Call for playtesters: Tiny Machines - a portal to quantitative biology

Image from the labs of S.-H. Roh and S. Wilkens

The world of a cell is a subtle, elegant, and chaotic place. Molecules are crammed closer together than attendees at a rock concert, computers self-replicate with ease, and millions of tiny machines cooperate to build a living city. Imagine yourself, shrunk to the nanoscale, exploring the cell. What might you see? 

We believe that a combination of spatial mental modeling, spaced repetition, and Fermi estimation problems can give motivated learners a useful and beautiful intuition for the cell. In our experience, this intuition is very fun to learn, but requires some activation energy to get past the basics. Therefore, we're making a short (1-2 hour) portal to give autodidacts a running start to the quantitative study of biology.

If you're interested in trying the study materials, email You should include a few sentences on why you'd like to try this, and quick notes on your prior experience learning biology or other scientific disciplines. We'll spend an hour teaching you the basics, in exchange for feedback on the experience.

  • Who: We're looking for motivated learners. You should be comfortable with basic mental math, but don't need the equivalent of an undergraduate math degree. We'd especially like to work with people who are proficient in a quantitative (non-biology) science or engineering field. 
  • When: We'll give you a 1-hour tutorial this week (10/19-10/26), and access to a resource page with our study material (videos, Anki decks, and written problems) in exchange for your feedback.
  • What we ask: You'll need an hour of focused, intellectually energized time (so, pick a time that would personally optimize for this).
  • Why: Biology thought experiments, to us, are more addictive than a video game and a source of deep joy and beauty. We can't help but want more people to experience this. It may also be useful to build this intuition, but we're probably biased in our assessment of the utility by our enjoyment of the process.

- Laura Deming + Joanne Peng

    Tips for newcomers to longevity biotech

    Most newcomers trying to learn ‘longevity biotech’ unintentionally take an ineffective approach. Here are some things I wish I'd focused more on when first entering the field, in addition to obsessively learning the fundamentals of aging biology. It's not a comprehensive list, just a few things I find some people skipping over when they start.

    1. Build mental models of how we manipulate human biology, on the molecular level.
    2. We use a diverse set of drugs, ranging from ~100 atom small molecules to entire cells, to manipulate human biology. Often, recent discoveries add new tools to our repertoire. Learning about new tools can help you envision newly possible drugs. Additionally, each tool has a unique set of challenges (manufacturing, delivery, specificity/toxicity). It's easier to think about your tools if you understand them well.  For example, if your mental model of an 'antibody' is a ~100K atom protein with multiple covalently bonded parts, which is actively recycled in circulation, you'll be able to anticipate certain challenges to using it that you'd be blind to if all you knew was that it was one of many tools you could select.

      Celine Halioua has some example modality questions here. It's important to know that we've recently approved many more modalities than ever before (gene therapies, nucleic acids, cells), so there are many new areas to explore, and technical challenges to overcome in building new platforms.

      Some parts of the human body are much easier to deliver drugs to than other parts. Some drugs are hard to get into a cell, without them degrading. Drugs have a half-life, and won't stay in circulation forever, and how frequently you dose them can affect what drugs are economical to develop. The fields of pharmacodynamics and pharmacokinetics are helpful places to learn what we know about these kinds of issues. I remember in my first year being shocked to learn how hard it was to get molecules above a certain size into the cell, that many companies had been started just focused on this issue, and that there were a large number of well-characterized targets which might be very effective to drug if we could solve this problem - there are many interesting challenges like this.

      Derek Lowe has some great textbook recommendations for PK/PD, as well as general histories of drug discovery. I'd recommend spending enough time to get an overview of the kinds of things covered by the subject, but not memorizing things like allometric scaling laws (you won't really remember them fluidly until you need to use them in practice). You should try to be very curious, and guess at questions these fields should have answered but haven't.

