# het. haploid genotypes present (see file.hh ); many commands treat these as missing Warning

[Rodrigo Canovas]

Hi,

I recently have being working over my bim/bed/fam files with plink1.9 and I have been reciving the warning # het. haploit.... 

I have done several testing trying to get rid of this warning with no result. Now I am more interested in understanding what is

the meaning of this warning in my case.

I try to use --split-x -b37  no-fail (and yes, I checked that my data is in build37), but this did not change anything. After that I

checked if the sex (--check-sex) was part of the problem, but all my individuals are set as correct. Finally, I checked the SNP

location of the ones reported in .hh and noticed that all of them are in chr23 and that none of then fall in the boundary zones

defined for b37 (all are in a potions >2699520 and <154931044).

Any idea? Anything else that I could try? what is plink trying to tell me?

Best,

RC

[Christopher Chang]

Are all of the heterozygous haploid calls in PAR1/PAR2, or are none of the calls positioned there?  This is unclear from your last message.

If they're all there, yes, "--make-bed --split-x b37" is supposed to be the solution; can you post the .log file from your failed attempt to use --split-x in that case?

[Rodrigo Canovas]

Hi,

sorry for been unclear.  None of the calls are there. This is the log file:

andom number seed: 1534390884
16047 MB RAM detected; reserving 8023 MB for main workspace.
504403 variants loaded from .bim file.
1231 people (386 males, 845 females) loaded from .fam.
1231 phenotype values loaded from .fam.
Using 1 thread (no multithreaded calculations invoked.
Before main variant filters, 1231 founders and 0 nonfounders present.
Calculating allele frequencies... done.
Warning: 775 het. haploid genotypes present (see chr_af.hh ); many commands
treat these as missing.
Total genotyping rate is 0.99905.
504403 variants and 1231 people pass filters and QC.
Among remaining phenotypes, 1231 are cases and 0 are controls.
Error: No X chromosome loci have bp positions <= 2699520 or >= 154931044. (Use
'no-fail' to force --make-bed to proceed anyway.)

If I add no-fail the warnings are still presented.

[Christopher Chang]

Those are almost certainly variant-calling errors.  You can use "--make-bed --set-hh-missing" to set those calls to missing.

[Rodrigo Canovas]

That command should erase the snps from the .bim file?

So far what I have done is:

>plink1.9 --bfile file --split-x b37 no-fail --make-bed --out file_new

>plink1.9 --bfile file_new --make-bed --set-hh-missing --out final_file

this generated final_file (bim/bed/fam/hh) but I can still find the rsids from hh in the bim file.

[Christopher Chang]

Most of the genotype calls for those rsids are correct, so there's no reason to get rid of them entirely.

[R K]

How do you know that these are almost certainly variant-calling errors?  I have come across the same problem with the checks and after, I have written --set--hh missing I only have 200 instead of 600 sex labelling errors, I had originally. I would like to confirm how do we know whether and when to use the -set-hh

Thanks