Hacker News Reader

Comment by roughly | Hacker News Reader

If you're not in the bio field, you may not be aware just how routine genetic modification has become. For the larger companies, projects can involve thousands of different iterations on DNA sequences, all of which are created, put in a cell, grown, sequenced, and analyzed. It's absolutely routine at this point - it's not a science or even engineering challenge project it's an operations project.

The one that really freaked me out, though, was learning about organoids - lab-grown clusters of organ cells used to research the effects of genetic disorders. Liver, kidney, and heart organoids are fine, but these are also used to study brain disorders, in which brain organoids - clusters of neurons - are grown from the stem cells of affected individuals. These clusters of neurons have been known to start firing and even display synchronized patterns...

View on HN

I accidentally grew a heart organoid (of sorts) once. Dissected an embryonic chicken heart, cultured it to make "fibroblasts" (mesenchymal stem cells), forgot about them, left them in an incubator for week, and when i came back, the cells had proliferated and covered the dish in a sheet of cardiac muscle, which was beating irregularly.

My takeaway is that cells absolutely love doing stuff. I don't mean to downplay organoid work, it's a huge achievement, but ultimately what it's about is giving cells a chance to do what they're programmed to do.

A project I was involved in was using optical flow to quantify the beating of cardiomyocytes grown in culture. I've done plenty of tissue culture but watching sheets of cells actually beat at a reasonable heart rate was still pretty wild.

Oscillations are also sort of a "freebie": if Thing A can temporarily inhibit the activity of Thing B and vice versa, activity will start pinging back and forth.

It's certainly a neat step forward towards studying brain activity, but the idea that organoids are sitting there dreaming is a bit much.

They can play pong. As far as I'm concerned that's already scary and raises ethical questions.

how far away is this from a 'two guys with the same name fucking around in a garage' (Apple mythical origin story) sort of setting?

Like when can I be a mad scientist making irregular beating sheets of chicken heart muscle in my basement?

There are a bunch of biohacker spaces around, and they’ll often run classes - small scale lab work isn’t actually that restrictive, cost or resource wise.

Also, The Odin - https://www.the-odin.com/ - sells DIY kits and classes (and is just generally one of my favorite groups of weirdos)

You'd need:

fresh fertilized chicken eggs

lab gloves

an egg incubator

a dissecting microscope

tissue culture dishes

phosphate-buffered saline

a pipette pump

10 ml pipettes

two pairs of dissecting tweezers

70% ethanol solution

kimwipes

lab gloves

a steady hand

tissue culture medium (DMEM is fine)

fetal calf serum

penicillin-streptomycin mix

a micropipette

micropipette tips

a tissue culture incubator

a tank of carbon dioxide

ideally, a laminar flow cabinet, but reasonably careful technique and the pen/strep would probably see you through

So, when? Depends if you have Prime really.

The thing is, growing sheets of cells is not a big step towards doing anything interesting. You need to do chemical and genetic manipulation of the cells, and that's the hard part.

I think you could do that today? Incubating a cell culture is definitely possible. You can either purchase used lab equipment, or even build your own incubation chamber.

I don't think OP's chicken heart experiment requires genetic tinkering, but even if it did, my understanding is that CRISPR is surprisingly accessible to hobbyists. For example, https://www.the-odin.com/diy-crispr-kit/.

> My takeaway is that cells absolutely love doing stuff. I don't mean to downplay organoid work, it's a huge achievement, but ultimately what it's about is giving cells a chance to do what they're programmed to do.

What gets tricky is when you’re trying to program the cells to do something different, because they really, really want to do a whole lot of other stuff, and rarely does making your weird enzyme or chemical top that list.

I am a stick.

SKELETON EDIT

Is this a reference to The Way of Kings?

Yes! Well, Stormlight Archive anyway.

As an experimental neuroscientist that has recorded fairly extensively from brain organoids, I would advise that these are HIGHLY underdeveloped cultures. When patch clamping, I had to depolarize neurons in a human brain organoid to -25mV to trigger action potentials (normally -60mV is sufficient; healthy neurons resting membrane potential is -70mV) Despite imaging organoids for hours with light sheet—a functional imaging technique that allows for observation of nearly all neurons in the organoid simultaneously, I did not observe any spontaneous action potentials, only diffusive waves of calcium activity.

Of course, not all organoids are created equal, and the protocols around extracellular matrices are improving constantly, but for folks interested in systems neuroscience the organoid field is still too underdeveloped to ask interesting questions around functional activity.

