Naphak Modhiran

Excited to share our recent work describing the broadly cross-reactive and protective flavivirus NS1 antibody in @DWatterson_ @ProfPaulYoung and Prof. George Gao and Yi Shi. 1/n science.sciencemag.org/content/371/65…

New funding for infectious disease clinical trials! 📣  We’re funding randomised controlled trials to optimise existing vaccines and therapeutics for infectious diseases. This research must be led by those closest to the challenge. The lead applicant must be based in a low- or middle-income country in Africa, South Asia, or South-East Asia. Trials should focus on one of the following disease areas: • mycobacterial infections (tuberculosis, leprosy or infections with non-tuberculous mycobacteria) • bacterial infections (sexually transmitted infections, lower respiratory tract infections or bloodstream infections) • invasive fungal infections • dengue, leishmaniasis or schistosomiasis If you have an innovative idea for a randomised controlled trial proposal, but need to build your team, develop your clinical trial protocol or generate pilot data, you can apply to our Infectious Disease Clinical Trial Development Award. Delivered in partnership with @NIHRresearch and @FCDOResearch Learn more about eligibility and how to apply ⤵️ wellcome.org/research-fundi…

Now online! High-resolution in situ structures of hantavirus glycoprotein tetramers dlvr.it/TRCff9

By isolating immune cells from pediatricians, researchers employ rational design to engineer cocktails of monoclonal antibodies that neutralize #RSV and #metapneumovirus in rodents without generating resistance. @ScienceTM scim.ag/3ZG2w2T

New study identifies a potential genomic signature linked to avian influenza viruses that can sustain mammalian transmission: reduced GC content. Key for pandemic risk assessment. #Influenza #OneHealth #Virology nature.com/articles/s4156…

The PRIDA Network is now accepting applications for its 2026 Medical Microbiology Postgraduate Diploma, a fully online programme for doctors & pathologists in LMICs. 🗓 Teaching: Mar–Oct 2026 ⏰ Apply by Feb 2 Learn more: rb.gy/3lh47x

Yet another reason to screen people over 65 years of age for type I IFN autoantibodies: Type I IFN autoantibodies underlie chikungunya live-attenuated vaccine encephalitis | PNAS pnas.org/doi/10.1073/pn…

Newest @adaptyvbio Competition results are out! The task was to design a protein to neutralize Nipah virus (PDB: 2VSM), let's see how the state of the art AI protein design tools faired. The Baker Lab's RFdiffusion/RFantibody remains a contender, with 4 of the top 10 using the structure-based design method along with rational design and other models including ESMFold and Sapiens. Nick Boyd and the team at Escalante's Mosaic de novo design model shows promise as well. Nick originally got the highest in silico result, and this seems to have correlated well with the wet lab validation as well. He got number 8 and 17 of the top hits! 𝘗𝘳𝘰𝘵𝘰𝘤𝘰𝘭𝘴 𝘰𝘷𝘦𝘳𝘷𝘪𝘦𝘸 BoltzGen has a lot of entries (n=163) but a low hit rate around 1.23 percent in this dataset. BindCraft has moderate volume (n=88) with a hit rate around 5.68 percent. RFdiffusion (n=42) shows the highest hit rate among the main buckets shown here around 21.4 percent ProteinMPNN (n=29) is close around 20.7 percent. Escalante's Mosaic (n=9) has the highest hit rate at 88.8 percent. I actually would not necessarily use these datapoints as limitations of the tools themselves, as the quality of the binders would meaningfully differ depending on the team submitting. 𝘔𝘦𝘵𝘳𝘪𝘤𝘴 Boltz IPSAE seems to reasonably predictive of binding, although primarily as a filter and not a predictor: Selecting top 1% by min_ipsae: 63.6% binders (7/11) Top 0.5%: 83.3% binders (5/6) Top 10%: 18.4% binders (19/103) Among already-binding designs, these scores don’t strongly correlate with KD (it's better at “binder vs not” than “best binder”). 𝘗𝘳𝘢𝘤𝘵𝘪𝘤𝘢𝘭 𝘵𝘢𝘬𝘦𝘢𝘸𝘢𝘺𝘴 Boltz2 min_ipsae can be a legitimate hard filter for evaluating designs. I would suspect this is true of AF2 min_ipsae as well. Want ~25 designs? pick very top tail (this dataset suggests you can get >30% hit rate). Want ~100 designs? top ~10% is ~2× baseline hit rate. Binding does not mean function. Only 19 designs have neutralization annotations, and while the full neutralizers trend tighter, epitope/geometry is the primary factor. Congrats to @julian_englert and the team at @adaptyvbio on their hard work organizing and running the experiments. Stay tuned for deeper analysis of the results in the coming days! I'm excited to see what new methods will be published alongside the competition, the top three protocols have not been released yet. ———— @tamarindbio is the leading provider of the cutting edge molecular design tools like AlphaFold, ProteinMPNN, BindCraft, Germinal, and most of the protocols applied to this Adaptyv challenge! Use them at large scale through our web interface or programmatic API.

Extremely interesting paper - HA subtypes have wildly different sequences, but share structure and function. Deep mutations scanning of H3, H5, H7: ~50% of HA sites have very diverged preferences for different amino acids 👉biorxiv.org/content/10.648…

First paper of the year out: Structural and functional characterization of the antigenicity of influenza A virus hemagglutinin subtype H15 cell.com/cell-reports/f…

VC Biotech Financing by therapeutic area - $9.15B raised in 2H25 vs $8.9B in 1H25 - In 1H25, there were 32 deals $100M+ vs 36 in 2H25 - 4Q notably picked up mirroring public market sentiment - ~2/3 of total $ invested were in clinical-stage companies Source: Evercore

New online! Whole-genome spatial transcriptomic imaging with RAEFISH dlvr.it/TQGLF3

I’m such a @aliceyting lab stan.

After 1000 days of war, #Sudan faces one of the world’s largest humanitarian crises. People’s health needs are immense. WHO is supporting hospitals, treatment centers for severely malnourished children, vaccination campaigns, and more. But to reach those most in need, we urgently require safe access and reliable funding. Above all, people in Sudan need peace.

Big news from Boltz today: we’re launching Boltz Lab, a new platform with new small-molecule + protein design agents, announcing Boltz PBC and a $28M seed round, and sharing a multi-year partnership with Pfizer. More below! 🚀

Second paper of the year out: Cross-binding antibodies capable of neutralising diverse hantaviruses are produced in response to Puumala virus infection thelancet.com/journals/ebiom…

Cool ribo paper here!

Engineering bispecific exosome activators of T cells to target immune checkpoint inhibitor-resistant metastatic melanoma go.nature.com/496Ins7

Molecular and biophysical requirements for B cell receptor dependent enhancement of dengue virus infection biorxiv.org/content/10.648… #biorxiv_immuno

Iterative Epitope Expansion Enables Structure-Guided de novoDesign of IgE-Binding Miniproteins Targeting the FcεRI Interface biorxiv.org/content/10.648… この研究は、構造が複雑で設計が難しい抗体エピトープに対して、de novo タンパク質バインダーを設計することを目指しています。 最初から大きな結合面を狙うのではなく、結合しやすい小さな領域から徐々に標的を広げていく「エピトープ拡張」という考え方を採用しています。 AIによる構造設計と人の判断、実験を繰り返し組み合わせることで、設計を段階的に洗練させていく戦略が取られています。 このアプローチは、従来の一括設計では難しかった複雑な結合界面に対して、生成モデルをうまく誘導する方法として重要だと考えられます。

Nice to see BoltzGen as most used method 🤗