Converting from STRUCTURE .prn to PLINK .bed for use in fastSTRUCTURE or ADMIXTURE

189 views
Skip to first unread message

Y Y

unread,
Sep 8, 2022, 1:46:22 PM9/8/22
to structure-software
Currently I've got an admixture model working in STRUCTURE, 3113 individuals and 7615 loci. I want to try using fastSTRUCTURE and ADMIXTURE as well, especially ADMIXTURE's supervised learning function, since ~100 individuals are of unknown origin.

Is there any way to produce a PLINK .bed file from the .prn file I currently have?

Vikram Chhatre

unread,
Sep 8, 2022, 1:50:01 PM9/8/22
to structure-software
What is a '.prn' file? If you have a structure file, PGDSpider will convert to fastSTRUCTURE format. Plink can convert it to ADMIXTURE format (details in ADMIXTURE manual).

On Thu, Sep 8, 2022 at 11:46 AM Y Y <universal...@gmail.com> wrote:
Currently I've got an admixture model working in STRUCTURE, 3113 individuals and 7615 loci. I want to try using fastSTRUCTURE and ADMIXTURE as well, especially ADMIXTURE's supervised learning function, since ~100 individuals are of unknown origin.

Is there any way to produce a PLINK .bed file from the .prn file I currently have?

--
You received this message because you are subscribed to the Google Groups "structure-software" group.
To unsubscribe from this group and stop receiving emails from it, send an email to structure-softw...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/structure-software/5f92edc7-1dc6-4927-b9a2-3e52456f84d8n%40googlegroups.com.

Y Y

unread,
Sep 9, 2022, 1:26:13 PM9/9/22
to structure-software
Hmm, it's just a STRUCTURE matrix. 
Top row is "locus1 locus2 locus3 ... locus7515 locus7516"
Then each row is "SampleName -999 1 1 1 3 3 ... -999 1 1 20" where missing=-999, A=1, C=3, G=7, T=20"

fastSTRUCTURE takes .bed and structure files, so this would file would work if not for it being haploid (fastStructure claims to be diploid only).
ADMIXTURE takes .bed only

Trying a couple of conversions in PGDSpider.
For the STRUCTURE input's settings, I use Haploid ploidy, No phase information, -999 missing code, SNP data type, marker names included (7615), individual names included, No PopData, None recessive alleles.

When I try converting to VCF, an error: "locus1 excluded as it doesn't have a location information", for all loci and the output is empty.

I also try converting to PED, which gives- "Data in individual "S95" does not have the requested ploidy (diploid). Alleles will be coded as homozygot diploid!"
Should I just continue on to ADMIXTURE with this PED?
Reply all
Reply to author
Forward
0 new messages