Posts by Category
ClinVar holds 2.4 million variants of uncertain significance, 52% of its records. Missense variants are 84% uncertain, intron variants 26%, canonical splice sites 11%. Illumina's new SpliceAI2 aims at the splicing slice of that gap.
ClinicalTrials.gov shows trials listing KRAS G12D starting at 2 in 2023, 24 in 2024, and 20 so far in 2026, with 7 in Phase 3. The first RAS inhibitor for pancreatic cancer was approved on August 26.
FDA launched a Rare Disease Framework Initiative on October 14, saying only about 5% of rare conditions have an approved treatment. CDER's annual reports show orphan-designated drugs at 50% to 54% of novel approvals every year since 2021.
The 2026 Nobel Prize in Chemistry went to Henri Kagan and Kenso Soai for non-linear effects and autocatalysis in asymmetric synthesis. PubMed shows a field of about 2,000 enantioselective papers a year around an autocatalysis niche of fewer than ten.
FDA's list of novel drug approvals shows 46 in 2026 through October 9. That equals the whole of 2025 and is 39% more than the same date last year. What the pace does and does not tell us.
The 2026 Nobel Prize in Medicine went to optogenetics. PubMed shows the field grew from 2 papers in 2006 to 1,745 in 2025, while ClinicalTrials.gov lists nine interventional optogenetic trials, all aimed at the retina.
Regeneron says trevogrumab preserved almost 70% of thigh muscle on semaglutide. The whole-body measure in the same trial shows 20% to 43%. Why both are right, and what is still unknown.
Google's AlphaGenome Atlas predicts all 9 billion single-letter changes in the human genome. ClinVar holds 4.6 million records, and 51.6% of them are of uncertain significance. What the atlas fills in, and what it leaves open.
IQVIA counts $130 billion of biopharma M&A in the first half of 2026, 98% of all of 2025. With IPOs at 11 last year, acquirers have become the exit.
Mounjaro and Zepbound earned $14.9 billion of Eli Lilly's $23.0 billion second-quarter 2026 revenue. What that concentration says about the GLP-1 wave and the bioeconomy.
23andMe genotyped more than 14 million people. Its market value peaked near $6 billion in 2021. In 2025 a nonprofit bought it out of bankruptcy for $305 million, about $22 per person.
About 12% of US adults take a GLP-1 drug, against an estimated 2% in Europe. Yet Mounjaro now sells more outside the US than inside it. We pulled ten quarters from Eli Lilly's SEC filings.
Semaglutide went from 6,000 Wikipedia page views a year in 2016 to 2.7 million in 2024. CRISPR peaked in 2017 and has fallen 78%. A look at what the public reads about biotechnology.
On VariantBench the ten best AI configurations pass 40 to 50% of attempts. Their cost per correct answer ranges from $0.67 to $8.49, and the agent harness changes the bill as much as the model.
We compared nine indicators with their own best year. Trial starts and structure releases are at records. Venture funding is at 61% of its peak and biotech IPOs at 11%.
Sequencing a human genome fell from $95 million in 2001 to about $525 in 2022. GenBank now holds 8.2 trillion bases. We follow the cost curve and the archive to find where the bottleneck went.
FDA approves about the same number of novel drugs every year. The pipeline underneath is changing: bispecific trials are up nearly 16-fold since 2015, while mRNA trial starts peaked in 2021.
The Protein Data Bank holds 260,724 experimental structures. The AlphaFold database holds over 200 million predictions. Yet experimental releases hit a record in 2025. What that means for protein design.
Biotech IPOs on US exchanges fell from 104 in 2021 to 11 in 2025, below even the 2022 and 2023 trough. IPOs raised $3 billion in 2025, the lowest in ten years.
PitchBook counts $33.8 billion across 1,171 biopharma venture deals in 2025, against $55.7 billion across 1,039 in 2021. The average round fell from about $54 million to about $29 million.
A 2024 analysis found AI-discovered molecules succeed in 80 to 90% of Phase I trials but only about 40% of Phase II trials. The samples are small, and the gap says something about what AI has and has not solved.
Synthetic bacterial genomes grew about sevenfold from 2008 to 2019 and have stayed near 4 million bases. The race since then has been about rewriting the genetic code.
Casgevy won approval in late 2023 at $2.2 million per patient. ClinicalTrials.gov shows new gene-editing trials per year rising from 8 in 2016 to 34 in 2024, a steady climb that started before the approval.
The National Academies and the Bureau of Economic Analysis cite $959 billion and $403 billion for the US bioeconomy in 2016. McKinsey forecasts $2 to $4 trillion a year. Why the numbers disagree.
Hugging Face's open genomic model Carbon is said to match Evo 2 at far higher speed. Its own model card shows it ahead on three of six tasks and behind on three. The real story is how it reads DNA.
The two plots in almost every omics paper, explained with a labelled simulation: what each axis means, what to look for, and the three mistakes that mislead readers.
FDA approved 50 novel drugs in 2024. Fifty-two percent were for rare diseases and 48% were first in class. What the mix says about where drug development is heading.
