A year after "Liberation Day," Trump is back with a 100% tariff on pharmaceutical imports, which sounds dramatic until you read the fine print. Companies that cut ‘Most Favored Nation’ deals with the White House? Exempt. Everyone else? Good luck, I guess...

Nearly every major pharma player has already made their deal: lower some U.S. prices, promise to build domestically, maybe list a few products on the government's direct-to-patient TrumpRx platform. AbbVie, J&J, the usual giants. They're through the door.

Now look at who's left outside. Midsized biotechs, many with just one or two products on the market, who say they simply can't afford to negotiate on the same terms.

Ten of them formed the Midsized Biotech Alliance of America earlier this year to push back. They argue that these companies drive a disproportionate share of breakthroughs in cancer and rare disease, and they're being asked to absorb costs designed for companies fifty times their size.

The U.S. keeps saying it wants to be the global leader in biotech innovation. But innovation doesn't come from the companies that already have 40 drugs on the market. It comes from the ones running on fumes with one shot at a first approval.

If the tariff structure rewards scale over science, who exactly is it protecting?

For more details: Full Article

BullFrog AI: A massive weekly change. The company announced a commercial agreement with a "Top 5 Global Pharmaceutical Company" to use their AI to identify drug targets for Major Depressive Disorder (MDD). (Very mysterious…)

Insilico Medicine: After dropping 12% last week, the stock roared back after Eli Lilly signed a massive AI drug discovery pact. Insilico received $115 million upfront, with the rest tied to milestones.

Relay Therapeutics: On April 2, the company confirmed they will present initial clinical data for Zovegalisib at an upcoming world congress.

Week after week, some of these moves remind me almost like it’s crypto trading…

Select the right answer! (See explanation below and source)

Rubedo Life Sciences' senolytic skin cream cleared Phase 1 with no serious adverse events and early signs of efficacy across psoriasis, eczema, and skin aging. Psoriasis patients saw a 20% reduction in skin thickness after one month with a clear dose-response, while 25% of eczema patients had significant itch reduction versus none on placebo. Rubedo has selected the 1.0% dose and already received FDA clearance to start a Phase 1b/2a trial in actinic keratosis in the U.S. [Link]

Cerevance's Phase 3 trial for CVN424 in Parkinson's disease has finished enrolling all 330 participants, with top-line data expected late Q3 2026. The 12-week study is testing once-daily oral doses across 49 sites and includes an AI-powered assessment system that evaluates patients' speech, cognitive, and motor skills through audio-visual conversation. [Link]

Formation Bio is about to dose its first healthy volunteers with BLKR201, a new CNS-targeted drug, with the Phase 1 trial expected to launch in May. The 128-person study includes cerebrospinal fluid collection, suggesting the drug is designed to cross the blood-brain barrier for a neurological or neurodegenerative condition. [Link]

Mount Sinai's AI "co-scientist" for Alzheimer's research won a $1 million prize for turning months of data analysis into minutes. Biomni-AD integrates genomics, proteomics, CRISPR screens, and clinical datasets through a natural language interface, letting researchers go from question to executable analysis plan in plain English. The platform will be made freely available to researchers worldwide through the AD Data Initiative. [Link]

UVA researchers built an AI drug design suite that accounts for how proteins flex and shift shape during binding, a limitation most existing tools ignore. The toolset, called YuelDesign, uses diffusion models to simultaneously generate both the protein pocket structure and the small molecule that fits into it, letting both adapt to each other during the design process. Companion tools handle binding site identification and chemical bond validation, and all three are freely available to the research community. [Link]

CRO stocks have tumbled since Anthropic launched its AI agents in February, but analysts say the selloff overestimates how much AI can actually replace. TD Cowen estimates a fully AI-enabled clinical trial setup would deliver only 10%-15% cost savings, and no evidence has emerged of pharma companies actually cutting CRO spending. The same analysts project AI could shave 11 months off a late-stage trial timeline, suggesting CROs that invest in the technology stand to benefit rather than be displaced. [Link]

For centuries, traditional Chinese medicine practitioners knew that Coptis (a bitter root) treated dysentery, and that Scutellaria (a flowering herb) handled lung infections. What they couldn't tell you is why at a molecular level.

That gap between "it works" and "here's the mechanism" is what drug developers call the mechanistic black box, and it's one reason the pharmaceutical industry largely abandoned plant-based drug discovery decades ago.

A new review in Life (Yin et al., March 2026) argues that AI is finally cracking that box open. The authors propose a three-layer framework that uses natural language processing, graph neural networks, and knowledge graphs to translate traditional medicine concepts into testable molecular hypotheses.

Think of it as a Rosetta Stone: the TCM phrase "clearing heat and detoxifying" becomes, in computational terms, a set of specific pathway targets like NLRP3 inflammasome activation and bacterial biofilm inhibition.

The case studies make this concrete. A Danshen (red sage) derivative called DS-2025 was mapped from ancient text references to "heat toxicity" through an AI platform that predicted high-affinity binding to PBP2a, a protein that makes MRSA resistant to standard antibiotics.

Lab work confirmed that DS-2025 disrupts bacterial membranes and messes with the transcriptional regulators that help bacteria form protective biofilms. Dual mechanism, validated in vitro.

This matters because the antibiotic pipeline is in crisis. The WHO's most recent pipeline review identified just 27 antibiotics in clinical development targeting priority pathogens, and only six were classified as innovative.

Antimicrobial resistance is projected to cause 39 million deaths worldwide over the next 25 years. Meanwhile, nearly half of all FDA-approved small-molecule drugs trace their origins to natural products. The raw material is there. The problem has always been figuring out mechanisms fast enough to matter!

The review isn't the only indicator that AI plus natural products is gaining traction. Enveda Biosciences has raised $300 million in the past year to decode nature's chemistry using AI, with its first drug candidate already in Phase 1b clinical trials for atopic dermatitis.

MIT's Antibiotics-AI Project has moved beyond screening existing compounds to generating entirely new molecules, with two lead candidates showing promise against drug-resistant gonorrhea and MRSA.

Although this review shows us that AI can be a tool to uncover candidates that already exist in nature, unfortunately, at this stage, many of the AI predictions remain validated only computationally or in vitro, not in patients. The authors also acknowledge that false-positive rates in virtual screening can exceed 90%, and the knowledge databases they're building on remain incomplete.

Still, the underlying logic is hard to dismiss. Traditional medicine systems across cultures have been running a millennia-long clinical trial on plant compounds. The question was never whether these remedies contain useful molecules. It was whether we could decode their mechanisms fast enough to turn them into real drugs.

Sources: [Research Article]

Theobromine is a compound in cocoa and chocolate, acts as a mild stimulant on the central nervous system and can contribute to improved mood and alertness. [Source]