Artificial intelligence Health, Wellness & Biotech Startups Venture

Nobody Has Cracked The Code On Drug Manufacturing. Why?

Illustration of lab worker looking through beeker

In conferences and biotech events, where startups present their elevator pitch and mingle with venture firms, you might hear industry players wistfully question why a certain type of startup hasn’t yet been pitched.

This magical startup combines logistics and pharmaceuticals to improve one of drugmaking’s biggest problems: manufacturing.

The process of scaling delicate petri dish-sized experiments into giant vats portioned out and sent to patients all over the globe is complex and expensive.

Search less. Close more.

Grow your revenue with all-in-one prospecting solutions powered by the leader in private-company data.

Pharma companies would throw money at the problem if they could; small changes in the manufacturing process can yield big savings in the long term, especially if that single process is going to be reproduced over and over again for as long as that drug is on the market.

While there are plenty of startups in the drug manufacturing world, very few (if any) have been able to approach the problem through designing efficient experiments.

Sara Choi, a Wing VC partner who invests in data-focused biotech care startups, has seen plenty of term sheets cross her desk since she started working for the firm in 2019. Not many have proposed an all-encompassing solution to the drug manufacturing issues that plague the industry.

“So far there hasn’t actually been a company to crack the design and experiment problem,” Choi said.

A high-stakes problem

In the early days of drug development, scientists are often hunched over a lab table, breath held, delicately separating cells and extracting proteins in small petri dishes. Under a microscope, in an excruciatingly controlled temperature and sanitary environment, these petri dishes hold what soon become new drugs.

When a bunch of them get poured into a giant, 1,000-liter vat, what was once a pure science problem meets with the cold, hard reality of logistics: Fragile concoctions that will suddenly need to endure reasonable temperature changes, long flights across the country and world, and being manhandled by doctors and nurses to treat their patients.

This is a complicated business.

It involves coming up with systems, ingredients and a recipe to make sure a drug can be made at scale as quickly, safely and cheaply as possible. The tiniest change, such as making a drug 10% more efficiently, can save billions of dollars. On the flip side, adding ingredients in the wrong order or in the wrong quantity can drastically affect a drug’s outcome and efficacy.

Experimenting in manufacturing — by conservative estimates — accounts for 16% of the total cost of developing a drug. Pharma companies are obsessive about searching for the right ingredients, creating a ripe market for this mythical startup.

“Understanding that if you change something here or change something there; what does that actually lead to in terms of the actual biomanufacturing?” said Choi. “That’s just a huge, huge, untapped area.”

Why can’t it be done?

A couple weeks ago I wrote about the data revolution spreading into biotech and startups leveraging data and artificial intelligence to make drug discovery faster and cheaper.

But making drug manufacturing more efficient appears to be an insolvable obstacle. As funding for data-oriented biotech startups that make health care more efficient has skyrocketed, why hasn’t a startup emerged with similar success in drug manufacturing?

This isn’t a pure science problem, which makes the process harder to solve. Any company to tackle design and experimentation won’t simply need biological data. They will also need data on the thousands of permutations of hundreds of ingredients in varying amounts, added at varying times. They then need to allow machine learning and AI to come up with best practices on what works for each drug.

The first step: Metabolomics

Although not a pure science problem, science may be the first step in solving this data-heavy manufacturing problem.

Metabolomics, a rather nascent field that maps out the molecules making up metabolic functions, could provide a wealth of data that manufacturing plants can use.

Advances in the data-intensive -omics technology, such as genomics and metabolomics, have allowed the pharmaceutical industry to leverage AI using said data. In drug discovery, AI is being used to sift through biology at a microscopic level, analyzing millions of permutations of different compounds to see what would make the best drug.

Who says the same can’t happen with drug manufacturing and metabolomics? Drugs are metabolized in the body, after all. For instance, a manufacturer may find that by using half of a certain ingredient, a drug metabolizes in the body and performs 40% more efficiently than before.

“We’re at this point where we have the computational data science tools to be able to make sense of metabolomics data and use it to drive decision-making and real improvements in drug development and production,” said Jack Geremia, co-founder of Matterworks.

Matterworks is a Boston-based biotech startup that combines metabolomics with AI to simulate the metabolic environment in which a drug may operate. After gaining information on how the environment responds, companies can make tweaks to the drug candidate, adding or removing contents and seeing how it performs in the body.

As it stands, metabolomics has seen very little action compared to its cousin genomics. Since 2000, there have been fewer than 40 startup funding deals in the sector, according to Crunchbase data. In 22 years, metabolomics startups have seen only $224 million to develop tools that would aid in disease diagnostics and drug development (which often go hand in hand).

While metabolomics data will be immensely useful in this field, drug manufacturing won’t make the strides it needs to without marrying biologics, chemistry, computational engineering and logistics.

“This is an exciting time in biotechnology because of the availability of massive amounts of data,” Geremia said. “At the same time, the field of data science to be able to handle and make sense and interpret that data is all coming together right now.”

Illustration: Dom Guzman

Stay up to date with recent funding rounds, acquisitions, and more with the Crunchbase Daily.

Copy link