As someone who recently left the field of experimental neuroscience, anecdotes like this hit at why media coverage of neuroscience is so sensationalized and reductive: it takes a lot of nuance and background to convey the impact of any given finding.

In a lot of cases being even slightly reductive of EXACTLY what was reported immediately destroys the meaning and context of a finding. It's a field that actively resists compression of information and is so heterogeneously interdisciplinary that having the correct context to interpret a finding even within the field is often difficult.

Why did you leave the field? The compensation?

That was certainly part of it. The structure of neuro academia meant that I would be stuck in a postdoc position for 3-10 years paying, likely, around the NIH minimum of around $55k. Which goes along with more responsibility and drastically less job security than a graduate student without any better chances of actually getting a faculty position.

Combine that with the fact that I didn't have the drive or disposition that the field selects for and I was out of there immediately after I got my degree for a more lucrative, more interesting (to me) job with a better work-life balance.

If I were to restart my career, this is likely the field I'd pursue. I read The Gene: An Intimate History by Siddhartha Mukherjee last year and just became enraptured in the possibilities.

I'm wondering if there's anything I can contribute to that field as someone with general programming capabilities.

I just finished that book, it was really good. I can't wait to read The Song of the Cell.

Except somehow the required education is much higher and the pay is much worse than software, right?

I’m personally surprised we don’t have 20x as many people working in biotech.

There's sort of a running joke with how the market treats biotech as well - a bio company could announce they'd cured death and get a 5% bump in their stock price, whereas a tech company says the word "blockchain" and doubles their share price. As a software engineer, you can get pay into the ballpark of what you'd get elsewhere, but you should probably just ignore the "equity" section of that offer letter.

i imagine a big reason for this is that we don't have a silicon valley of biotech pumped with VC money. and presumably a big reason for that is there's all kinds of red tape when it comes to health and humans. no such thing as "disrupting" if you're going to violate some intense laws and then get arrested for it or get your labs shut down. or even worse, the potential resulting violence from fundamentalists surrounding your campus, or stalking employees. scary stuff

TBF, San Diego and Boston are the centers for Biotech. The issues you raise are the correct ones.

Also Oyster Point in South San Francisco (north of Silicon Valley) is a massive biotech hub.

In some way hubs are even more important in biotech than in software because of the massive startup cost.

> no such thing as "disrupting" if you're going to violate some intense laws and then get arrested for it or get your labs shut down. or even worse, the potential resulting violence from fundamentalists surrounding your campus, or stalking employees. scary stuff

Or, you know, actually "disrupt" something, like endocrine system of half a million people, or the entire global economy.

For the past decade, "Uber for Biotech" was my go-to scary hypothetical to post in HN comments. I'm increasingly worried we may actually get one for real.

(And to be absolutely clear: I'm fine with the biotech part itself. I'm worried about the consequences of using it to frack markets, SV-style.)

Well, even getting past the human/medical side - one thing the computer/VC people miss coming into the field is Bio is both expensive and not amenable to the kinds of growth/pricing curves computers have been. It costs money to grow a cell strain in a tube, but to grow 100 cell strains, it costs… 100x as much. It takes 24 minutes for E.coli to reproduce, but in another couple years we’ll be able to get that down to… 24 minutes. There’s opportunities for improving the cost, reducing the spend on reagents (or finding better ones), reducing the number of strains required, that kind of thing, but the computer/internet/digital industry’s basically priced a Moore’s law curve in on everything, and that just doesn’t work in Bio. The capital costs are large and the time to iterate is really slow compared to digital.

Yup, Theranos attracted more than $700 million, was valued at $10 billion, and silently returned bogus test results for thousands of people. They weren't doing anything like altering people's DNA, immune systems, endocrine systems, etc.

Yes the barriers and regulations are (rightly) much higher than just business or consumer software, but I fully expect that the VC money will be, and in fact is already moving there. VC is very herdlike, and it'll only take a few really hot biotech hits to open the floodgates. I'd predict 5-10 years away, maximum.

The marginal and capital costs of biotech are way higher than tech. It is absolutely fascinating however.

In the research field (my old field) most things don't work!

"If it always worked, it would be business. Let's go the pub." -- me to grad student after 12h of imaging cells that took months to grow up :)

It’s because bioengineering does not yet result in huge capital outputs and productivity gains that other engineering fields enable abstractly.

Once that happens and the existing underemployed bio-scientists are printing money through wealth/value creation, you will get your 20x.

> I'm wondering if there's anything I can contribute to that field as someone with general programming capabilities.

Oh there Absolutely is!

