🐨

ask: $3.5k
6.30.2025

{Any amount helps me ship by June 30, 2025.}




aim

¶ --> (1) extend the HTML5/CSS of the Stewdio frame with javascript [stewdio.js] so that (2) outputs are batched + pushed for training. ..Estimated delivery = June 2025

Stewdio


S: banana -beta sheetT: blueberry -alpha helixU: lime -delta ??[EMOJI]I (lnq🧑🏿) randomly walk on a #threaded feedback loop to ctf🎮🪖 {&& 0b_¢} the Stewpot (linker from all 2,028 stews) ++ Stube^solve -jailbreaking my 🧠nervous filesystem [STU thread.kernel=$LNQ]--> ie. *\:==avatar?stewellis🦸🏿‍♂️.root.uid=0;[/exploit🔐]
🍌 🫐 🍋‍🟩
R O N A L D


¶Whatever monies you pledge here go toward this effort.[-Thank you.] When the task is completed, we'll have a tailored mobile browser-based/virtual 3d Rubik's Cube (3×3×3) which can be a demonstrable accelerator in drug design and macromolecular/peptide science for public benefit.[-You're welcome.] If I fail, you'll get your money debited back to you. It's a win-win./// short: the costs for a metered cloud compute exceed my current monthly gross from Adsense, and I would like to spend a pay cycle[g] speeding stuff up on the client-side so that revenue > expenses.


-ideation graphic of the Stewdio as a speedcube in a browser-

/// Keep in mind that these images are just ideations (aren't they cool, though?); the finished product doesn't look like that, per se. They exist to showcase how the Stube's user could theoretically interface with panels that discuss relevant computational chemistry. In practice, this is basically speedcubing.
¶Some people have asked, "Well, why don't you just use ai to train all the models, and then there's your answer?". --In a sense, that's what is being done industrially. One of the issues with ai is that it can only be trained on given datasets. As potent as it is, ai comes nowhere near the capacity of the human brain🧠 in terms of speed, scaling, or energy management. (psst! ai still hasn't solved a Rubik's Cube in less than 20 moves[n], which is the necessary threshold for finding optimal fold structures) Another question often asked is, "Why not just use off-the-shelf software[2][v] built specifically for said science?". --Mainly because those products are bloated, making the learning curve too wide for beginners (our target market), ultimately lowering our contribution knowledge base. The Cube, on the otherhand, is based on very simple math, so it is the better option./// trivia: the Rubik's Cube is the world's bestselling toy.



Solving the Cube requires a certain amount of ingenuity, regardless of the type of intelligence (articial vs. organic). 100M tokens (here token = second/data) worth of data should be enough for a human to creatively solve any scramble; I know this has to be true because the Rubik's Cube can be solved by a 3 year-old[c]. The cubing community already has an extensive library of algs and solves that I'm going to apply to peptides for a new training stack. Automation enables me to stube [ie. put into a virtual speedcube and completely solve by mapping folding products to their 1 of 27 respective cubic blocks/cells (stewc --> stube)] each molecule with relative quickness. Compiled results (grouped by #hashtag before being fed to a neural network) may then be submitted to their respective academic journals and/or labs. /// Notes: +I'm not a medical professional, but I anticipate plenty of follow-up questions to which I'll preface that any pharmaceutical or clinical discussion to be had should only come after solutions are posted. For now, this is an activity that I just do for myself as play therapy, with no further ambition. Even if strategies are adopted, residual scientific papers published, and a bunch of really smart people come to a consensus, it (as does any legally marketed medicine) will still need FDA approval.
+The 2024 Chemistry Nobel laureates won for their work on 'computational protein design'[h] (disclosure: certain images on this page were ai-generated using DeepMind). What I'm doing here with stubing is moreso related to the 'protein folding problem'; resolving the dynamics of Levinthal's paradox[l] exactly. Together, both halves constitute a breakthrough in drug discovery. Aside: this is part of my submission to the bounty for the imaging prize from the Michael J. Fox Foundation.




