I’ve spent most of my career writing code. But in early 2025, I decided to build something different: a home biohacking lab for education. I wrote about the building process, and today I’m going to share a few key experiments I did with it. All of these experiments and more are available in my lab notebooks on Benchling and linked throughout with public share links, so you can see the raw notes behind the narrative.

Overall, my goal was to show that I could insert the DNA of my choice into a host, in this case E. coli, and verify that it was being expressed. To do that, I trained up by doing some labs from kits provided by Bento Bio and Odin Bio. Finally, I enlisted the help of a tutor from Caltech to use the lab to reach my goal with a Whey fragment called Casein GMP.

I’ll put in a summary that I am delighted I was able to do such powerful things in my garage, but as well, for those interested in biosecurity, it was quite interesting how little oversight there was in this process. I could essentially order the exact DNA I wanted to be synthesized as part of a round piece of DNA called a plasmid. Occasionally, some vendors required an official address, not my home address, to ship to. This was easily obtained via a bio incubation space there’s one about 10 minutes away that was itself delighted to hear about my home biohacking setup and happy to help for a small fee and let me rent a shipping address there. Occasionally, vendors required that order came from an incorporated business, but this was also easily obtained via Stripe Atlas.

With models like Anthropic Mythos on the horizon, it’s no better time to get interested in biology and biosecurity. I think with dual-use technologies such as this, we are potentially entering an amazing era of biology productivity, and at the same time we need to think carefully about the capabilities these models unlock. Most of the things here I did without a mentor, and those with a mentor I could have potentially done, but the debug cycle would have been much longer. At the time last year, I found LLMs heavily useful, but for many things such as pipetting technique or gel debugging or other things, having the tribal knowledge of a seasoned biologist was useful.

I even went back and forth on whether I should publish all my lab notebooks and this process; however, I concluded that in the end, powerful models such as Mythos likely have access to these data sets via digitized programs such as Benchling or other biologists using these systems to debug and sharing their data. Ultimately, I’m not a believer in security via obscurity, so I think it’s better to share publicly that these capabilities are well within reach of the home biohacker.

The Lab Setup

As a refresher, the workspace lives in my garage. Here’s the core equipment:

• BentoLab: an all-in-one portable lab with a PCR thermocycler, centrifuge, and gel electrophoresis unit
• Egg incubator: repurposed for bacterial culture incubation at 37°C (yes, really)
• Instant Pot: used as an autoclave for sterilizing agar media
• Sous vide and bath: Used as a temperature controlled bath
• Styrofoam box: Used to put ice in for temperature shocking
• UV light: for checking GFP fluorescence
• Standard consumables: pipettes, falcon tubes, microcentrifuge tubes, petri dishes, LB media, kanamycin

Most of this was sourced from The Odin kits, BentoLab’s Bioltechnology 101 kit, and general lab supply catalogs. The total setup cost was a fraction of what a university lab bench runs.

Experiment 1: The Athlete Gene: Extracting and Genotyping My Own DNA

The goal: Extract DNA from cheek swabs, run PCR to amplify the ACTN3 gene (the “athlete gene”), and visualize the results on a gel.

Why ACTN3? It’s one of the most-studied variants in sports genomics. The R577X polymorphism affects the production of α-actinin-3, a protein found exclusively in fast-twitch muscle fibers. People with two copies of the X variant (homozygous 577XX) produce no α-actinin-3 at all — and this genotype is significantly underrepresented among elite sprinters. It’s a clean, well-characterized target for a home lab experiment.

DNA Extraction

The extraction protocol was straightforward using the BentoLab’s built-in centrifuge (notebook):

• Salt water mouth rinse (30 seconds of vigorous cheek swishing)
• Transfer to a 1.5mL tube, centrifuge at 4,000g for 90 seconds to pellet the cheek cells
• Pour off the supernatant, resuspend the pellet
• Transfer 20μL aliquots to PCR tubes
• Heat in the thermocycler at 99°C for 10 minutes to lyse the cells
• Centrifuge again at 8,000g to separate the cell debris
• Transfer the supernatant (containing the DNA) to fresh tubes

I ran two samples — mine and my son Casey’s. We labeled everything meticulously because mixing up family DNA samples would be… well, it would make for an awkward blog post.

PCR & Gel Electrophoresis: The Learning Curve

This is where things got interesting. And by “interesting” I mean “I ran a lot of gels that looked terrible.”

Attempt 1: DNA ladders (notebook) — My first gel run was just testing different DNA ladder concentrations and volumes. I made a 10x dilution of the ladder (45μL DI water + 5μL ladder), cast the gel with DNA SafeStain, and ran it at 90V. There was an odd hole in the middle of the gel, and then the electrophoresis box started acting up. Welcome to experimental biology.

