LDpred2: Character in chromosome column causes error

[deepchocolate]

Describe the bug
If using sumstats that contain contain character in the chromosome column this will cause an error as below. In this case due to the presence of "X".

Expected behavior
This is unclear.

Additional context
As of now, happens when users join up chromosome and position from HRC data to sumstat files generated by meta-analyses.

[espenhgn]

I would've assumed bigreadr assumes that the sumstats chr column should only be numeric, but couldn't find a clear statement that values in each column all must have the same type. Can bigreadr::fread2 read the file as on this page?

Anyway, perhaps this issue is better opened at https://github.com/privefl/bigsnpr/issues instead? There may be a sort of similar case with snp_match() reporte here: privefl/bigsnpr#120

[deepchocolate]

I would've assumed bigreadr assumes that the sumstats chr column should only be numeric, but couldn't find a clear statement that values in each column all must have the same type. Can bigreadr::fread2 read the file as on this page?

Anyway, perhaps this issue is better opened at https://github.com/privefl/bigsnpr/issues instead? There may be a sort of similar case with snp_match() reporte here: privefl/bigsnpr#120

Yeah, I think if the intended use is to only do numeric matching on chromosome the snp_match function should list these rows as excluded. But for our purposes It's probably faster just to filter these rows out

[espenhgn]

The vctrs::vec_in function may accept NA for missing values (https://vctrs.r-lib.org/reference/vec_match.html), so replacing 'X' by NA could also by an option.