James Dewar @jamesmdewar.bsky.social · 25/09/2026I think Anthropic, like other AI companies, is doing wonderful things for society with its models. I’m incredibly grateful for the free access it provides scientists like me. I just wish it would let me use Claude to the best of my ability to contribute to society through my own research. 010
James Dewar @jamesmdewar.bsky.social · 25/09/2026None of this shows that Anthropic’s commercial interests motivate its biology restrictions. But the current setup is also exactly what you might expect if Anthropic wanted to preserve a head start in areas where Claude has unique strengths. 100
James Dewar @jamesmdewar.bsky.social · 25/09/2026... One way to reduce this conflict would be to publish clear eligibility rules and approval timelines for biological research, with ordinary access available by default to verified scientists at institutions with appropriate biosafety oversight and additional review reserved for high-risk work. 100
James Dewar @jamesmdewar.bsky.social · 25/09/2026🚨 The core conflict is that Anthropic supplies the platform, controls access to it, and now competes in the same discovery space. 🚨... 100
James Dewar @jamesmdewar.bsky.social · 25/09/2026... If Claude's capabilities extend to slightly different domains compared to other models then Anthropic has an incentive to explore those opportunities itself before others do. 100
James Dewar @jamesmdewar.bsky.social · 25/09/2026Restricting Claude might seem to offer no financial advantage when competitors can use other models with similar capabilities. However, Claude doesn’t need to perform better for restriction to make commercial sense; just differently. ... 110
James Dewar @jamesmdewar.bsky.social · 25/09/2026... Indeed, Anthropic’s CEO Dario Amodei has publicly indicated that Anthropic views AI-driven pharmaceutical discovery as highly commercially valuable: www.dwarkesh.com/p/dario-amod... 110
James Dewar @jamesmdewar.bsky.social · 25/09/2026This is troubling because Anthropic has a potential conflict of interest. Reverse transcriptases already power gene-editing tools, so Anthropic could benefit commercially from the same kind of research it restricts others from pursuing with Claude. ... 120
James Dewar @jamesmdewar.bsky.social · 25/09/2026...Yet using Claude with fewer biology restrictions requires separate approval through Anthropic’s Life Sciences Verification Program, without clear published approval thresholds or timelines. 110
James Dewar @jamesmdewar.bsky.social · 25/09/2026... NIH-funded genetic-engineering research already falls under stringent biosafety rules overseen by institutional biosafety committees. Extending faculty verification to confirm that oversight should provide a straightforward route to using Claude for ordinary biological research. ... 100
James Dewar @jamesmdewar.bsky.social · 25/09/2026Anthropic already verifies scientists quickly: my application for its scientist benefit needed little more than my faculty page and said most applications were approved within five minutes. ... 100
James Dewar @jamesmdewar.bsky.social · 25/09/2026At the same time, Claude Opus 5.5’s biology filter blocks requests from scientists for help finding those same enzymes in the same viruses. 110
James Dewar @jamesmdewar.bsky.social · 25/09/2026Yesterday, Anthropic announced that it had used Claude to discover a new reverse-transcriptase system in viruses that infect bacteria: www.anthropic.com/news/claude-... 100
James Dewar @jamesmdewar.bsky.social · 25/09/2026Anthropic is using Claude for biomedical discovery with commercial potential while restricting outside scientists from using Claude for biomedical research. This creates a potential conflict because the same company controls access to the tool and competes in the discovery space. Thread 👇 130
James Dewar @jamesmdewar.bsky.social · 11/08/2026Excited to share that my review on the role of topoisomerases during DNA replication is now accepted for publication in JMB! www.sciencedirect.com/science/arti... 011
