Brandon Richard retweeted
As psychedelic treatments approach potential FDA approval, psychiatrists need to be ready for patients’ questions.
To help prepare them, Johns Hopkins, Yale, NYU, and Columbia developed a standardized psychedelic medicine curriculum for psychiatry trainees. The first multi-site evaluation of the course was just published in Academic Psychiatry.
After the course:
💡 Confidence counseling patients about clinical research and legal care nearly doubled
🔬 Understanding of the limitations of current evidence rose sharply
📈 Knowledge quiz scores improved
⚖️ Trainees gained a more balanced, critically informed view of the evidence, without inflated enthusiasm for the field
We’re continuing to develop educational and training programs for professionals across a range of fields, so stay tuned!
doi.org/10.1007/s40596-026-0…
$TEM Tempus presented new clinical validation data at ESMO 2026 for an AI-enabled pre-screening panel in advanced non-small cell lung cancer (NSCLC), developed through its ongoing collaboration with Pfizer.
As the number of targeted therapies for lung cancer grows, identifying the right biomarkers requires running multiple, time-consuming tests on limited biopsy tissue. In practice, tumor tissue often runs out before testing is finished, forcing patients to go through invasive repeat biopsies or delaying targeted treatment.
Tempus is addressing this diagnostic bottleneck by pulling molecular signals directly from standard hematoxylin and eosin (H&E) digital pathology slides that are already generated during routine care.
Key takeaways:
• Massive Real-World Validation: The multi-center analysis evaluated 60,459 advanced NSCLC cases across 56,969 patients using routine whole-slide pathology images.
• High Analytical Sensitivity: The AI panel achieved more than 80% sensitivity across all 12 evaluated genomic and IHC biomarkers, with sensitivity exceeding 90% for several specific targets.
• Intelligent Diagnostic Triage: Still in development, the panel isn't meant to replace established sequencing (NGS) or immunohistochemistry (IHC) assays. Instead, it acts as a triage tool to help clinicians prioritize which confirmatory tests to order first, preserving limited tissue samples and getting patients on targeted therapies faster.
• Pharma Alignment: Conducting the study with Pfizer highlights a clear shared incentive. Pharma companies want to remove testing hurdles so more eligible patients actually get identified for precision therapies.
• Expanding the Digital Pathology Suite: This builds directly on Tempus’s existing AI biomarker work, sitting alongside Paige Predict (which analyzes H&E slides to predict 123 biomarkers across multiple cancer types when tissue is insufficient for sequencing) and Tempus Preview (which provides early biomarker predictions about 24 hours after a sample arrives at the lab).
In short, Tempus is using its multimodal data library to turn a routine pathology slide into an automated triage tool, helping doctors conserve tissue, eliminate testing delays, and route patients toward personalized treatments much faster.
ALT Tempus Presents Data Demonstrating AI-Enabled Pre-Screening Can Help Identify Patients for Biomarker Testing in NSCLC October 19, 2026 Study presented at ESMO evaluated more than 60,000 advanced NSCLC cases and found greater than 80% sensitivity across all 12 biomarkers using routine H&E pathology images. $TEM
Study presented at #ESMO26, from our collaboration with @pfizer, evaluated more than 60,000 advanced #NSCLC cases and found greater than 80% sensitivity across all 12 biomarkers using routine H&E pathology images.
Read on: tempus.co/4jpwO4X
$PRME Regulation can make or break a thesis in bio. Between 2024-2026, the FDA has been overhauling how it evaluates genetic medicines and platform technologies like Prime editing, creating a potential tailwind for this type of tech.
Roughly 95% of rare diseases still have no approved treatment. Instead of forcing ultra-rare programs into slow, legacy clinical trial models designed for mass-market pills, the FDA has announced several policy changes to modernize the process:
• Plug-and-Play Platform Reuse (May 2024 & Summer 2026)
Once a delivery vehicle or manufacturing method (like a standardized LNP or AAV vector) is approved, sponsors no longer have to start from scratch. Under Section 506K and recent genome-editing guidance, companies can cross-reference prior safety, biodistribution, and manufacturing (CMC) data across different drug candidates using the same platform.