    3. Build financial models for drug portfolios and venture biotech portfolios to understand when you can predict a positive return on capital.
    4. Building a biotech company involves spending a lot of money before you make money, usually, and working with products that sometimes have a high expected failure rate. If you want to build a biotech, you'll need money, and the people who can give it to you will probably want to know how you'll deal with these facts. To deal with these things, people in the industry model success probabilities and use portfolios of risky assets to try to achieve an overall positive return. Additionally, many of the major companies today are started by venture funds, which have evolved standard internal playbooks. I'd spend some time (but not an insane amount) studying the business models of Third Rock, BridgeBio, Flagship, Arch, Atlas (all different). It's worth understanding what they've learned to do, although I wouldn't recommend copying their specific business models unless you can start with the same reputation and scale.

      You should try to recreate the portfolios of some funds or companies and build basic NPV models (which take into account the fact that money today is more valuable than money in the future) to try to understand their decisions. Read S-1s (available on the 'investor' tab of public company websites) - companies give you large amounts of specific internal information about how they made decisions, for free, because they have to. 

      Good texts on the history of biotech include one on of Genentech (Sally Smith Hughes has online transcripts of most of the interviews here - incredible resource), Arch Venture Partners, Vertex at the start and later. The 8th Day of Creation is wonderful. I thought Merchants of Immortality was a good description of the early days of the longevity field (I don't think it's a good playbook for the future, just historical context). Bob Nelson talks are more blunt than most. HBS used to have okay case studies on many firms (FlagshipMillennium - I particularly like this one). Blogs are a good resource. Nathaniel Brooks Horowitz has a good list here. Steve Holtzman has a wonderful and very unappreciated blogpost, with lots of industry insider back and forth, hereIn the PipelineLifeSciVC, and Plengen are great. I think Josh Elkington has some good analyses at Axial

      Remember - you won't get all the way just by reading the above. You have to think for yourself, model for yourself, and try to recreate some of the logic behind the decisions you're reading about from scratch. Otherwise, it won't really be helpful to have all that context.

    5. Don't take anyone's word for it that we can change human longevity with our current tools - prove or disprove this for yourself.
    6. Wanting it to be true doesn't mean it is. The laws of nature don't care what you want. 

      Things that convinced me: evolutionary modularity + observed cross-species genetic pathways that increase lifespancross-species very diverse lifespan differences. Also, the fact that lifespan is variable so either genetic or environmental factors control that and we can control both of those. But you should form your own model, and critique mine! 

      You can’t understand and reason about aging biology unless you understand actual biology and most people don’t have good models of that. I like things that are really quantitative (so, showing that bias here), but I think doing thought experiments from religiously will give you a better intuition for bio. You should Anki a bunch of a constants from that book and practice Fermi estimation problems for different properties of the cell. 

    One of the most important facts about the next decade in longevity biotech is we will probably run the first longevity trial (either directly measuring aging or a proxy indication) and if we do enough of these well one of the interventions will probably work. If you join the industry now, this will be a major factor in everything you do.

    Therefore, It's also very important to understand possible clinical strategies for first aging therapeutic approval. It's probably good to know about the TAME trial, and look for similar projects. Longevity biotech companies are working hard on many different approaches to this problem (I wish I could list some specifically here, but I don't want to select only a subset of the ones we've invested in - our portfolio is partially public here). BioAge and Cambrian aren't in our portfolio right now, but I think they're also doing good work, and thinking in a sophisticated way about the best first indications to target.

    Why are we so much bigger than atoms?

    Quick notes from Schrodinger's 'What is Life':

    We combine two paradigms:

    Quantum world: Operates on the nanoscale. Energy is quantized, which means continuous noise will not constantly disrupt processes or destabilize structures. Therefore, you can do useful things.

    Continuous world: Operates on the macro scale. Due to the law of large numbers, noise is scaled down by sqrt (number of atoms involved). Therefore, continuous noise is less of a problem and you don't need to rely on quantum effects.

    We multicellular organisms are able to operate on both of these scales. I'm not sure how to justify why this is important, but it seems quite useful! 

    Footnote: In the quantum world, the kBT unit of energy rules triumphant - in the nanoscale, many different types of energy become comparably large to that unit. It's on these energy quanta that the cell operates.

    I'm looking for a thought partner

    Many of my adult friends feel sad because they can't do science more. For most, it was a source of joy in college, but their current profession is related and orthogonal to it. You job requires you to know enough science to be dangerous, and keep up with the literature but original foundation thinking or exploring vastly new areas is not going to be part of your manager-oriented schedule. I relate to this, and have many areas of science I'd like to seriously explore over time.