James Dewar @jamesmdewar.bsky.social · 04/08/2026Excited to speak later this month at the ORISE week-long online symposium on AI Opportunities in STEM. It’s free, open to all, and designed for faculty and trainees. orisesymposium2026.vfairs.com Timing feels perfect: I’ve seen a real uptick in interest in this topic over the past few weeks. 000
James Dewar @jamesmdewar.bsky.social · 03/08/2026Excited to share my piece in @nature.com Careers on why scientists are already trained for AI, plus ten places to start. 🧑🔬🤖 www.nature.com/articles/d41... Hope it helps! 🙏 Incredibly grateful to editors Jack Leeming and @jperkel.bsky.social for the opportunity and collaboration.nature.comWant to get more from AI? Treat every prompt like an experimentTaking a scientific approach to artificial-intelligence queries makes every output a result to be checked, says James Dewar. Here are ten tips for doing it right. 021
James Dewar @jamesmdewar.bsky.social · 04/06/2026Excited to see this published and to be able to read the whole story! Exciting science and a tour de force. www.nature.com/articles/s41... Ragini also gave a phenomenal talk on this at GS-Connect (gs-connect.us).nature.comCentromeric footprints preserve telomere integrity in ALT cancers - NatureCentromeric DNA repeat insertions and CENP-A chromatin assembly are identified as genomic signatures that preserve telomere integrity in ALT cancer cells. 041
James Dewar @jamesmdewar.bsky.social · 28/05/2026Huge thanks to the Dewar lab members who did this work: Sara Conwell (first author), Khushi Patel, @svnnahweeks14.bsky.social , and Steven Dahmen, and to our collaborators Matthew Cranford, @dkcortez.bsky.social, David Long, and Bill Dunphy 🙏 010
James Dewar @jamesmdewar.bsky.social · 28/05/2026Overall, single and convergent collapsed forks can elicit distinct repair outcomes. Single-fork collapse resolves the broken end without restarting synthesis. Convergent collapse completes DNA synthesis through error-prone DSB repair. 110
James Dewar @jamesmdewar.bsky.social · 28/05/2026Similar outcomes occurred at SSBs generated by Cas9 nickase (H840A nCas9), indicating these findings extend to other SSB origins, including CRISPR-induced breaks relevant to genome editing. 110
James Dewar @jamesmdewar.bsky.social · 28/05/2026Convergent collapse produced precise deletions and templated insertions. We didn't detect these products following single-fork collapse, demonstrating distinct repair outcomes between single and convergent replication fork collapse. 100
James Dewar @jamesmdewar.bsky.social · 28/05/2026Surprisingly, repair at convergent collapsed forks proceeded through annealing-dependent end joining following resection, largely independent of RAD51. This differs from the expectation that homologous recombination dominates repair at collapsed forks, and in the discussion we suggest why. 110
James Dewar @jamesmdewar.bsky.social · 28/05/2026At convergent collapsed forks, fork convergence rapidly converted single-ended DSBs to double-ended DSBs. These deDSBs underwent efficient DSB repair that completed DNA synthesis, a distinct outcome from single-fork collapse. 110
James Dewar @jamesmdewar.bsky.social · 28/05/2026Leading, but not lagging, collapse caused "secondary collapse" via nuclease resection that disassembles the sister fork. This is very similar to the recently described "replication reset" in bacteria 🦠 (pubmed.ncbi.nlm.nih.gov/41297800/).pubmed.ncbi.nlm.nih.govChecking your browser - reCAPTCHA 110
James Dewar @jamesmdewar.bsky.social · 28/05/2026At single collapsed forks, leading or lagging collapse generated single-ended DSBs that RAD51 efficiently resolved into stable D-loops, but synthesis didn't detectably restart, consistent with work in bacteria by @winterhalterlab.bsky.social 🦠 (pubmed.ncbi.nlm.nih.gov/41298506/).pubmed.ncbi.nlm.nih.govChecking your browser - reCAPTCHA 120
James Dewar @jamesmdewar.bsky.social · 28/05/2026When a fork encounters an SSB, it converts to a double-strand break (DSB). We analyzed how DSBs are resolved and whether DNA synthesis is completed downstream. 100
