`--keep` does not handle IID-only files properly when the main dataset contains FIDs

[dvg-p4]

Conditions

With a main dataset that contains FID information, do a --keep ids.txt operation, where the ids.txt file contains a single column of individual IDs, optionally with #IID header.

Expected behavior

The dataset will be filtered to only those IIDs, without regard to FID.

If the first line starts with '#FID' or '#IID', it will be treated as a header line. As long as the first columns are "#FID IID", "#FID IID SID", "#IID", or "#IID SID", PLINK 2 will do the right thing.

Observed behavior

No samples are matched, plink2 errors out with "Error: No samples remaining after main filters."

Full reprex

plink --dummy 10 10 --out input
# [...]
# Dummy data (10 people, 10 SNPs) written to input.bed + input.bim + input.fam .

head -n3 input.fam
# per0 per0 0 0 2 1
# per1 per1 0 0 2 2
# per2 per2 0 0 2 1

echo $'#IID\nper0\nper2\nper4' > ids.txt
cat ids.txt
# #IID
# per0
# per2
# per4

plink2 --bfile input --keep ids.txt --make-pgen --out output
# PLINK v2.00a5.12LM AVX2 Intel (25 Jun 2024)    www.cog-genomics.org/plink/2.0/
# (C) 2005-2024 Shaun Purcell, Christopher Chang   GNU General Public License v3
# Logging to output.log.
# Options in effect:
#   --bfile input
#   --keep ids.txt
#   --make-pgen
#   --out output
# 
# Start time: Thu Jul 18 21:04:31 2024
# 380297 MiB RAM detected, ~273638 available; reserving 190148 MiB for main
# workspace.
# Using up to 96 threads (change this with --threads).
# 10 samples (10 females, 0 males; 10 founders) loaded from input.fam.
# 10 variants loaded from input.bim.
# 1 binary phenotype loaded (5 cases, 5 controls).
# --keep: 0 samples remaining.
# Error: No samples remaining after main filters.
# End time: Thu Jul 18 21:04:31 2024

[dvg-p4]

Same issue if the input is a plink2 fileset:

$ plink2 --bfile input --make-pgen --out p2_input
[...]
Writing p2_input.psam ... done.
Writing p2_input.pvar ... done.
Writing p2_input.pgen ... done.
End time: Thu Jul 18 21:06:36 2024

$ plink2 --pfile p2_input --keep ids.txt --make-pgen --out output
PLINK v2.00a5.12LM AVX2 Intel (25 Jun 2024)    www.cog-genomics.org/plink/2.0/
(C) 2005-2024 Shaun Purcell, Christopher Chang   GNU General Public License v3
Logging to output.log.
Options in effect:
  --keep ids.txt
  --make-pgen
  --out output
  --pfile p2_input

Start time: Thu Jul 18 21:07:04 2024
380297 MiB RAM detected, ~273639 available; reserving 190148 MiB for main
workspace.
Using up to 96 threads (change this with --threads).
10 samples (10 females, 0 males; 10 founders) loaded from p2_input.psam.
10 variants loaded from p2_input.pvar.
1 binary phenotype loaded (5 cases, 5 controls).
--keep: 0 samples remaining.
Error: No samples remaining after main filters.
End time: Thu Jul 18 21:07:04 2024

[dvg-p4]

...it DOES work if there is no FID column in the main dataset input, though:

$ cp p2_input.pvar no_FID.pvar
$ cp p2_input.pgen no_FID.pgen
$ awk 'BEGIN {FS = "\t"; OFS = "\t"; printf "#"}; {print $2, $3, $4}' p2_input.psam > no_FID.psam
$ head -n3 no_FID.psam
#IID    SEX PHENO1
per0    2   1
per1    2   2

$ plink2 --pfile no_FID --keep ids.txt --make-pgen --out output
[...]
--keep: 3 samples remaining.
3 samples (3 females, 0 males; 3 founders) remaining after main filters.
1 case and 2 controls remaining after main filters.
Writing output.psam ... done.
Writing output.pvar ... done.
Writing output.pgen ... done.

[chrchang]

https://www.cog-genomics.org/plink/2.0/input#sample_id_convert

IID-only means FID is treated as 0.