• Single-Arm Approvals for Ultra-Rare Diseases (Sept 2025)
For severe genetic conditions affecting less than 1,000 U.S. patients, randomized placebo trials are practically impossible. Building on previous policy, the FDA’s Rare Disease Evidence Principles (RDEP) now permit full marketing approval based on a single-arm trial, provided it’s backed by clear natural history data and confirmatory biomarker readouts.
• Mutation Bundling via Draft "Plausible Mechanism" (Feb 2026)
Gene editing therapies often target distinct, patient-specific mutations within the exact same gene. Under the Plausible Mechanism Framework, sponsors could bundle multiple mutation-specific editors into a single trial and BLA under a master protocol. In addition, new mutation variants could potentially be added post-approval without requiring full-scale clinical trials.
• Restored Non-Dilutive Capital & Division-Wide Alignment (2026)
Congress and the White House reauthorized the Rare Pediatric Disease Priority Review Voucher (PRV) program, restoring an asset historically worth about $100M+ in cash upon approval. Capping this off, the FDA’s recent rare disease framework whitepaper standardizes Bayesian stats and flexible CMC inspections across all CDER and CBER review divisions to eliminate inconsistent standards.
In short, for modular genetic medicines with hard biomarker readouts, the FDA is starting to dismantle the traditional 10-year development gauntlet in favor of a faster, capital-efficient, plug-and-play framework. While some of these shifts are still in a draft or non-binding guidance status, they're shaping up to be a regulatory sweet spot for therapies like prime editing that directly repair a root genetic mutation and deliver objective, binary biochemical proof.
$PRME The FDA's new Rare Disease Framework Initiative is well-timed for Prime, especially considering their $108.8M in cash as of Q2 (runway into 2027).
Roughly 30M Americans have a rare disease ($997B annual burden), yet only about 5% have an approved therapy. Historically, the FDA’s 18+ rare disease programs were applied unevenly across review divisions (for example, the FDA rejecting uniQure’s plan to file a single-arm BLA in Huntington’s and mandating a sham-surgery trial).
The new framework aims to fix this division-by-division inconsistency by calibrating evidence and manufacturing rules to what is realistically attainable in small populations. This initiative caps a 13-month wave of CBER and CDER reforms to make regulatory flexibility the default norm.
Here are 4 ways the FDA's 2025–2026 rare disease overhaul could directly impact Prime:
1) De-Risking PM359's 1H 2027 BLA
While seeking a CGD partner to focus capital on in vivo liver programs, Prime is advancing PM359 (granted RMAT in June) toward a 1H 2027 BLA on Phase 1/2 data.
Unlike Huntington's (which has about 40,000 U.S. patients), p47phox CGD affects fewer than 1,000 U.S. patients and has a clear biochemical readout in functional neutrophil activity (exceeding 20% restores immune function, and PM359 hit 69–83% in NEJM). This checks every box of the FDA's Rare Disease Evidence Principles (RDEP) criteria for a single-arm filing.
2) Removing CMC Bottlenecks & Unlocking a $100M+ PRV
Historically, ex vivo stem cell therapies have been delayed by manufacturing (CMC) hurdles rather than clinical efficacy. Pairing the framework's risk-based CMC rules with CBER's Jan 2026 policy (dropping the 3-PPQ-lot mandate and allowing concurrent batch release and post-approval spec tweaks) could sharply cut the time and cost to make PM359 BLA-ready.
With Rare Pediatric Disease status (recently extended by Congress through 2029), approval also unlocks a Priority Review Voucher (PRV) worth $100M–150M+ in non-dilutive cash.
3) Scaling the "Universal Liver LNP" Platform
PM577a (Wilson disease, H1069Q), its R778L follow-on, and PM647 (AATD, PiZ) share the same GalNAc Universal Liver LNP and Prime Editor mRNA, swapping only the guide RNAs (pegRNA/ngRNA). Two 2026 FDA policies turn this modularity into a major moat:
• Platform Data Reuse (June 2026 Guidance): Safety, biodistribution, and CMC data from PM577a (first patient dosed Oct 5) can be reused across programs (evidenced by PM647 clearing its IND on Sept 24, just two months after PM577a).