James Dewar @jamesmdewar.bsky.social · 28/05/2026🧬 ⛓️💥 Our paper is now out in @natsmb.nature.com ! We examined the fate of replication forks that collapse at single-strand DNA breaks (SSBs) in vertebrates 🐸, comparing the consequences of collapse at single vs convergent forks. www.nature.com/articles/s41...nature.comDistinct repair outcomes from single and convergent replication fork collapse - Nature Structural & Molecular BiologyHere the authors show that collapse of single replication forks causes recombination, end fusions or degradation but does not restart DNA synthesis. In contrast, collapse of convergent forks completes... 3176
James Dewar @jamesmdewar.bsky.social · 08/05/2026Excited to share a preprint of my review: "How Topoisomerases Relax and Disentangle the Genome During DNA Replication" 🧬🪢🧵✂️🙏 papers.ssrn.com/sol3/papers....papers.ssrn.comHow Topoisomerases Relax and Disentangle the Genome During DNA ReplicationDNA replication generates topological stress that must be relaxed and molecular entanglements that must be resolved. Unwinding of the parental duplex creates po 142
Reposted by James DewarAlberto Ciccia @albertociccia.bsky.social · 02/05/2026Happy to share our new review in @annualreviews.bsky.social on abasic sites and their impact on DNA replication written by @angelotaglialatela.bsky.social. www.annualreviews.org/content/jour...annualreviews.orgEndogenous Sources of Abasic Sites and Implications for DNA Replication: Mechanisms of Fork Stalling and RecoveryApurinic/apyrimidinic (AP) sites, also known as abasic sites, are among the most frequent DNA lesions, arising spontaneously or as intermediates in base excision repair. Their structural impediment to... 13017
Reposted by James DewarAmerican Cancer Society @americancancersoc.bsky.social · 30/04/2026Congratulations to all of the new members of the National Academy of Sciences! We are proud to see ACS Professor Dr. Johannes Walter recognized among these new members for his important work on DNA replication and repair. #CanSky #OncoSky 🧪 121
Reposted by James DewarCharles Winterhalter @winterhalterlab.bsky.social · 29/04/2026What are the genetic and molecular determinants underpinning the repair of bidirectional replication fork collapse in bacteria? PhD position open in my lab in collaboration with the brilliant @heathmurraylab.bsky.social. Please share and contact me if interested: www.findaphd.com/phds/project...findaphd.comIdentifying bacterial DNA repair requirements enabling stable genetic inheritance during bidirectional replication fork collapse at Newcastle University on FindAPhD.comPhD Project - Identifying bacterial DNA repair requirements enabling stable genetic inheritance during bidirectional replication fork collapse at Newcastle University, listed on FindAPhD.com 01615
James Dewar @jamesmdewar.bsky.social · 29/04/2026Updated the genome stability feed. Should be comprehensive but please let me know if relevant topics are missed. 000
James Dewar @jamesmdewar.bsky.social · 28/04/2026Now widely sharing two AI preprints: 1. A practical guide to using LLMs in science and research: zenodo.org/records/1917... 2. A framework for integrating LLMs into education, with Madhvi Venkatesh: osf.io/preprints/ed... These are living documents, openly licensed for re-use. 010
James Dewar @jamesmdewar.bsky.social · 27/04/2026Excited to finally see read story from Marta Markiewicz-Potoczny and Eros Lazzerini Denchi! www.nature.com/articles/s41...nature.comNonsense-mediated mRNA decay safeguards telomeres in pluripotent stem cells - Nature Cell BiologyMarkiewicz-Potoczny et al. report that the nonsense-mediated mRNA decay pathway protects telomeres in the absence of core Shelterin component TRF2 by regulating the stability and abundance of TRF1 at ... 000
James Dewar @jamesmdewar.bsky.social · 11/03/2025Awesome resource and super proud of Dewar lab alum tatakavlashvili.bsky.social who co-led the work.tatakavlashvili.bsky.socialTata Kavlashvili (@tatakavlashvili.bsky.social)@DamonRunyon.org Timmerman Traverse Fellow w Agnel Sfeir at MSK • PhD in Dewar lab @vanderbilt.edu • from Tbilisi, Georgia 🇬🇪 basic scientist fascinated by mtDNA+🏔️⛷️🎭🗽 030
