Behavior when no rsID or chromosome position is available?

[Fiwx]

Description of the bug

Some scores, such as PGS000026, have SNPs which lack rsIDs and positions. How does pgsc_calc handle these situations? Is an error thrown? (I don't believe so.) Is the score simply calculated without using the locations?

Example:

rsID	chr_name	chr_position	effect_allele	other_allele	effect_weight	locus_name	variant_description
	19		e2		-0.47	APOE	e2 allele of the APOE gene
	19		e4		1.03	APOE	e4 allele of the APOE gene
rs4266886	1	207685786	C	T	-0.09	CR1	
rs61822977	1	207796065	A	G	-0.08	CR1	
rs6733839	2	127892810	C	T	-0.15	BIN1	

Command used and terminal output

No response

Relevant files

No response

System information

No response

[Fiwx]

It seems like it simply skips the SNPs, and says "unmatched" in the log file. However, it does not set "exclude" to true. This is less important for scores which have a small % of SNPs matching most of the time, unless, like in this case, the missing SNPs have by far the largest effect size. A workaround would be to fix the scorefile and then use a custom scorefile.

[Fiwx]

If a SNP is skipped (e.g. to a problem in the genome file), is it also skipped in the reference genome score calculation?

[nebfield]

Scoring files submitted to the PGS Catalog are harmonised to include missing optional fields like genomic coordinates and other_allele

e.g. for PGS000026: https://ftp.ebi.ac.uk/pub/databases/spot/pgs/scores/PGS000026/ScoringFiles/Harmonized/

By default if you use --trait_efo, --pgp_id, or --pgs_id, we fetch the harmonised data from the PGS Catalog that matches your genome build.

Matching only ever uses genomic coordinates and allele information (rsids are never used).

[Fiwx]

Got it. For this example, I believe it will skip APOE e2 and e4 as there are no coordinates, and simply calculate the score without it.

[smlmbrt]

Got it. For this example, I believe it will skip APOE e2 and e4 as there are no coordinates, and simply calculate the score without it.

Correct, there are other ways to genotype APOE and enter it into your genotyping data. If you know which is the e2/e4 allele in your data you could customise the scoring file so that it will match the chrom:pos:ref:alt in your dataset and it should be included as a match.