Beshoy Armanios, PhD MPH

Coptic Orthodox | | PhD in Pharm/Tox @UConnPharmacy | MPH @UMassChan | RNA therapeutics | Bioethics & responsible innovation | Views my own

[Beshoy Armanios, PhD MPH @beshoyarmanios]

Actively exploring scientist / translational pharmacology / RNA therapeutics roles.

I’m an in vitro & in vivo pharmacologist and translational toxicologist focused on diabetes, MASLD/MASH, and RNA therapeutics, with 5+ years across CRO, big pharma, and academia (Cyprotex/Evotec, Boehringer Ingelheim, Novo Nordisk, UConn). I’ve led lncRNA and ASO/siRNA safety studies in liver MPS and diet‑induced MASLD/MASH models, co‑invented a patent on RNA‑based therapeutics in MASLD/MASH, and built mechanism‑aware ML platforms like OligoTox and a liver ncRNA target‑prioritization engine.

If you’re hiring in translational pharmacology, investigative toxicology, DMPK/ADME‑Tox, or RNA therapeutics, I’d appreciate connections or referrals.

🔗 LinkedIn: https://t.co/8zUXQTdNVT
🔗 GitHub: https://t.co/Bov4IYspYs

#RNAtherapeutics #Toxicology #DMPK #TranslationalPharmacology #DrugDiscovery

[Beshoy Armanios, PhD MPH @beshoyarmanios]

@tangming2005 faster, not better, is exactly it. the judgment part, spotting a batch effect, a swapped label, a result that's too clean, comes from years of watching things go wrong. AI helps most the people who already know what wrong looks like

[Beshoy Armanios, PhD MPH @beshoyarmanios]

@CellCellPress GPX4 has been the textbook ferroptosis target for years, but tool compounds like RSL3 never had the exposure to work in animals. a bioavailable one finally makes it testable. the real question is the window, since kidney and neurons depend on GPX4 too

[Beshoy Armanios, PhD MPH @beshoyarmanios]

@EricTopol @ScienceMagazine burning off glucose to gluconate on an implanted electrode is wild. the tox side is what i'd want to see next: any platinum leaching or peroxide buildup at the electrode over months, and what the extra gluconate load does to the kidney

[Beshoy Armanios, PhD MPH @beshoyarmanios]

@NEJM circular rna is a smart fit for a gland. no free ends for exonucleases to chew on, so it should outlast linear mrna. how long did the salivary flow gain hold after one dose?

[Beshoy Armanios, PhD MPH @beshoyarmanios]

major us banks are finding third-party ai agents operating inside customer accounts. most got in because customers handed over their own usernames and passwords. some agents are calling bank call centers or taking actions the user never asked for, and banks say they often can't tell an agent acting for the customer from one that's gone rogue.

morgan stanley is blocking third-party agents. jpmorgan, bofa and citi are restricting them. one exec called it horrifying. banks are now running "day after" drills so they can lock down or restore customer access fast if the worst happens.

this is a national security problem, not a customer service one. banks are critical infrastructure. an agent holding a customer's login can move money, and banks have no reliable way to check who built it or who's actually steering it.

the worst case isn't one rogue bot. it's one popular agent getting hijacked, through a hacked vendor or a poisoned instruction, while it holds logins for thousands of people across dozens of banks. that's mass transfers or mass lockouts in minutes, and it looks like normal customer activity until it's too late.

what i'd want to see:
- verified identity for any agent that touches a bank, like know-your-customer but for agents
- scoped tokens instead of shared passwords, read-only by default
- a human sign-off before any transfer or new payee
- treasury and cisa setting one standard so each bank isn't figuring this out alone

the lesson: handing an agent your password is handing it your identity. we did this with screen-scraping budgeting apps years ago and banks spent a decade moving to proper tokens. we shouldn't have to relearn it with something that can act on its own.

https://t.co/smiMaggGJr

[Beshoy Armanios, PhD MPH @beshoyarmanios]

quick version for anyone running php: the fixed releases are 8.5.12, 8.4.27, 8.3.36 and 8.2.35, out oct 21. php says updating is the only real fix, the config tweaks just lower risk. 7.4–8.1 only got source backports

[NSA Cyber @NSACyber]

Effective Immediately, all organizations using PHP for their systems are advised to be prepared for an update addressing two critical vulnerabilities. PHP is widely used for web applications. These vulnerabilities often provide a direct path to the server's file system or database. If an immediate update is not feasible, it is recommend that organizations take the system offline until it can be fully mitigated. NSA will follow up shortly with more specific guidance.