purpose

¶This website @uuelco.ME averaged $0.21 rpm (~$3,500/month) in Adsense from October 2024 - April 2025./// A rule-of-thumb is that the more traffic a site gets, the smaller (closer to $0.01 - 0.05) its rpm[a], leading to greater revenue. --> DISABLE/TURN OFF YOUR AD BLOCKER. (see how)
¶Those averages are fluxual and from tedious manual input alone. A lot (~x104/day) more tries can be generated at a faster clip with automation, leading to an increase in two vital ad metrics: traffic + clicks. Obviously, I could just script a Rubik's Cube simulator (something like this), change a few of the colors around to match my branding, and call it a day (here's sample code for that). But, biochem requires more robust paneling (for cluster analysis, notebooks, script creation, etc.) to run a full computation. I want to dedicate time to constructing the solution.

I gotta get the Stewdio transpiled/code refactored into a javascript (.js) canvas[d] so that functionality is automated. This step is crucial to increasing the maximal number of brute force attempts involved in online speedcubing (the chat output is intentially kept to a bare minimum of prompt = 1, but offsite, all results are archived to spreadsheets for ml #threading --congruent to a trainable ai model). The stewdio.js will require sandboxing in a cloud service (eg. GC, Azure, AWS) before deployment. --> what ships: code dump, executable


-ideation graphic (sample) of the Stewdio as an unsolved speedcube-


¶Engineering this (ported or from scratch) might take an estimated 3-14 days (70-300 man hours) - plus a similar amount of time for testing, depending on finances. Once live, self-sustaining ad revenue should come immediately afterwards. When this is ready, we'll do a livestream demo.


¶I'll then have enough of a personal energy budget (permuting peptides for me generally consumes an enormous amount of calories) moving forward to begin work on solving (I foremost care about the topologies, not so much the functions of) the following 9 molecules [each are related to Parkinson's/Huntington's/Alzheimer's disease(s)]:

-alpha-synuclein
¶As far as Parkinson's (PD) is concerned, aggregation of misfolded α-syn [PARK1] into Lewy bodies plays a role in neurotransmitter release and synaptic vesicle management. The misfolding causes lewy toxicity (ie. compromised DNA repair), resulting in cell (neuronal) death. If this spreads to other neurons, then you've got a progression of the disease (in the brain AND/OR the heart). Keep in mind that the brain really begins directly behind the retina (which is where clumps of the protein can intially be examined without the use of MRI). #ParkinsonsDiseasesucks

¶The isoform of this synuclein that I will be tackling has 140 amino acids.

-STING
¶"STimulator of INterferon Genes" (acronym) is a signaling protein for detecting cytoplasmic DNA (cytoDNA). Endogenous cytoDNA can contribute to sterile inflammation (ie. no pathogens) from hyperactivation, leading to cancer (dysregular cell growth) and/or neurogeneration. <-- real nasty ish!
amino acid count: 379

-PINK1
PTEN-induced kinase 1 is important for maintaining mitochondrial health and quality control, as it targets damage/degradation/depolarization to the organelle (which supplies cells with their energy source). Without PINK1 properly functioning (from, say, genetic mutations), it can't recruit enough parkin from the cytoplasm to mediate dysfunctional mitochondria. Mutations of the gene lead to the death of dopaminergic neurons, causing Parkinson's.
amino acid count: 581

-huntingtin (Htt)
¶Huntingtin is essential to human life, as it is instrumental in longterm memory storage, and the axonal flow mechanisms (ie. organelle transport throughout the cell). It's a huge protein to begin with (3,144 amino acids), but is abnormally so in terms of the Huntington's Disease (HD) where the htt gene's DNA CAG trinucleotide is repeated a 36th time (1 more than the usual 10-35), leading to a mutant protein that then becomes fragmented and misfolds (the keyword in pathologies). These misfolds interfere with normal cellular+motor functions (those available in early development) that prove critical to personality, cognition, and emotion.
-granulin (GRN or PGRN)
¶Primarily located within the cell's lysosome (the organelle responsible for catabolic degradation of macromolecules, namely proteins/enzymes and carbohydrates), progranulin has been implicated in both cell proliferation and cell inhibition.

