| |
| Managing Editor's Note: There's a growing disconnect between what the headlines say, what the data says… and what the market is actually doing. |
| Our colleague, Larry Benedict, calls this sort of disconnect a "market divergence." And it tends to create lots of short-term mispricings across the market. |
| It can be tricky to navigate market conditions like these. But Larry has a strategy to take advantage of it. He's airing a special briefing on Thursday, August 7, at 2 p.m. ET, to break it all down. |
|
| The World's First AI Genome Editor |
| By Jeff Brown, Editor, The Bleeding Edge |
| |
It's one of the most profound tech convergences taking place right now… |
| And it continues to be deeply overlooked by the markets. |
| The convergence of advanced semiconductor technology, generative artificial intelligence (AI), and biotechnology has created a once-in-a-lifetime setup for a golden age in biotech. |
| We wouldn't know this convergence is happening, though, just by looking at the equally weighted SPDR S&P Biotech ETF (XBI). |
| |
| SPDR S&P Biotech ETF (XBI) 5-Year Chart | Source: Bloomberg |
| If we put our blinders on and only use the chart above as an indication of whether or not the biotech industry was doing well, we'd be missing the entire picture. |
| The chart suggests not much is happening at all, trading at almost half the index's 2021 high… as well as at the same level as 10 years ago – the 2015 peak. |
| |
| With technology advancing as quickly as it is in nearly every sector, it's hard to believe that any tech sector chart would look like this. |
| The biotech industry was dealt a horrible – and artificial – blow, caused by the unscientifically based policies implemented during the pandemic. |
| Compounding the problem was the multi-trillion-dollar fiscal deficits. This resulted in a highly inflationary environment and much higher interest rates, implemented in an effort to stop inflation spiraling out of control. |
| High-interest-rate environments tend to be bearish for small-capitalization stocks, specifically those that have high research and development spend and little in the way of revenues and, more importantly, free cash flows. |
| That's why the sector chart doesn't tell the whole story. |
| While politics and fiscal policy are deeply linked to the sector's bear market, it doesn't mean that the industry isn't screaming forward with development. |
| This is why reading research, following public and private investment trends in the sector, and understanding how the sector is evolving and repositioning itself is so important. |
A perfect example is an announcement last week from Profluent Bio, an early-stage private biotech company that has been around since 2022. |
| The company's latest development has implications for the entire biotech sector. |
| AI-Developed Genome Editor |
| Profluent published research on what it calls OpenCRISPR-1. |
| CRISPR is, of course, the genetic editing technology discovered back in 2012 and has since been commercialized with lifesaving genetic therapies, which I've written about extensively here at Brownstone Research going back to 2015. |
| OpenCRISPR-1 is the world's first human genome editor created by artificial intelligence. That's an important distinction. Rather than human-designed genome editors, Profluent has leveraged advanced semiconductor technology, generative AI, and biotechnology – specifically related to proteins and genetic editing – to build an optimized genome editor using AI. |
| It's an incredible undertaking, one that would have been impossible even five years ago. Without the confluence of these technologies and the rapidly declining cost of computational resources, even large, profitable companies would not have been able to do this, let alone a small, early-stage company like Profluent. |
| |
| OpenCRISPR-1 Architecture | Source: Profluent Bio |
| As shown above, Profluent developed a protein language model (PLM) that was trained on 500 million protein sequences. This model was then fine-tuned with a CRISPR-Cas PLM (CRISPR– associated protein language model), which used about 5.1 million CRISPR-related proteins for additional training, specifically related to CRISPR genetic editing. |
| It's worth noting here that Profluent was able to leverage Google DeepMind's AlphaFold2, a protein structure prediction model built by AI, to predict the protein structures for 2,000 randomly chosen proteins. |
| While it was a small percentage of Profluent's training data, it was able to leverage the work of Google DeepMind, as AlphaFold is also an open-source model. |
| To make OpenCRISPR-1 that much more useful, Profluent included CRISPR-associated proteins (Cas) from 45 different protein families. |
| |
| OpenCRISPR-1 Cas Protein Families | Source: Profluent Bio |
| One of the early misnomers in the industry was that a single protein or approach would have universal utility. That has never been the case. |
| I've long maintained that different proteins will be used for different genetic editing therapies. To simplify, genetic therapies will be optimized to use the best proteins and editing approaches to get the job done in the most effective, safest manner. |
| And this is where Profluent's work gets really interesting. |
| Recommended Links
Larry Benedict Is Unveiling a Whole New Way to Trade Volatility Larry Benedict made $274 million for his clients – even with market crashes. Now, he's revealing a new way he's trading through the strange disconnect gripping today's markets. A setup he calls "The Market Divergence." This isn't about guessing where the market's going. It's about responding in real time – and potentially getting in and out within 24 hours. Larry will share all the details this Thursday, August 7, at 2 p.m. ET – including the name of his favorite "divergent" stock to watch right now.