• Mutation Bundling (Feb 2026 Plausible Mechanism Framework): Editors targeting different mutations in the same gene (like Wilson's ATP7B variants or Cystic Fibrosis CFTR variants) can be tested under one master protocol, bundled into a single BLA, and expanded to rarer variants.
4) Fast-Tracking 2027 Liver Readouts via Biomarkers
Instead of waiting 5–10 years for clinical cirrhosis or emphysema outcomes, the framework supports accelerated registration using direct mechanistic biomarkers:
• PM577a (Wilson): Measuring how effectively the liver clears copper into bile (via 64Cu PET imaging), blood ceruloplasmin (copper-carrying protein) levels, 24-hr urinary copper, and chelator withdrawal.
• PM647 (AATD): Restoring normal M-AAT protein levels and reducing toxic Z-AAT protein accumulation.
In short, the FDA is systematically removing the biggest historical roadblocks in gene editing for rare diseases, giving Prime a faster, clearer, and more capital-efficient path to validate its platform and monetize its pipeline.
Brandon Richard retweeted
We will report financial results for the third quarter ended September 30, 2026, before U.S. markets open on Monday, November 2, 2026.
The live audio webcast will be accessible through the “Events” section of the Tempus Investor Relations website, as well as here: tempus.co/47USb73
$ABCL Thad Hughes no longer works at AbCellera but it's interesting to see the background and experience of someone who helped build their AI/ML capabilities. Formerly worked at Google and his work at AbCellera led to a patent.
Presentation from 2021: youtu.be/9v38lwvlsgc?is=eucf…
Brandon Richard retweeted
Visit our poster at The Menopause Society 2026 Annual Meeting. We will be presenting Phase 1/2 results on ABCL635, an investigational, potential best-in-class antibody drug for the non-hormonal treatment of vasomotor symptoms.
Read more about ABCL635 here:
abcellera.com/pipeline/abcl6…
$TWST New Twist interview, always enjoy hearing Patrick Finn talk. Here, he explains how Twist acts as the physical wet-lab engine for generative biology with applications ranging from AI-driven discovery to MRD. Key highlights:
• Silicon Scale: By writing DNA on silicon chips rather than legacy plastic plates, Twist writes roughly 32 Million oligonucleotides a day with capacity for about 3 million clonal genes a year. That parallel throughput is what makes rapid design-build-test-learn iterations viable.
• Closing the Manufacturability Gap: AI models spit out highly engineered designs with inverted repeats, homopolymers, and extreme GC content that traditional synthesis struggles with. Twist’s complex gene platform raised sequence acceptance from 96% to 99.5%, delivering difficult designs in 12 to 15 days at industrial scale.
• Pure Infrastructure Player: Twist acts as a dedicated wet-lab execution layer, not a model builder. They don't take customer screening data to train competing proprietary drug discovery models, protecting partner IP.
• Halving Liquid Biopsy Sequencing Costs: Detecting rare ctDNA signals in tumor-informed MRD (minimal residual disease) requires massive sequencing depth (often 50,000x). Twist’s uniform target enrichment panels prevent wasted reads on off-target regions, cutting total sequencing costs for diagnostic partners by up to 50%.
• Pooled Testing at Scale (The MSK Case Study): When Memorial Sloan Kettering submitted 100,000 AI-designed antibodies, Twist synthesized the entire library as a single pool on a chip, ran display selections, and sequenced the binders. Real biological data goes directly back to train the next iteration of the model.
• The Bottleneck is Moving Downstream: AI removed the design constraint, and silicon synthesis is removing the build constraint. The bottleneck is now sliding downstream toward functional characterization, complex multi-specific formats, and chemistry manufacturing and controls (CMC).
Hear it straight from the source! Twist President & COO @genesynthesizer joins Ross Katz on the Data in Biotech Podcast to discuss antibody and AI drug discovery, liquid biopsy, MRD & Complex Genes: m.youtube.com/watch?v=mLwu8T…
@TwistBioscience
$PRME The FDA's new Rare Disease Framework Initiative is well-timed for Prime, especially considering their $108.8M in cash as of Q2 (runway into 2027).
Roughly 30M Americans have a rare disease ($997B annual burden), yet only about 5% have an approved therapy. Historically, the FDA’s 18+ rare disease programs were applied unevenly across review divisions (for example, the FDA rejecting uniQure’s plan to file a single-arm BLA in Huntington’s and mandating a sham-surgery trial).
