Christof Angermüller

Excited to announce our latest paper on ML-guided design of single-domain antibodies against CoV-2; A collaboration between @GoogleDeepMind, @AAlphaBio and @LumenBio.

Check it out: biorxiv.org/content/10.110…

After only 3 rounds of ML-based design and wet-lab testing, we were able to significantly improve the efficacy against multiple CoV2 strains, including Delta and Omicron. Designed Abs are diverse with up to 15 mutations, providing a therapeutic hedge against future pathogens.

Excited to announce our latest paper on ML-guided design of single-domain antibodies against CoV-2; A collaboration between @GoogleDeepMind, @AAlphaBio and @LumenBio.

Nature review: Scientific discovery in the age of artificial intelligence go.nature.com/45fUazH

"Sensitive remote homology search by local alignment of small positional embeddings from protein language models" ESM2 embeddings used " as input to the highly optimized Foldseek, HMMER3, and HH-suite search algorithms" biorxiv.org/content/10.110…

Contextualizing protein representations using deep learning on protein networks and single-cell data t.ly/M1iY6

Self-play reinforcement learning guides protein engineering t.ly/RP3RY

The Patent and Literature Antibody Database (PLAbDab): an evolving reference set of functionally diverse, literature-annotated antibody sequences and structures bit.ly/3OpR1aG

Sequence-based drug design as a concept in computational drug design go.nature.com/3Dqd1fn

“Large language models are universal biomedical simulators” Another use of GPT-4 for biomedical sciences research biorxiv.org/content/10.110…

Machine learning optimization of candidate antibody yields highly diverse sub-nanomolar affinity antibody libraries go.nature.com/3CEaRby

Mol-Instructions: A Large-Scale Biomolecular Instruction Dataset for Large Language Models bit.ly/3NnnKvU

Contrastive learning in protein language space predicts interactions between drugs and protein targets bit.ly/3qP3L1w

PoET: A generative model of protein families as sequences-of-sequences bit.ly/3qVm2u8

ProteinChat: Towards Achieving ChatGPT-Like Functionalities on Protein 3D Structures bit.ly/43rXvLp

ChatGPT-powered Conversational Drug Editing Using Retrieval and Domain Feedback arxiv.org/abs/2305.18090

A Multimodal Protein Representation Framework for Quantifying Transferability Across Biochemical Downstream Tasks bit.ly/3OW2hMO

BiomedGPT, a unified biomedical generative pretrained transformer model for vision, language, and multimodal tasks. Achieves state-of-the-art performance across 5 distinct tasks with 20 public datasets spanning over 15 unique biomedical modalities. arxiv.org/abs/2305.17100

“Self-driving laboratories to autonomously navigate the protein fitness landscape” Four independent trajectories “quickly converged on thermostable enzymes that were at least 12 C more stable than the starting sequences” biorxiv.org/content/10.110…

A general model to predict small molecule substrates of enzymes based on machine and deep learning go.nature.com/3OueNCU