--score ignore-dup-ids parameter seems to fail

[Anton]

Hi all,

I am trying to apply the --score function to my file containing effect sizes. The file has some duplicate SNP IDs, which is why I want to pass the ignore-dup-ids parameter.

However, when I do this I still get the following error:

Error: Variant ID '10:2365978' appears multiple times in --score file

I am using the PLINK v2.00a2.3LM 64-bit Intel (24 Jan 2020) build on Linux.

Here's the command I'm using:

plink2 \

--bgen mybgenfile.bgen ref-first \

--sample mysamplefile.sample \

--score weights.txt \

1 \

2 \

3 \

header \

ignore-dup-ids

Here's my log:

PLINK v2.00a2.3LM 64-bit Intel (24 Jan 2020)
Options in effect:
  --bgen mybgenfile.bgen ref-first
  --sample mysamplefile.sample
  --score weights.txt 1 2 3 header ignore-dup-ids

Random number seed: 1615541288
128722 MiB RAM detected; reserving 64361 MiB for main workspace.
Using up to 24 threads (change this with --threads).
--bgen: 2834076 variants detected, format v1.2.
2124 samples imported from .sample file to plink2-temporary.psam .
--bgen: plink2-temporary.pgen + plink2-temporary.pvar written.
2124 samples (681 females, 1443 males; 2124 founders) loaded from
plink2-temporary.psam.
2834076 variants loaded from plink2-temporary.pvar.
1 quantitative phenotype loaded (2011 values).
Calculating allele frequencies... done.

Error: Variant ID '10:2365978' appears multiple times in --score file.

Any ideas as to what I am missing here?

Thanks!

[Christopher Chang]

The ignore-dup-ids modifier addresses the reverse scenario: there is a duplicate ID in your .bim/.pvar file, but not in the --score file.  It won't be changed to handle your scenario, because that prevents --score from supporting multiallelic variants properly.  If there is a duplicate ID in the --score file, you really should backtrack and fix how the --score file is generated: make sure variant IDs are unique at that point.

[Anton]

I see.

Thank you for replying so quickly!