The new framework aims to fix this division-by-division inconsistency by calibrating evidence and manufacturing rules to what is realistically attainable in small populations. This initiative caps a 13-month wave of CBER and CDER reforms to make regulatory flexibility the default norm.
Here are 4 ways the FDA's 2025–2026 rare disease overhaul could directly impact Prime:
1) De-Risking PM359's 1H 2027 BLA
While seeking a CGD partner to focus capital on in vivo liver programs, Prime is advancing PM359 (granted RMAT in June) toward a 1H 2027 BLA on Phase 1/2 data.
Unlike Huntington's (which has about 40,000 U.S. patients), p47phox CGD affects fewer than 1,000 U.S. patients and has a clear biochemical readout in functional neutrophil activity (exceeding 20% restores immune function, and PM359 hit 69–83% in NEJM). This checks every box of the FDA's Rare Disease Evidence Principles (RDEP) criteria for a single-arm filing.
2) Removing CMC Bottlenecks & Unlocking a $100M+ PRV
Historically, ex vivo stem cell therapies have been delayed by manufacturing (CMC) hurdles rather than clinical efficacy. Pairing the framework's risk-based CMC rules with CBER's Jan 2026 policy (dropping the 3-PPQ-lot mandate and allowing concurrent batch release and post-approval spec tweaks) could sharply cut the time and cost to make PM359 BLA-ready.
With Rare Pediatric Disease status (recently extended by Congress through 2029), approval also unlocks a Priority Review Voucher (PRV) worth $100M–150M+ in non-dilutive cash.
3) Scaling the "Universal Liver LNP" Platform
PM577a (Wilson disease, H1069Q), its R778L follow-on, and PM647 (AATD, PiZ) share the same GalNAc Universal Liver LNP and Prime Editor mRNA, swapping only the guide RNAs (pegRNA/ngRNA). Two 2026 FDA policies turn this modularity into a major moat:
• Platform Data Reuse (June 2026 Guidance): Safety, biodistribution, and CMC data from PM577a (first patient dosed Oct 5) can be reused across programs (evidenced by PM647 clearing its IND on Sept 24, just two months after PM577a).
• Mutation Bundling (Feb 2026 Plausible Mechanism Framework): Editors targeting different mutations in the same gene (like Wilson's ATP7B variants or Cystic Fibrosis CFTR variants) can be tested under one master protocol, bundled into a single BLA, and expanded to rarer variants.
4) Fast-Tracking 2027 Liver Readouts via Biomarkers
Instead of waiting 5–10 years for clinical cirrhosis or emphysema outcomes, the framework supports accelerated registration using direct mechanistic biomarkers:
• PM577a (Wilson): Measuring how effectively the liver clears copper into bile (via 64Cu PET imaging), blood ceruloplasmin (copper-carrying protein) levels, 24-hr urinary copper, and chelator withdrawal.
• PM647 (AATD): Restoring normal M-AAT protein levels and reducing toxic Z-AAT protein accumulation.
In short, the FDA is systematically removing the biggest historical roadblocks in gene editing for rare diseases, giving Prime a faster, clearer, and more capital-efficient path to validate its platform and monetize its pipeline.
The FDA today launched the Rare Disease Framework Initiative to reimagine rare disease drug development.
The new framework will adapt aspects of the agency’s long-standing approach designed for larger diseases. It will adopt an optimized, fit-for-purpose framework for rare disease drug development that considers the nuances of a specific disease, patient needs, product type, and attainable evidence. fda.gov/news-events/press-an…
FDA will host a meeting on Oct. 28 at the White Oak campus to collect feedback on the Rare Disease Framework Initiative from patients, disease advocates, industry, and other stakeholders. fda.gov/news-events/fda-meet…
Brandon Richard retweeted
I'm a simple man: 70% to 80% of rare disease are genetic, that's the lions share right there. In the original 2019 $PRME editing paper the authors calculated that prime editing can, in principle, correct up to ~89% of the ~75,000 known pathogenic human genetic variants in ClinVar. (Haven't found a clean number for base editing but probably around 30% to 60%)... Ofc delivery is an unsolved problem and I'm not saying that they can cure any single one of them. What I'm saying is that $BEAM & $PRME are the only ones who have a ball to shoot! Only thing that limits them is cash.
