> allow this program to execute over multiple weeks or months. um also we need a fully connected network where all neurons are connected to all neurons from the beginning i think. this is biologically implausible for super-large neuronal networks but maybe okay for smaller networks.
02:54 < kanzure> if there is a conflict of multiple neurons claiming the same neuron (not sure how to detect this since it would literally see self-self but whatever) then you might be able to backtrack the neuronal identity to an earlier or more abstract identity from the genome, or otherwise re-negotiate with peers.
02:55 < kanzure> eventually a mechanism of synaptic pruning would be required to eliminate wrong/unspecified synapses.
02:56 < kanzure> this is based in part on the botnet/freenet-style oskar sandberg virtual address swapping technique: https://diyhpl.us/~bryan/papers2/security/The%20RedKing%20Hivemind%20-%202022.pdf
02:59 -!- WizJin_ [~Wizzy@2402:a00:184:8354:28ae:ac51:4309:fcb4] has quit [Read error: Connection reset by peer]
02:59 < kanzure> without full connectivity, you would have to grow new synapses and target them somehow and then figure out who you connected to. and then your identity might change based on who you connected to? or something? or your (new) local topology may not have any role in the final converged network according to the genomic information... dunno.
03:00 -!- WizJin_ [~Wizzy@2402:a00:184:8354:28ae:ac51:4309:fcb4] has joined #hplusroadmap
03:00 < kanzure> the alternative strategy is the "genomic bottleneck" concept where you statistically encode neural connectivity by a general genomic program that grows a brain with a certain architecture from scratch.
03:10 < kanzure> maybe you can do better than statistical connectivity encoding with a mitosis strategy. maybe for local connectivity.. not for long-range projections though, which seem to be targeted? or fenn's balloon expansion thing.
03:13 < fenn> with the balloon expansion, long range connections would have the same level of specificity as short range connections
03:13 < fenn> because they *are* short range connections, at least at the start
03:24 < fenn> "## Postscript: A note on dual uses
03:24 < fenn> While all of my examples are malicious, a
03:24 < fenn> network like this can obviously be used for good."
03:50 < kanzure> okay sure, if you pattern the long-range connections in early life when the neural network is small, and then let growth happen after that patterning, then they become long-range projections later.
04:46 < kanzure> ah, i like being right:
04:46 < kanzure> "Network cloning using DNA barcodes" https://arxiv.org/abs/1611.00834 (2016)
04:46 < kanzure> "Network cloning using DNA barcodes" https://pmc.ncbi.nlm.nih.gov/articles/PMC6511037/ (2019)
05:03 < fenn> it took 3 years to get published?
05:03 < fenn> sheesh
05:55 -!- TMM [hp@amanda.tmm.cx] has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
05:55 -!- TMM [hp@amanda.tmm.cx] has joined #hplusroadmap
06:10 -!- gl00ten [~gl00ten@2001:8a0:7ee5:7800:46d9:f5c:17a2:432] has joined #hplusroadmap
06:51 -!- WizJin__ [~Wizzy@2402:a00:184:8354:28ae:ac51:4309:fcb4] has joined #hplusroadmap
06:54 -!- WizJin_ [~Wizzy@2402:a00:184:8354:28ae:ac51:4309:fcb4] has quit [Ping timeout: 252 seconds]
15:03 -!- catalase [~catalase@user/catalase] has quit [Remote host closed the connection]
15:11 < hprmbridge> kanzure> "Induction of experimental cell division to generate cells with reduced chromosome ploidy" https://www.nature.com/articles/s41467-025-63454-7
15:11 -!- catalase [~catalase@user/catalase] has joined #hplusroadmap
15:11 < hprmbridge> kanzure> "Somatic cell nuclear transfer (SCNT) enables the direct reprogramming of somatic cells into functional oocytes, albeit with a diploid genome. To address ploidy reduction, we investigated an experimental reductive cell division process, termed mitomeiosis, wherein non-replicated (2n2c) somatic genomes are prematurely forced to divide following transplantation into the metaphase cytoplasm of
15:11 < hprmbridge> kanzure> enucleated human oocytes. However, despite fertilization with sperm, SCNT oocytes remained arrested at the metaphase stage, indicating activation failure. Artificial activation using a selective cyclin-dependent kinase inhibitor successfully bypassed this arrest, inducing the segregation of somatic chromosomes into a zygotic pronucleus and a polar body. Comprehensive chromosome tracing via
15:11 < hprmbridge> kanzure> sequencing revealed that homologous chromosome segregation occurred randomly and without crossover recombination. Nonetheless, an average of 23 somatic chromosomes were retained within the zygote, demonstrating the feasibility of experimentally halving the diploid chromosome set. Fertilized human SCNT oocytes progressed through normal embryonic cell divisions, ultimately developing into embryos
15:11 < hprmbridge> kanzure> with integrated somatic and sperm-derived chromosomes. While our study demonstrates the potential of mitomeiosis for in vitro gametogenesis, at this stage it remains just a proof of concept and further research is required to ensure efficacy and safety before future clinical applications."
