Abstract: Predicting the pathogenicity of variants across the proteome remains a barrier to the implementation of universal precision medicine. To address this, the Missense Tolerance Ratio (MTR) was developed as a novel approach to assess the deleteriousness of variants. Its 3-dimensional successor, MTR3D, was demonstrated powerful at discriminating pathogenic from benign variants. However, this was limited to the fraction of the human proteome which had been modelled experimentally or could be modelled using SWISS-MODEL software, which relies on the existence of a close homologue. By leveraging the recent improvements in protein folding predictive technology, we have expanded the scope and power of MTR3D. By mapping large datasets of human variation to AlphaFold2 models, we have calculated scores for 1,311 proteins which were entirely missed by the original version of MTR3D and improved its overall coverage of residues in the proteome by 40.33%. Not only did harnessing AlphaFold2 structures improve coverage, MTR3D-AF2 also has a significantly improved ability to distinguish clinically established pathogenic from benign variants. Regions of intolerance as determined using MTR3D-AF2 scores (MTR score of <0.5) were found to be enriched with pathogenic variants at a rate ~9 times greater than that of benign. This enrichment was only 5 times greater when using regions assessed by the original version of MTR3D. MTR3D-AF2 calculations performed using larger size windows proved to be even more discriminatory. Using a window size of 14A to calculate MTR3D-AF2 scores, pathogenic variants were found to be enriched in regions considered intolerant at a rate ~21 times greater than that of benign variants. MTR3D-AF2 scores are freely available at https://biosig.lab.uq.edu.au/mtr3daf2/