This current project is effectively a cache database for an earlier tool. The earlier project is where the main impact is - predicting the functional effect of a variant in the human genome.

For this updated project, they effectively ran the earlier tool for all possible variants in the genome (3 billion bases in the genome * 3 single nucleotide variants = 9 billion). It’s a very useful project for people like me, but the major advance was already done (admittedly by the DeepMind team).