Most ambitious submission: RiNALMo: General-Purpose RNA Language Models Can Generalize Well on Structure Prediction Tasks

The GIS team (Tin Vlašić, Wan Yue, Mile Sikic) who developed RiNaLMo, clinched the Most Ambitious Submission out of the 144 submissions received at the Machine Learning for Life and Material Science workshop at the International Conference on Machine Learning (ICML) held from 24 to 27 July 2024, one of the top AI conferences in the world. Presented by Tin Vlasic at the ICML, RiNALMo was developed at GIS’ Laboratory for AI in Genomics, led by Prof Mile Šikić in collaboration with Dr Wan Yue and Dr Roland G. Huber (BII). It is currently the largest RNA language model that shows previously unseen generalisations on downstream tasks such as secondary structure prediction, which opens new opportunities in developing RNA-based drugs. RiNALMo has 650 million parameters pre-trained on 36 million non-coding RNA sequences from several available databases.

Read here.