search

ANGSD / angsd

Program for analysing NGS data.
231 stars 51 forks source link

assoc5 Issue with weights being nan or inf #499

Open katcatalano opened 2 years ago

katcatalano commented 2 years ago

Getting the following error with doAssoc5. The same genotype probability input file was used for doAssoc5 with a different phenotype and it worked fine... Any idea what's going on here? Thanks! -> angsd version: 0.933 (htslib: 1.9) build(May 6 2020 21:25:11) Issue with weights being nan or inf

ANGSD commented 2 years ago

Dear @katcatalano

Can you supply more information including commands used and full error message.

TonyKess commented 2 years ago

I've run into this issue when populations in GWA are highly structured - including PCA covariates has helped. Using some method to normalize quantitative phenotype has worked as well

katcatalano commented 2 years ago

Thanks for the suggestions! Full command is:

angsd -yQuant angsd/SL_pheno.yquant -cov angsd/pc1-10.cov -doAsso 5 -doMaf 4 -minCount 16 -nThreads 20 -beagle angsd/ensembl_appc_angsd_probs_noMaxdepth_maf0.05_minQ30.beagle.gprobs.gz -fai /data/katrinac/APPC_sequencing/Genome/ensembl/Amphiprion_percula.Nemo_v1.dna.toplevel.fasta.fai -out angsd/sl_assoc5_maf0.05_noMaxdepth_unlinkedPC1-3cov_minQ30_minCount16

Using PCA covariates seems to reduce the issue with weighting being Nan or Infinity, but doesn't eliminate it. The phenotype is pretty normally distributed, so I'm not sure that's the issue...

Screen Shot 2022-07-13 at 12 22 51 PM

Screen Shot 2022-07-13 at 12 23 39 PM

krcorder commented 1 year ago

@katcatalano Did you ever find a solution to this? I'm having similar issues. Thanks!