src/compute_genomic_mapping.py is a heuristic tool to map variants from an older reference to a new one.
It is a honest, if futile, attempt at covering a frequent or semi-frequent variant definitions
from a wild zoology of edge cases in UCSC variant definitions,
and map them to GTEx variant definition.
It doesn't handle all possible cases but should handle most.
See the following GTEx variant:
chr variant_pos variant_id ref alt num_alt_per_site rs_id_dbSNP150_GRCh38p7
...
chr10 47225796 chr10_47225796_G_GTCC_b38 G GTCC 1 rs142868522
This variant explictly declares alleles G and GTCC (rs142868522 in b38)
but dbSNP declares alelles G and GTCCTC.
IN snp125_hg17, there is a variant that, after liftover, maps to GTEx definition: rs10631152
but it has been delisted. What to do then, in this cases?
Variant rs11348115 in UCSC https://www.ncbi.nlm.nih.gov/snp/rs11348115
is listed as -/T (delT) but is displayes as deletion of TTTTTTTTTTTTTTTT
Many snps have multiple entries in the UCSC file: see for example rs3007516.
Both rs9794226 and rs11813951 have the same position, and have the same alleles after swapping strands.