Click here to register with a single click…
(When you click the link, your email address will be added to the event’s guest list.)
Futurist Eric Fry Says It Will Be A "Summer of Surge" For These Three Stocks One company to replace Amazon… another to rival Tesla… and a third to upset Nvidia. These little-known stocks are poised to overtake the three reigning tech darlings in a move that could completely reorder the top dogs of the stock market. Eric Fry, Senior Macro-Investment Analyst of InvestorPlace, gives away names, tickers and full analysis in this first-ever free broadcast. Watch now…
| |
| Open Sourced, Free for Use |
| Because the OpenCRISPR-1 genome editing model was trained on such a diverse data set and a wide range of protein families, it outputs unique proteins optimized for precise applications that are hundreds of mutations away from any known natural protein. |
| Put another way, OpenCRISPR-1 is able to design custom proteins for use in genetic editing therapies that human scientists would have no way of designing themselves. |
| The precision and custom proteins for specific applications speak to the efficacy of a potential gene therapy. |
| As for safety, the OpenCRISPR-1 model provides for two critical elements: |
- OpenCRISPR-1 demonstrated significant improvements in the reduction of off-target edits. Off-target edits are unwanted or incorrect edits to a genome that can cause unwanted side effects or inefficacy of the genetic editing therapy.
- OpenCRISPR-1-designed proteins showed lower immunogenicity to those proteins that are known to nature. Immunogenicity speaks to the immune response by the human body to a genetic therapy. In the case of a genetic therapy, a lower immunogenicity is desirable, as it allows the genetic therapy to get its work done without being disturbed by the immune system. A high immunogenicity in this context would result in the immune system attacking the genetic editing therapy, thus reducing or eliminating any efficacy.
|
| So, what's the significance of these developments? |
| Well, Profluent has not only made OpenCRISPR-1 open source, but it's also open-sourced its CRISPR-Cas Atlas, which is "the most extensive dataset of CRISPR systems to date." And that means that any biotech company, public or private, as well as any academic institution, now has access to these tools for developing genetic-editing therapies. |
| It's a catalyst, an accelerant to technological development in genetic editing and biotech. And it's free for all to use. |
For those curious, you can find the AI model and documentation available on GitHub right here. |
| Growing the Size of the Pie – Everybody Wins |
| The incredible part is that Profluent is a small company, having only raised $44 million since its founding. |
| This confluence of technologies is not only accelerating technological development at a pace we've never seen before… It's allowing breakthroughs and industry-altering developments to be built. And not just by massive companies with billions of cash to invest, but also by small groups of intelligent, motivated individuals who will take advantage of the incredible resources available to them today. |
| You might be thinking, if Profluent is open-sourcing its technology, how is it going to be successful? |
| Well, Profluent has a number of its own proprietary foundational models for genetic editing and protein design that it will use for the development of novel genetic editing therapies. |
| It will be able to partner with other biotech companies in development deals, and/or license out its own assets and models to others in the industry. |
| It's a smart strategy to accelerate growth, creating a much larger opportunity. And it will ultimately reduce the amount of time it takes – from discovery to FDA approval – for genetic therapies designed to cure disease caused by genetic mutations. |
| Jeff |
| |
No comments:
Post a Comment