15:11 < hprmbridge> kanzure> but what about maternal methylation or imprinting?
15:12 < hprmbridge> kanzure> "donor oocyte cytoplasm" that's cheating, unless its interspecies ooplasm transfer.
15:17 < hprmbridge> kanzure> "We have successfully utilized the residual metaphase activity and meiotic machinery in the cytoplasm of enucleated human MII oocytes to implant donor MII spindles (known as spindle transfer) or PB1 genomes (PB1 transfer) to induce de novo spindle formation."
15:21 < hprmbridge> kanzure> well it was not interspecies. if you need human donor oocytes you're not exactly making lots of progress here...
16:04 < hprmbridge> kanzure> https://fdamuseum.com/
16:35 -!- Gooberpatrol66 [~Gooberpat@user/gooberpatrol66] has quit [Quit: Konversation terminated!]
17:23 < hprmbridge> kanzure> "Rescue of oocytes recovered from postmortem mouse ovaries" https://www.jstage.jst.go.jp/article/expanim/72/1/72_22-0043/_article/-char/ja/
17:28 < hprmbridge> kanzure> "We examined the effect of ooplasm aging on the developmental potential of SCNT embryos in bovine (Tani et al., 2003) and in mouse (Liu et al., 2007). When bovine cumulus cells at the G0 phase and M-phase are fused every hour after activation for 6 h, the potential to develop into blastocysts after SCNT gradually decreased with time after activation. Ooplasm 2 h and more after activation did not
17:28 < hprmbridge> kanzure> support the development of G0-phase cumulus nuclei to blastocysts. When M-phase cumulus cells were fused, the ability to develop into blastocysts dramatically decreased beginning at 6h postactivation. These findings clearly demonstrate that some reprogramming factors present in the ooplasm decrease after artificial activation."
17:29 < hprmbridge> kanzure> okay, but that's fairly remarkable. how many times does a single oocyte support reprogramming? did they only check hourly fusion reprogramming events?
17:37 < hprmbridge> kanzure> "Two-staged nuclear transfer can enhance the developmental ability of goat–sheep interspecies nuclear transfer embryos in vitro" https://link.springer.com/article/10.1007/s11626-010-9363-6
17:38 < hprmbridge> kanzure> "In the first stage nuclear transfer (FSNT), GFFs were injected into the ooplasm of enucleated sheep metaphase-II oocytes, then non-activated reconstructed embryos were cultured in vitro, so that the donor nucleus could be exposed to the ooplasm for a period of time. Subsequently, in the second stage nuclear transfer, FSNT-derived non-activated reconstructed embryo was centrifuged, and the donor
17:38 < hprmbridge> kanzure> nucleus was then transferred into another freshly enucleated sheep oocyte. Compared with the one-stage nuclear transfer, two-stage nuclear transfer could significantly enhance the blastocyst rate of goat–sheep interspecies clone embryos, and this result indicated that longtime exposure to xenogenic ooplasm benefits the donor nucleus to be reprogrammed."