In both PD and AD, low granulin levels have been shown to be pathological.
amino acid count: 593 (progranulin)

-SERPINF2
¶(Serpin Family F Member 2) The serpin proteins are protease ('worker bee' enzymes that break proteins down into even smaller peptides) inhibitors. SERPINF2's role is to inhibit plasmin, itself a enyzme for degrading (breadking down) fibrin (the main protein in blood clotting), thereby maintaining hemostasis.
amino acid count: 482

-GSTP1
¶(Glutathione S-transferase P1) An enzyme (meaning that it catalyzes) with a critical role in cellular detoxification at the hydrophilic level (phase II).

Single nucleotide polymorphisms lead to different (and varied) responses to exogenous compounds (eg. drugs), including glutathione metabolism. Applied to neuropathology, GSTP1 can be like an overprotective parent, prohibiting drugs from reacting with or binding to their intended target(s), thus chaperoning oxidative stress.
amino acid count: 209

-BIN1
¶Bridging Integrator 1 (aka Amphiphysin II) is tricky because of its various Jekyll-Hyde isoforms, which it assumes through splicing. It is involved primarily with muscle regulation (myo motility).

In at least one of its isoforms, BIN1 is known for being a tumor suppressor protein (helping prevent cells from growing and dividing too rapidly). That sounds good, but its other responsibility within cells is to kill them (programmed 'off-switch'). Now, couple that with neurons, and you could - depending on the isoform - get conditional myotonic atrophy, or neurofibrillary tangles which cause diseases like Alzheimer's.
amino acid count: 409-593 (isoform-dependent)

-siglec-3 (CD33)


¶The siglec (sialic acid-binding immunoglobulin-like lectin) family of transmembrane glycoproteins (the 's' could just as easily be for 'sugar') regulates (cf. inhibits) immune cell function, particularly those responsible for causing inflammation in the peripheral nervous system.

Pathologically, polymorphic CD33 expresion in brain immune cells is linked to Alzheimer's.
amino acid count: 363


-lnq👨🏿‍🦱



FAQ

--refunds?


¶If I do not succeed in accomplishing the task, then you're due to get your money right back. Otherwise, nope. Refunds are not allowed. Transactions are final. I can't have you disrupting development with a withdrawal of resources. The entry is fair and very low ($1+), so your risk is negligible. An electronic record is preferred over cash💵 for purposes of any would-be disputes. As an optional bonus, you can have your name put into the sourcecode at no extra charge, identifying you as a 'project sponsor'. Thank you for your continued support.

--crowdfunding


Crowdfunding is the mechanism for pooling monies online for a specific cause or pledge. For reference, the JOBS Act of 2012 (more specifically, Title III) made fundraising using websites/web portals lawful.

¶The web portal in use here = PayPal[h].


¶How this works: these are over-the-counter (otc) digital/electronic transactions; essentially an IOU -you pay me what I ask up front, and I deliver on my end in return. Your name, billing/shipping address, and email address are collected by the payment processor, and then added to a spreadsheet (JSON file). Simple. {It is standard practice to include your email address and mailing address with your purchase.}

--Google Adsense


--> DISABLE/TURN OFF YOUR AD BLOCKER. (see how)

Adsense is a program from Google that allows websites and apps to make money [pay metrics = impressions, clicks] from their >Web real estate by inserting some computer code/script into the HTML of the site. The entire process is automated by Google, and I have no control over its functionality besides adjusting a few parameters on my end. My Adsense Publisher ID = pub-5560663513404554. Even if you don't make a pledge, at least notify the site that you were/are a visitor by clicking on some ads. -Thanks☺️

¶Millions of websites use the service, and web realtors must apply, be accepted, and remain in good standing (advertisers want their ads shown only on friendly websites).

¶I use Adsense primarily for its spiders (web crawlers or bots) that give me important analytics data, such as: page impressions, clicks, etc..

DISABLE/TURN OFF YOUR AD BLOCKER. (see how) <--


1 comment: