New AI Framework Quickly Spots Concerning COVID Variants in Genomic Data
-
Scientists developed an AI framework with a new clustering algorithm to identify concerning COVID-19 variants early from large genomic datasets.
-
The framework combines dimension reduction and explainable clustering to process sequences quickly compared to manual methods.
-
It analyzed 5.7 million sequences in 1-2 days on a laptop, allowing more researchers to spot variants with less resources.
-
The new method works alongside phylogenetics as an alert tool for major new variants without needing to generate phylogenies.
-
It breaks down sequences into "words", represents them as numbers, then groups similar sequences to spot lineages of interest.