Is equivariance all you need? #machinelearning #proteins

Pretty wild that there have been at least 4-5 #antibody specific structure prediction tools this year, all based on #deeplearning

1. DeepAb/IgFold
2. ABLooper/ImmuneBuilder
3. Equifold
4. t-AbFold

I think this calls for a new antibody version of CASP that measures both computationally-relevant and experimentally relevant metrics!

https://twitter.com/ideasbyjin/status/1593977142748094465?s=20&t=CSHAGpTCypOC8DV9_hEZEg

Anyone at #PEGS this week?

Really excited for Jane's keynote at the upcoming #PEGS in Barcelona! Also I'll be giving a talk next Wednesday - catch you at the #machinelearning session for #antibody discovery!
https://www.linkedin.com/posts/alchemab-therapeutics-ltd_pegseurope-antibodies-drugdiscovery-activity-6995819047317602304-7OCx?utm_source=share&utm_medium=member_android

Research Topics of interest:
* Protein #structureprediction - #antibodies and TCRs

* #NLP and #languages in general. E.g., there are many Korean expressions that I wish I could say in English and vice-versa, and I wonder if LLMs will ever "know" the frustration! (Meta: thinking about this makes me 답답해)

* Few/zero shot learning, self supervision

* repertoire scale #NGS analyses

$ whoami

I'm a bioinformatician / ML researcher, and I've been the main architect of the AI stack at Alchemab since the early days. Our main focus is on discovering antibodies that contribute to resilience against disease, using ML to support this work.

Drop me a line if you're interested in what we do, or if you're in Cambridge (UK). We can grab a coffee! #introduction