PLINK1.9 check sex problem

Question

[panacotaforcota]

Hello!

I use plink1.9 with --check-sex option to check sex in data with only one sample. I use --read-freq with a file contained frequencies of variants from chrX of 1000G (obtained by --freq).

My problem is that for all samples I have F=-1 and SNPSEX=2. Full log file:

PLINK v1.90b7 64-bit (16 Jan 2023)             www.cog-genomics.org/plink/1.9/
(C) 2005-2023 Shaun Purcell, Christopher Chang   GNU General Public License v3
Logging to plink.log.
Options in effect:
  --allow-extra-chr
  --check-sex
  --chr X
  --exclude range PRS.txt
  --read-freq 1000G_chrX.frq
  --vcf 5098_23andme_vcf_qc_task_filtered_rm_dup.vcf
  --vcf-half-call m

7617 MB RAM detected; reserving 3808 MB for main workspace.
--vcf: plink-temporary.bed + plink-temporary.bim + plink-temporary.fam written.
(901465 variants skipped.)
24454 variants loaded from .bim file.
1 person (0 males, 0 females, 1 ambiguous) loaded from .fam.
Ambiguous sex ID written to plink.nosex .
--exclude range: 91 variants excluded.
--exclude range: 24363 variants remaining.
Using 1 thread (no multithreaded calculations invoked).
Before main variant filters, 1 founder and 0 nonfounders present.
Calculating allele frequencies... done.
Total genotyping rate is 0.999959.
--read-freq: .frq file loaded.
24363 variants and 1 person pass filters and QC.
Note: No phenotypes present.
--check-sex: 6683 Xchr and 0 Ychr variant(s) scanned, 1 problem detected.
Report written to plink.sexcheck .

I don't know what the reason is.

Answer 1

[chrchang523]

From the documentation at https://www.cog-genomics.org/plink/1.9/basic_stats#check_sex : "0.66 is, of course, still much larger than 0.2, and in most contexts it still justifies a female call. We suggest running --check-sex once without parameters, eyeballing the distribution of F estimates (there should be a clear gap between a very tight male clump at the right side of the distribution and the females everywhere else), and then rerunning with parameters corresponding to the empirical gap."

Unfortunately, your dataset has only 1 sample, so there is no real "distribution of F estimates" to eyeball there. However, since you are using allele frequencies from 1000G, you can try eyeballing the distribution of --check-sex F estimates on THAT dataset, after filtering down to the SNPs overlapping your sample.