James Dewar @jamesmdewar.bsky.social · 10/03/2025"Resolution of collapsed forks is separate from completion of DNA synthesis" biorxiv.org/content/10.1... "Rescuing bacterial genome replication: essential functions to repair a double-strand break and restart DNA synthesis" biorxiv.org/content/10.1... 000
James Dewar @jamesmdewar.bsky.social · 10/03/2025Also @heathmurraylab.bsky.social were absolutely AMAZING colleagues as we discussed our complementary studies and how to put them out to the community. We are incredibly grateful to have them as colleagues in the field! March 10, 2025 at 6:22 PM 120
James Dewar @jamesmdewar.bsky.social · 10/03/2025These differences make intuitive sense because helicase reloading proteins in bacteria are absent from vertebrates. Conversely, converging forks are common in vertebrates but rare in bacteria. Overall it looks like the response to fork collapse is shaped by evolution. 100
James Dewar @jamesmdewar.bsky.social · 10/03/2025In their study it was because re-loading of the replicative helicase was needed to restart synthesis. In our case it was because fork convergence was needed to generate a double-ended DSB. 100
James Dewar @jamesmdewar.bsky.social · 10/03/2025We were really surprised that D-loops in and of themselves did not efficiently restart DNA synthesis following fork collapse in vertebrates. However, @heathmurraylab.bsky.social came to the same conclusion independently in bacteria! 110
James Dewar @jamesmdewar.bsky.social · 10/03/2025Overall, our results indicate that resolution of replication dependent DSBs is separate from completion of DNA synthesis while revealing flexibility in how PARP inhibitors can induce fork collapse. 100
James Dewar @jamesmdewar.bsky.social · 10/03/2025Similar events happen at simple SSBs (‘nicks’), CRISPR-Cas induced SSBs and abasic site SSBs. PARP inhibition can stabilize SSBs to promote fork collapse, but we find that this differs based on the composition of the underlying SSB. 100
James Dewar @jamesmdewar.bsky.social · 10/03/2025We also discovered that “secondary collapse” events can take place. These involve extensive exonuclease degradation of the single-ended DSB to disassemble the divergent fork. 100
James Dewar @jamesmdewar.bsky.social · 10/03/2025Fork convergence formed single-ended DSBs that were converted to double-ended DSBs and rapidly resolved by end joining. End joining was largely accurate and resulted in efficient completion of DNA synthesis. 100
James Dewar @jamesmdewar.bsky.social · 10/03/2025Leading or lagging collapse induced single-ended DSBs that were efficiently resolved by RAD51 to form D loops. However, restart of DNA synthesis was not detected. 100
James Dewar @jamesmdewar.bsky.social · 10/03/2025We analyzed the downstream consequences of fork collapse to understand how DNA synthesis is completed after a single-strand break (SSB) is converted to a double-strand break (DSB) by replication. 100
James Dewar @jamesmdewar.bsky.social · 10/03/2025🧬⛓️ Our latest preprint examines the fate of collapsed forks in vertebrates 🐸 and our colleagues in @heathmurraylab.bsky.social asked similar questions in bacteria 🦠 to reveal how evolution shapes the response to fork collapse (see🧵) www.biorxiv.org/content/10.1... www.biorxiv.org/content/10.1...biorxiv.orgResolution of collapsed forks is separate from completion of DNA synthesisReplication fork collapse at single-strand DNA breaks (SSBs) poses a serious threat to genome stability. Using Xenopus egg extracts, we show that a replication fork encountering an SSB on either the l... 1116
James Dewar @jamesmdewar.bsky.social · 13/01/2025In our latest preprint we analyzed how abasic (AP) sites affect DNA replication: - AP sites stall DNA synthesis ⛓️💥 - Leading APs stall replisomes but lagging APs don’t 🛑/✅ - TLS is crucial for bypass of both leading and lagging strand AP sites 🛠️🧬 www.biorxiv.org/content/10.1...biorxiv.orgLeading and lagging strand abasic sites differentially affect vertebrate replisome progression but involve analogous bypass mechanismsAbasic sites are one of the most frequent forms of DNA damage that interfere with DNA replication. However, abasic sites exhibit complex effects because they can be processed into other types of DNA d... 000