17:41 < hprmbridge> kanzure> "Reprogramming of two somatic nuclei in the same ooplasm leads to pluripotent embryonic stem cells" https://academic.oup.com/stmcls/article-abstract/31/11/2343/6449156 "double SCNT"
17:42 < hprmbridge> kanzure> "In conclusion, we have shown that the oocyte's reprogramming capacity is in excess of a single nucleus"
18:32 < kanzure> you should be able to inject a neuron nucleus into a synaptosome
18:32 < kanzure> via nuclear transfer
18:32 < kanzure> with emulsifiers or somatic cell fusion you could optionally expand the compartment.. probably.
18:40 -!- stipa_ [~stipa@user/stipa] has joined #hplusroadmap
18:43 -!- stipa [~stipa@user/stipa] has quit [Ping timeout: 255 seconds]
18:43 -!- stipa_ is now known as stipa
18:57 < kanzure> https://www.github.com/BoxuanZhaoLab/connectome-seq-code
19:11 <+gnusha> https://secure.diyhpl.us/cgit/diyhpluswiki/commit/?id=003f1247 Bryan Bishop: germline upgrades to assist with brain uploading >> http://diyhpl.us/diyhpluswiki/mapseq/
19:11 < kanzure> i don't like how the ikiwiki table of contents anchor tags are randomly generated and not related to the section title
19:13 <+gnusha> https://secure.diyhpl.us/cgit/diyhpluswiki/commit/?id=24b8ff88 Bryan Bishop: oops fix cell lineage tracing link >> http://diyhpl.us/diyhpluswiki/genetic-modifications/
19:17 <+gnusha> https://secure.diyhpl.us/cgit/diyhpluswiki/commit/?id=a8afff87 Bryan Bishop: fix mapseq page, move connectome-seq >> http://diyhpl.us/diyhpluswiki/mapseq/
19:38 -!- gl00ten [~gl00ten@2001:8a0:7ee5:7800:46d9:f5c:17a2:432] has quit [Ping timeout: 256 seconds]
19:58 <+gnusha> https://secure.diyhpl.us/cgit/diyhpluswiki/commit/?id=8b4706f8 Bryan Bishop: consider more genetic changes for brain-computer interfaces >> http://diyhpl.us/diyhpluswiki/genetic-modifications/
20:09 -!- gl00ten [~gl00ten@87-196-72-24.net.novis.pt] has joined #hplusroadmap
20:24 -!- gl00ten2 [~gl00ten@nat-vpn-fm.net.ul.pt] has joined #hplusroadmap
20:25 -!- justanotheruser [~justanoth@gateway/tor-sasl/justanotheruser] has quit [Remote host closed the connection]
20:25 -!- justanotheruser [~justanoth@gateway/tor-sasl/justanotheruser] has joined #hplusroadmap
20:27 -!- gl00ten [~gl00ten@87-196-72-24.net.novis.pt] has quit [Ping timeout: 256 seconds]
20:53 -!- Gooberpatrol66 [~Gooberpat@user/gooberpatrol66] has joined #hplusroadmap
21:39 -!- gl00ten2 [~gl00ten@nat-vpn-fm.net.ul.pt] has quit [Ping timeout: 240 seconds]
22:03 -!- flyback [~flyback@2601:540:c700:2380:d4f1:bf17:a77d:bd7a] has quit [Ping timeout: 244 seconds]
22:32 -!- gl00ten2 [~gl00ten@2001:8a0:7ee5:7800:46d9:f5c:17a2:432] has joined #hplusroadmap
22:36 -!- gl00ten2 [~gl00ten@2001:8a0:7ee5:7800:46d9:f5c:17a2:432] has quit [Remote host closed the connection]
22:47 -!- gl00ten [~gl00ten@2001:8a0:7ee5:7800:46d9:f5c:17a2:432] has joined #hplusroadmap
22:52 -!- gl00ten [~gl00ten@2001:8a0:7ee5:7800:46d9:f5c:17a2:432] has quit [Remote host closed the connection]
22:54 -!- gl00ten [~gl00ten@2001:8a0:7ee5:7800:46d9:f5c:17a2:432] has joined #hplusroadmap
23:40 -!- flyback [~flyback@2601:540:c700:2380:e5ce:ed50:2863:67af] has joined #hplusroadmap
--- Log closed Wed Oct 01 00:00:55 2025