AminoAnalytica: Revolutionizing Drug Development through AI in Biopharma

This is The Startup Breakdown, the only newsletter looking into the new Y Combinator companies that you need to know. By joining this growing community of hundreds of founders, investors, and other startup aficionados (think i spelled that right?), you're getting a firsthand look at the companies that may be the next Airbnb, Coinbase, or Reddit.

Table of Contents

• AminoAnalytics: Pioneering AI in Biopharma for Eff …

  • Problem

  • Solution

  • Metrics:

• Links of the week:

  • New Nvidia chips

  • Trump denies campaign calls

  • How to jailbreak ChatGPT

  • Starlink breaks another record

  • Google layoffs

• How am I doing?

AminoAnalytics: Pioneering AI in Biopharma for Efficient Drug Development

Problem

We’ve all had that one idea that seems great in our heads, but it doesn’t quite work out in practice.

For my mom (and her ankle), that was ice skating.

For Elon Musk, that was buying Twitter.

For me, it was betting on myself and trying to build an AI personal finance app during my senior year instead of getting a real job. Still recovering from that one.

Biopharma companies, particularly those researching protein-based drugs, encounter this scenario frequently, thinking of a brilliant formula for a treatment that looks great on paper, but trying to turn it into a product and manufacture it is like eating peanut butter with tartar sauce - you might be able to do it once, but good luck repeating it.

Solution

AminoAnalytics is building a solution for this, developing a platform that allows biopharma companies to replace the manual trial and error process of manufacturing feasibility testing with AI-tomation, a prime example of how AI in biopharma is revolutionizing the industry.

The team is bringing protein property prediction to your computer, saving millions of dollars, maximizing the likelihood of commercially viable drug production, and allowing more treatments to be rolled out quickly. This also frees up researchers to focus on developing drugs for a wider range of ailments, showcasing the potential of AI in biopharma.

Metrics:

Traction:

• Won Imperial College AI startup competition → YC

Market Size:

• TAM: $4.8 billion

• SAM: $3.36 billion

• SOM: $100.8 million

Competition:

• Insitro, Cradle, Atomwise

• Insilico Medicine, Recursion Pharmaceuticals, Cyclica

• PathAI

Team:

• Abhi Rajendran, CEO: materials engineering with the Mercedes Formula 1 Team, LLM’s to predict protein manufacturability at Imperial College London

• Matteo Peluso, CTO: Masters in Bioinformatics and PhD in Molecular Life Sciences at Zurich (dropped out to build cool stuff)

• Adam Wu, COO: materials science and engineering at Imperial College London

Risks:

• Data access: the team’s models can only be trained and refined with massive quantities of data, and biopharma companies will be hesitant to give up the fruits of millions of dollars of labor

• Efficacy of models: Even with all the data in the world, building accurate, dependable models is tough. Under- or over-fitting models can lead to unreliable results and churned customers

• Reputation-focused industry: healthcare isn’t exactly keen on the whole “innovation” thing, and even coming from trusted names (which AminoAnalytica is not yet), it will be tough to convince a lab to take a bet on a new tool or software

What I like:

• Data-heavy industry: While obtaining data won't be easy, there’s a massive supply of usable data, making it a feasible space to train a vertical LLM, which is a key aspect of leveraging AI in biopharma

• Experienced and connected to researchers: yes, the team is young and relatively inexperienced, but they did devote the last few years of their life to this very specific topic, and through their research, they have begun to build up a reputation and network in the research sphere, making it more likely that they’ll be able to access the data needed to build their product

• Industry size and potential: the total biopharma industry is >$400 billion, and it’s one of the spaces most likely to throw multi-million contracts around like they’re dollar bills

Opportunities:

• Data licensing partnerships: accessing data will be tough, and the altruistic “data drop” on their website isn’t likely to supply the quantity of information needed to build a world-changing prediction model; the team can determine a realistic price, charge 20% more than this for most customers, and approach labs and research centers about a 20% (or even lower) discount for x years if they will supply their data for training purposes

• Championing open-source medical: when researching the company, I came across a link to what appears to have been an open-source model on Hugging Face allowing users to play around with the AminoAnalytica LLM; given their mission, they could open-source this model (or at least a limited version of it) without disclosing training data and weights to build up a passionate user base that may contribute ideas and data to building out the product, further advancing AI in biopharma

• Data cleaning or analysis for labs: step number one for the company should be just obtaining as much data as possible, and one of the ways to potentially access this data will be through providing some sort of service for labs and researchers to clean their unstructured data for them, and in exchange, they can use this data to train their models

AminoAnalytics is at the forefront of using AI in biopharma to transform drug development. Their innovative platform offers significant benefits in terms of cost savings, efficiency, and the potential to bring more treatments to market quickly.

Links of the week:

How am I doing?

I love hearing from readers, and I’m always looking for feedback. How am I doing with The Startup Breakdown? Is there anything you’d like to see more or less of? Which aspects of the newsletter do you like most?

Cheers to another day,