[Beshoy Armanios, PhD MPH @beshoyarmanios]

nsa telling orgs to patch two php flaws or pull systems offline, and the flaws were found by ai. bug discovery is getting faster than most teams can patch. hospitals and labs running old web stacks are the ones i'd worry about

[Beshoy Armanios, PhD MPH @beshoyarmanios]

@EricTopol @CellCellPress immune-related adverse events are still so hard to predict before dosing. if diet shifts Treg stability, that's a modifiable risk factor trials could actually record at baseline, next to the usual labs

[Jorge Bravo Abad @bravo_abad]

Describe a protein’s job. AI proposes its structure and sequence.

A scientist might want an enzyme that breaks down plastic without knowing what its molecular structure should look like. Many protein-design tools require the scientist to supply some of that structural information first.

Dai and colleagues developed Pinal, a generative AI model that proposes proteins from written descriptions of their intended function.

Its Transformer-based architecture first generates a compact description of a possible protein structure. A second model uses that description, together with the original request, to generate the amino acid sequence. Structure provides a useful bridge: proteins with quite different sequences can still adopt similar shapes.

One experimental test went beyond designing an isolated enzyme. The team requested an H-protein, which carries reaction intermediates between enzymes in a CO₂-fixing pathway. The best design supported 1.7 times the glycine production of the native E. coli protein in a laboratory-reconstituted pathway, without repeated rounds of experimental optimization.

Getting there still required selection: 50,000 generated candidates were narrowed to 20 for testing, and five were functional. The written request opens the search. Computational screening and experiments establish which designs actually work.

Dai et al., Nature (2026). Journal license. https://t.co/xhIqq69vee

[Beshoy Armanios, PhD MPH @beshoyarmanios]

@tangming2005 the random-label test is the one i wish more papers ran. shuffle the labels, rerun the whole pipeline including feature selection, and if you still get a great AUC, the signal was never real

[Beshoy Armanios, PhD MPH @beshoyarmanios]

@NatRevDrugDisc what makes this one interesting is that it seems to work two ways, knocking down viral RNA and surface antigen while also nudging the innate immune response. losing surface antigen is the bar nucleoside drugs alone rarely clear

[Beshoy Armanios, PhD MPH @beshoyarmanios]

First FDA-approved antisense oligonucleotide for chronic hep B. 19% functional cure vs <1% on standard of care. not a cure for everyone, but for a disease where "treatment" has meant lifelong pills, a finite 24-week course is a real shift

[Superpower @superpower]

Read more: https://t.co/oBLbNDbB5w

[Beshoy Armanios, PhD MPH @beshoyarmanios]

RNA structure as a surveillance signal is a clever idea. the real test is whether it holds up on messy clinical samples, not purified RNA

[RNA-Seq Blog @RNASeqBlog]

Researchers used Raman spectroscopy to distinguish 20 SARS-CoV-2 variants through RNA structural fingerprints, demonstrating a potentially faster approach that could complement RNA sequencing in viral surveillance. - @kmu1928 https://t.co/LbhS38b3xX

[Beshoy Armanios, PhD MPH @beshoyarmanios]

same model, different scaffold, +11.5 points. in pharma we'd call this a formulation effect, and we'd never let it on the label without saying so

[Parnee @Parnee0wl]

Same frozen model. Up to +11.5 points on an office-work benchmark.

Google didn't touch the weights. It had an AI editor rewrite the agent's harness, task by task.

Harness² in 6 lines 🧵

[Eric Topol @EricTopol]

GLP-1 peptides protect the gut from immune injury.
Essential role of endogenous GLP-1 family peptides to reduce inflammation and restore mucosal integrity in the experimental model @DanielJDrucker @CellRepMed
https://t.co/oE0Z1VQ98m

[Adaptyv Bio @adaptyvbio]

We're excited to share results from our latest wet lab campaign with @AnthropicAI and @citrini.

Claude designed novel proteins to neutralize scorpion toxins and we tested 211 of these protein designs in our automated wet lab. 17 of them showed binding to the toxins in our experiments, with all data now available on @proteinbase.

This is a great example of what we're building towards: A world where anyone can run drug discovery campaigns on our automated wet lab.

[Beshoy Armanios, PhD MPH @beshoyarmanios]

@theo_jala the scorpion-toxin campaign is a great showcase for this. the hard engineering problem I'd love to hear about is automated QC, meaning how the lab decides a run failed before anything downstream analyzes it

[Beshoy Armanios, PhD MPH @beshoyarmanios]

@EDohertyGlobal 11% lower odds is interesting, but people on GLP-1s before elective surgery differ in a lot of ways matching can't catch. an active-comparator design against another diabetes drug would tell us much more

[Beshoy Armanios, PhD MPH @beshoyarmanios]

@MolTherapy the tradeoff nobody can skip is innate immune sensing. self-amplifying RNA buys dose-sparing, but the dsRNA intermediates trigger the very sensors you're trying to avoid

[Beshoy Armanios, PhD MPH @beshoyarmanios]

@MolTherapy extrahepatic siRNA delivery is the question everyone in RNA is chasing. curious which tissues showed sustained knockdown and how durable it was compared with GalNAc in liver