AlphaGenome Variant Impact scores integrated into Ensembl VEP

AlphaGenome Variant Impact scores integrated into Ensembl VEP Following the recent launch of Google DeepMind’s AlphaGenome Atlas, the Ensembl Variant Effect Predictor (Ensembl VEP) has now integrated AlphaGenome Variant Impact (AVI) scores. This integration enables researchers to easily access AI-generated scores which estimate how likely genetic variants across the whole genome are to be deleterious. Identifying variants likely to be involved in disease from the millions observed when sequencing the genome of an individual remains a challenge. Extensive work on variants which change a single amino acid has resulted in a wealth of tools for evaluating the potential impact of missense variants. However, variants outside protein-coding regions are less well studied and therefore more difficult to assess. The new AVI scores help address this problem. The AlphaGenome AI model, developed by Google DeepMind, comprehensively predicts how single nucleotide variants in human DNA sequences impact a wide range of biological processes regulating genes. The new AVI score summarises this information into a single metric, which can be used to rank potential variant deleteriousness. AVI scores are particularly valuable as few tools currently estimate the impact of variants on gene regulation. Ensembl VEP is a tool developed at EMBL-EBI, which predicts the molecular effect of variants on gene transcripts and protein sequence, as well as regulatory regions. It reports reference data including gene and variant phenotype associations, and population allele frequencies to make it easier for scientists to prioritise and interpret genetic variants. Ensembl VEP has now been extended to integrate AVI scores, alongside other evidence, into its extensive variant annotation reports. The scores can be used in the command line package which is commonly used in high-throughput variant annotation, by downloading data from [description/link]. AVI scores can also be accessed via the Ensembl VEP web interface, which provides links to the Google DeepMind AlphaGenome Atlas website for details on the features influencing the score. Both interfaces support filtering on AVI score, and other attributes, for variant prioritisation. This facilitates easy incorporation into a range of human variant interpretation workflows. Edit