Merry Christmas and a Happy New Year, everyone! I hope you’ve had a wonderful year and wish you all an even better one ahead!

I'm thinking about organizing a conference about AI and non-AI approaches for protein design and computational structural biology in Oslo (Norway) next year. If this happens, would you be interested in attending or speaking there?

P.S. Attached photos are the reason why Oslo 🏞️🌄

My last paper with Shuguang Zhang is finally published!

We explored QTY (L → Q, I/V → T, F → Y) mutagenesis to make hydrophobic transmembrane proteins more water-soluble. It shows that simple methods, without thousands of GPUs, can be useful and fulfill their tasks.

www.nature.com/articles/s41...

Importantly, we evaluated the suitability of cutting-edge structure prediction tools for developability tasks and conducted molecular dynamics simulations to validate the accuracy of structural DP calculated on predicted data

Combining high-throughput antibody sequencing data with structural modeling revealed that the human-engineered antibodies were localized within the natural developability landscape, highlighting the relevance of natural antibodies in therapeutic mAbs design

We found that the native developability landscapes exhibited variations across species and antibody chain types. And, structure DPs were harder to predict than sequence DPs using non-exhaustive machine learning algorithms

For our study, we utilized a dataset comprising over two million natural and human-engineered antibodies, alongside their predicted structures and constructed a developability atlas encompassing 40 sequence and 46 structure computationally determined DPs

Our study explores the plasticity of DPs, including redundancy, sensitivity and predictability within the native antibody repertoire. We also investigate how human-engineered antibodies relate to the native developability landscape

Antibody developability reflects the feasibility for therapeutic development based on antibody sequence and structure physicochemical developability parameters (DPs). While challenging to optimize, the native antibody repertoire offers valuable developability insights

Our work on antibody developability is out! We present a cartography of the developability landscapes of native and human-engineered antibodies. This study was a team effort of Habib Bashour, @evasmorodina.bsky.social, Matteo Pariset, and @jahnzhong.bsky.social from @victorgreiff.bsky.social, UiO

Inspired by DALL-E3's protein illustrations, I played with Bing Image Creator. Not very scientific, but lots of fun! All attached images were generated with AI 💻