This is a meaningful step to bring better coordination around FDA initiatives to make the process for developing treatments for rare diseases more efficient. Will help accelerate new policy development.
bloomberg.com/news/articles/…
Brandon Richard retweeted
Base editing and prime editing clinical trial landscape, including today’s GSD type I clinical readout from @BeamTx.
Brandon Richard retweeted
The FDA approved 45 novel drugs in the first nine months of 2026, versus 32 over the same stretch in 2025 (about 41% more). Those approvals expand options for diagnosis and treatment, including for people living with serious and rare diseases.
I expect this number to keep going up.
$TWST I find it odd that some are saying that there is no data bottleneck in AI x Bio, and others are saying it's already over...
Meanwhile, billions of dollars in investments (from Biohub, Meta, the DOE, the NIH, Google Deepmind, and others) were recently announced to generate the massive amounts of data needed to train biological AI models as part of the Virtual Biology Initiative.
Plus, Anthropic recently unveiled their wetlab in SF, launched a protein design competition, and seems to just be getting started with their push into bio.
On top of that, in a recent interview, Max Jaderberg, president of Isomorphic Labs (recently rumored to be raising a new round at $40B), said:
"In AI, if you don't feel limited by data or you don't feel limited by compute, you're doing something wrong. Especially in the space of biology in chemistry, you can't just go to the internet and download data here. There are some amazing historical data sets that have been created, but actually we believe that a lot of the data that's going to power this space needs to be generated specifically for the models."
Well connected companies with deep pockets run by smart people are commiting a lot of money toward generating massive amounts of data to train biological models in the coming years and even the government is getting involved.
It's real, and it's not over.
Despite the crazy price action, I continue to believe Twist is very well positioned for this theme, especially in light of biosecurity concerns, competition from China, and an increasing focus on domestic biomanufacturing.
$ABCL $TWST
The data bottleneck in AI for Biology is bigger than many people realize... This quote is eye opening:
"If you think about the scale of data that we need now, I think we've seen plenty of presentations that outline that, but I'll tell you an anecdote from us where almost exactly 15 months ago we would think, okay a project of about 500 [DNA] sequences to make and test and characterization assay, that was a big project for us.
That was a lot of data.
Now, in about the last six months, we've done over 200,000. And now routinely, pretty much every week, we see projects come in between the 5,000 to 10,000 mark.
That's just over the last year, and that's the scale that the market has really changed.
And that's been driven by the AI market, by antibody discovery tools that are finding more and more candidates."
— Twist CSO, Colby Souders
Twist and AbCellera have spent years generating proprietary, high-quality biological data that makes them uniquely positioned for the AI biotech boom.
Source: youtu.be/5tvY_9BReeE
Glad some people are addressing Ginkgo's finances. Seeing other accounts pushing it as an AI x Bio name, but most conveniently never mention the financials... Important to understand that many of those "bulls" are traders. Always do your own research.
$DNA
구조조정을 시작한 2024년 3분기부터 매출, 조정 EBITDA, Excess Space 비용을 정리한 표.
1) 매출: 2026년 1분기 Biosecurity 사업 분사 이후 매출은 2천만 달러로 전년 동기 대비 반토막 났음
- 그나마 위로 하자면, 2분기에 더 떨어지지는 않고 아주 약간 상승한 수준. 부디 바닥이길 희망
2) 부동산 임차료: @Ginkgo 는 부채는 없지만, 과도한 부동산 임차료가 있기 때문에 이를 잘 봐야 한다. 구조조정의 영향으로 연구시설을 합치면서 빈 공간이 늘어 이 비용이 분기 약 930만 달러에서 2026년 1분기 1,584만 달러까지 70% 가까이 올랐다가, 2분기부터는 조금씩 개선되는 추세
- 26년 2분기 기준으로 이게 조정 EBITDA의 거의 40%를 차지한다. 코로나 버블의 후유증이다;;; CEO를 탓하지 않을 수 없음;;
3) 2025년 2분기까지 매출이 유지되다가 비용 절감 효과로 반짝 조정 EBITDA가 - 3천만불 아래로 유지되었으나, 이후 매출 폭망하면서 적자폭이 더 커진 상황
4) 그래도 2026년 1분기 매출, 조정 EBITDA, 유휴 부동산 비용 모두 최악의 바닥 찍고 2분기부터는 조금씩 반등하는 모습을 보인다.
