The reference genome for variants in humsavar.txt is GRCh38 (hg38) and is updated every release. If you want the genomic coordinates for these variants you can get them via the proteins API (https://www.ebi.ac.uk/proteins/api/doc/#/variation) and setting the sourcetype filter to ‘uniprot,mixed’; but also please read our help pages about how to retrieve large datasets. We have the variants represented in genome annotation tracks (BED and bigBED formats) that you can retrieve from our FTP site: ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/genome_annotation_tracks/ and going to the human datasets. Note here as we are representing protein information at the genomic level we provide the location for the three nucleotides that make up the codon.
We currently do not supply GRCh37 (hg19) coordinates for the humsavar.txt variants; you will have to use the appropriate GENCODE set or ENSEMBL’s VEP set to use GRCh37 on the variants.
This humsavar.txt file contains dbSNP accessions. Checking two of them manually both led to entries for GRCh38 so the current reference genome, also called hg38. This should be what you are looking for. For genomic coordinates you should download an annotation (GTF) file for the genome, e.g. from GENCODE, and then filter for the genes you want. GTFs contain the coordinates. GENCODE has both annotations for hg19 and hg38, choose what you prefer.