SIFT scores are calculated based on multiple sequence alignments of protein orthologues (SIFT Aligned Sequences). Scores differ between builds 37 and 38, because MSA (Multiple sequence Alignments) can be different depending on the genome used. MSA are accessible by clicking on the missense prediction tool button in the Variant Panel. PolyPhen-2 (in AlamutTM Visual Plus) does not use MSA, just the human protein sequence and the substitution information.