Attempt 2: With my tutor Kian (notebook) — Kian (a grad student who mentors me on lab technique) and I ran another gel at 70V for 30 minutes. Key lesson: the 10x dilution was way too dilute. We could barely see anything. Note for next time: dilute only 2x.

Attempt 3: The real run (notebook) — Full PCR with the ACTN3 program (92 minutes), followed by electrophoresis. This time we used SeeGreen tabs instead of DNA SafeStain. Result: “it sucked :(” — direct quote from my notebook. The SeeGreen tabs weren’t compatible with the BentoLab’s blue light transilluminator. Lesson learned: always verify your visualization chemistry matches your hardware.

Every failed gel was a debugging session: wrong concentration? Wrong voltage? Wrong stain? It’s the same systematic thinking as software debugging, just slower and more expensive per iteration. I definitely got some good practice with lab technique that would help later.

Experiment 2: CRISPR and GFP Yeast (Odin Bio Kits)

The goal: Use a CRISPR kit to edit yeast and a separate GFP kit to make fluorescent yeast — both from The Odin’s beginner kits.

These kits are designed as entry-level genetic engineering experiments. The CRISPR kit targets a gene in yeast that, when knocked out, allows the yeast to grow on a selective medium. The GFP kit introduces the gene for green fluorescent protein.

Results

Checked the plates on April 14, 2025 (notebook), and:

• CRISPR plate: Lots of colonies growing. The selective medium was doing its job — only edited yeast could survive.

• GFP plate: Under UV light… it was glowing. That unmistakable blue-green fluorescence.

There’s something genuinely thrilling about seeing fluorescent protein expression for the first time especially in your garage. And being able to CRISPR or do bacterial transformation in the garage was a big step to ensuring I’d reach my major experimental goals for the first round of the lab.

Experiment 3: Bacterial Transformation with Kanamycin Selection

The goal: Transform E. coli with a plasmid carrying kanamycin resistance, select for successful transformants on antibiotic plates, and confirm the transformation worked. This was the most involved multi-day experiment. The first time we attempted, we realized that my “competent E. Coli cells” (ones that would uptake DNA rings called plasmids) were no longer so. So we attempted again using some cells that Kian provided.

The Protocol

Day 1 (May 29) (notebook): Made LB broth (200mL with 5g LB powder), prepared kanamycin stock (50mg/mL, 1000x), and autoclaved everything in the Instant Pot. Weighing out 250mg of kanamycin powder with a modified transfer pipette as a tiny spoon is peak garage-lab energy.

Day 2 (June 11) (notebook): The actual transformation. Thawed competent cells on ice, added 2μL of plasmid DNA, 30-second heat shock at 107.5°F, back on ice, then added LB broth and incubated at 98.6°F on the 3D-printed orbital shaker at 120 RPM. After an hour of recovery, streaked 200μL onto kanamycin plates and put them in the incubator.

Days 3–4: The wait. Temperature management was the biggest challenge. The egg incubator runs hot during the day, so I was turning it off and on to maintain 37°C. Not exactly a precision environmental chamber.

Day 5 (June 14) (notebook): Kian’s plates grew. The transformation worked. Only bacteria that successfully took up the plasmid (and its kanamycin resistance gene) could grow on the selective plates. Colonies on the plate = confirmed transformation.

What I Learned

• You can do pretty powerful things, as an amateur, for low cost: A double edge sword, as I was learning but also reflecting about the security implications of all this.
• Mentorship accelerates everything, AI assistance wasn’t enough. Having Kian as a tutor was key to making quick progress, but even then biology debug cycles are slow. AI helped me in the learning curve, but it wasn’t enough.
• Keep a lab notebook, but digitize it day of. I kept my lab notebook by hand as I didn’t want to use the computer while handling biological products. But when I went to use it later, it was difficult to parse, share, or recall. I recommend digitizing the day of the experiment. I used Benchling, which is professional-grade (and free for academic use). And next time I’ll do it day of.

What’s Next

I’m getting back into the garage lab with Kian this month. Still trying to decide on which experiments to do, but I’m leaning toward one with CRISPR and plants. Have any great ideas? I’m also going to be working on some concepts to increase biosecurity in ways that might still be friendly for home biohackers like me, while being less friendly to bad actors.

Interested in biohacking? The Odin makes beginner-friendly kits, and BentoLab is an excellent portable lab for getting started. You can also check out my original blogpost on the lab equipment and setup.