Announcement
COMPASS preprint introduces an atlas of 376 million intergenic DNA sequences
A researcher on the project says the team fine-tuned Evo 2 to find relationships between sequences with little similarity.
TLDR
The Arc Institute announced COMPASS as an approach to mapping underexplored noncoding DNA. A researcher on the project says the team fine-tuned its Evo 2 genomic language model on bacterial and archaeal sequences between genes, then organized 376 million of them into an atlas. The researcher says the team used repeating genomic patterns to rediscover CRISPR, VIPR and TIGR arrays and nominate new array-associated candidates.
Combined views
26.5K
11 Sources, first seen ago