    I'm interested in finding people who feel similarly, and want to start to personally address this. In short, I'm looking to find thought partners (and offering to support you in turn, in exploring new areas). Intellectual diversity is good, and I've met an enormous number of interesting people I'd never have otherwise encountered through cold emails to my inbox, so I'm posting this publicly in the hopes that a few similar souls might see it.

    Shared values are important - here's what I'd strive to bring to the table, and look for in a thought partner in turn:

    • Chain ideas together coherently
    • Researcher or a professional with academic interests (I'm the latter)
    • Consume papers avidly, and have taste on which you like
    • Intellectually humble enough not to make bold claims without caveats, but courageous enough to want to find general principles
    • You like to get your hands dirty when it comes to papers - your immediate instinct is to quickly try to run the program, call a friend in a lab to vet some info, or find an alternate way to visualize the equation. Not doing this feels a bit uncomfortable and dishonest, if you’re really interested in the question at hand
    • Interested in the history of science
    • You love explaining things to others - your eyes light up and you feel a rush of joy when able to explain something you understand deeply to an interested observer

    Specific interests I have:

    What is life? Does the answer to this involve entropy. (Jeremy England / finding signal in the noise from other writings on the connection between the two)

    Generally, exploring any part of physics deeply (particularly stat mech + information theory, which are also useful for the above).

    I'm am amateur, but the style here would be focused on finding new representations of physics concepts that we like and trying to make analogies between different fields. Getting the 'gist' of major ideas, but more technically oriented and focused on visual analogies than reading a popular science book.

    Understanding evolutionary dynamics (the mathematics of the dynamic equations and game theory concepts involved), and conceptually mapping the space. I have a strong personal interest in the impact of cooperation and group dynamics here (as in, I understand that group selection can be equivalent to any other explanation for altruism if you view it as book keeping, but still think there is something interesting there that has to do with communication between individuals in groups).

    My goal would be to chat once a week about intellectual topics of interest, and possibly work on some side projects together. I think starting with a low bar to build an intellectual bridge is generally a good idea.

    Email me if you're interested -

    Can someone build Roam x Mathematica?

    Can I ask for help/advice? 

    I really want a @Roam x @Mathematica tool for making explicit, data-filled knowledge graphs of new fields. What’s a good way to hack a pipeline until someone builds this?


    I often want to understand new fields. To do this, I need to both build an idea graph and compile evidence for nodes in the graph. 

    How I learn new fields:

    1. Build an idea graph (Roam, Google Sheets)
    2. Capture key papers + databases for nodes on the graph (local folders, Google Drive)
    3. Extract information from key papers + databases (Google Sheets, export to pandas or other programming)
    4. Graph extracted information (Prism, Matplotlib, Google Sheets charts, Jupyter or Mathematica notebook)


    It’s hard to link idea graph and primary evidence in a way that 

    1. Renders the primary source easily accessible
    2. Allows for fluid data manipulation
    3. Allows for easy data visualization
    4. (Wishlist) propagates uncertainty in one concept to dependent nodes (make Roam an actual PGM, or make PGMs more interpretable and editable)

    There a few reasons for this

    1. PDFs are a terrible, noisy filter for the raw data (see below)
    2. Solutions exist for 2+3 above (for example, Mathematica or Jupiter notebooks) but PDFs are such a terrible core source of information that it’s hard to easily add them to this stack
    3. Science is nuanced and any time you put things in spreadsheets it loses some important context. Papers capture some of this in an unstructured way. 

    My core issue with reason #3 is that you should be able to then add structure to your idea graph, and incorporate the primary evidence into the updated graph with notes on your uncertainties. This to me is the recursive cycle at the heart of this process. I think a more fluid tool would allow for many more layers of iteration here. 

    Appendix: Why I hate papers so much as a way to get scientific info

    How science works: 

    1. Collect data, put in spreadsheet (scientist)
    2. Make JPEG with spreadsheet  (scientist))
    3. Put JPEG in PDF (scientist))
    4. Extract JPEG from PDF (you)
    5. Extract data points from JPEG (you)
    6. Put data points in spreadsheet (you)

    The paper PDF is literally a noisy, compressed filter on the info you want