In general, there are four really high impact areas for computer folks - automation of lab equipment allows for scaling the number of experiments and the data that can be gathered by orders of magnitude; data science and engineering to sort, catalog, and investigate that data; building and scaling scientific computing systems to run analysis repeatably at scale; and even standard application building to manage and improve the overall process of tracking experiments and coordinating between groups.

The scientists and bioinformatics folks are very good at what they do, but someone who really understands how to build, manage, and scale software systems is an enormous force multiplier. If you're interested in the field, I really strongly recommend it.

Glorified tech support for scientists? No thanks.

That’s really not what the job is at this point. At least where I’ve been, there’s an understanding of the strengths of what compute systems and software can offer for accelerating the work that’s done and opening new avenues for experimentation and methods development - it’s very much a collaboration, not a one-sided tasking.

That said, you don’t get to just go off and faff around with code - you do have users and they do know more about you about the problem space you’re working in, so you do have to spend some time to learn to understand what you’re building and the context you’re operating in, which is different from most other tech jobs. The failure case for most software/computer folks coming into bio is not recognizing the difference between biology and physics before building systems.

Why? Do you have a bad experience?

If I want to do data engineering for such labs, how should I aporoach or even find such opportunities? I heard they hire graduate students for that in general.

I agree, I think we're pretty early in harnessing bioengineering, compared to what we do internally literally without thinking about it.

how far along is yeast mutation?

I've fantasized about retiring to somewhere remote and starting a yeast radiation garden, to experiment with using those micro-machines to re-assemble matter for us

> These clusters of neurons have been known to start firing and even display synchronized patterns...

This doesn't surprise me at all. Dr. Thomas DeMarse was training rat brain slurries (think brains thrown into a blender) to fly flight simulators in 2004. I've long suspected brain cells tend to self organize and in particular want to minimize change in inputs over time. (and yes this is a gross over simplification given that I know of at least a half dozen types of neurons).

Neurons don't do a substantial amount of self-organizing without hormone gradients and other external factors to cause them to organize. It's unfortunately inherently difficult to generalize the function and organization of neurons since there are not only hundreds, or more, distinct classes, but each class has thousands of inputs and outputs to the local and distant network that almost all have an important impact on their function, and the generation of this network of inputs and outputs and what factors affect their operation, is highly dependent on the exact subclass of neuron in question.

Mapping neural outputs of some small network to some operation is easy since all you're concerned with is the output, not the actual operation and organization of the network. Scientists could get macaques to play Pong using electrodes (like what Musk demonstrated with Neuralink) since, iirc, the 80s. This is also the exact reason that neural nets tell us nothing about the actual function of the brain.

In the past 6 months, I decided to leave traditional B2B SaaS completely and move to this space. I realize there is still amazing money to be made in traditional tech, but the more I read, research, and network in SynthBio/BioTech/TechBio I know this will be the most important field in our lifetime. From a mission and a financial perspective, I've never been more convicted about a space than this.

I believe everyone that was distracted by easy money flowing into crypto and B2B SaaS the last couple of years is looking at generative AI (rightfully so!), but the real opportunity in in BioTech/TechBio. VC cycle likely to follow AI for the time-being, but I'm betting a full career pivot on BioTech. Please email me if you are exploring or excited about the space!

...not only that, they also developed something like precursors to eyes. As in light sensitive cells.

https://www.smithsonianmag.com/smart-news/mini-brains-grown-...

The phase 1 trials for the Moderna vaccine were ready within 3 months of the first case being recorded in Wuhan. And much of that time was still just bureaucracy and logistics. Sequencing of the viral genome was done in a matter of days and so was the conversion of the DNA sequence to an mRNA vaccine. Then it was months of safety and efficacy studies. If we can prove the long-term safety of the process in it's entirety, we could see vaccines be produced in days.

https://www.nih.gov/news-events/news-releases/nih-clinical-t...

To add: anything to do with optogenetics.

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

TLDR: Input the genes into a nerve cell and now you can fire the nerves with light pulses.

The even crazier thing about opto is that it's old hat at this point. You can teach an attentive undergraduate lab assistant to do it and a sizeable portion of labs working with any animal short of a primate did or does use it to address questions.

Care to share the 'new hat' methods?

Really depends on what kind of work you're doing. CRISPR, flexible high-density microelectrodes, scRNA-seq, etc. are all borderline magic compared to the average understanding of biology.

I just remember being awed by opto around a decade ago and now it's just a tool that labs have. Which isn't to say there's not cutting edge opto.

Reply on news.ycombinator.com