5) 이번 분기에도 3개 지표를 핵심 지표로 보고, 체크해야 함. 2026년 1,2분기가 바닥인 점을 확인하는 숫자가 나온다면 정말 좋겠다.
$TWST Xaira recently demonstrated how its proprietary AI model, X-Design Vega, generated a 60,000-member VHH library and discovered 7 binders and 6 functional inhibitors against a previously "undruggable" GPCR target.
Twist is mentioned in the acknowledgements of the Nature paper cited in Xaira's article:
"Thank you... Twist Biosciences for access to their 400-bp oligo synthesis, which was invaluable for the high-throughput VHH experiments."
Some members of David Baker's team developed the algorithms RFdiffusion and ProteinMPNN, and later became Xaira's core design team.
Twist manufactured the physical 36,000-sequence library (9,000 across 4 disease targets) via 500 bp multiplexed gene fragments (MGFs) that allowed Baker's team to prove the concept experimentally in the Nature study.
Xaira Therapeutics (which Baker co-founded) took this proven workflow, licensed the IP, hired the key researchers, and scaled the approach resulting in the library for XA-4.
XA-4 is a complex G-protein coupled receptor (GPCR) with no known functional epitope or structural clarity. Traditional naive VHH library selections (four rounds) and llama immunization campaigns both failed to yield functional binders.
Modeling conformational flexibility across candidate epitopes, X-Design Vega generated a 60,000-member VHH library. Panning followed by functional screening yielded 7 confirmed cell binders and 6 functional inhibitors.
For more context, designing VHH single-domain antibodies requires synthesizing DNA sequences that span the entire variable heavy-chain domain (about 120 to 140 amino acids, corresponding to 360 to 420+ base pairs).
Historically, high-throughput pooled oligonucleotide synthesis was constrained to less than 300 nucleotides. Building sequences beyond 300 bp required multi-fragment assembly, which is expensive, low-throughput, and prone to combinatorial mispairing and dropouts.
Twist’s MGFs eliminated multi-piece assembly by printing the entire VHH domain (frameworks and all three CDR loops) as a single, unbroken double-stranded fragment up to 500 bp.
In short, Twist’s silicon-based synthesis of long, pooled gene fragments was the essential physical manufacturing bridge that allowed Baker’s team (now part of Xaira's core design team) to test distinct full-length VHH sequences, enabling the structural and biochemical discoveries published in Nature (the research cited in Xaira's new article).
ALT Using Al and Twist's Multiplexed Gene Fragments in the Hunt for the "Holy Grail" in De Novo Antibody Design $TWST
ALT "Thank you...Twist Biosciences for access to their 400-bp oligo synthesis..."
We’ve been quiet about protein design at @Xaira_Thera . Here’s what we’ve been building:
Introducing X-Design, with results from two active drug programs that show what we think this field needs to deliver: speed to a useful lead, and access to targets that have resisted discovery.
→ 7 weeks from DNA synthesis to a preclinical lead on an oncology target, with the epitope and reagents already prepared.
→ A functional antibody antagonist against a challenging GPCR where conventional library screening AND llama immunization had failed.
The bar we set for our model goes well beyond binding.
The antibodies need to be human-like, functional, cross-reactive across species, and developable. Properties that have to come together in the same molecule for a program to move forward.
That’s what makes these results exciting to me: designs tested in the lab, meeting the requirements of real programs in our pipeline.
My bet is that the biggest impact of AI protein design will be expanding which diseases and targets we can pursue.
We want to choose targets because the biology is compelling and to build the ability to make the right molecules.
This is our first look at X-Design. Much more to come.
See our blog: xaira.com/news/de-novo-desig…
Brandon Richard retweeted
We ❤️ the marriage of AI x Bio & the flurry of new research it is producing.
Our new flyer explores this relationship including:
Design → Build → Test → Learn, Lab-in-the-Loop workflows, better data for better models, and AI-powered antibody discovery.
@TwistBioscience
