treemix analyses not outputting vertices and edges files and -nans

148 views
Skip to first unread message

Laura Cespedes

unread,
Jun 26, 2024, 11:08:41 PM6/26/24
to structure-software
Hi all,

I have been trying to run a TreeMix analyses using a set of 109 individuals (160 SNPs, only using SNPs with not missing data, for a preliminary run). I prepared the input file by calling SNPs using the populations programs for STACKS, filtering further with VCFtools (only biallelic SNPs and no missing data) and finally converting the vcf to treemix format using the script here: https://github.com/tkchafin/polyrad_scripts/blob/main/vcf2treemix.py.

When I run treemix, I am running treemix in my cluster using Slurm (executables of my .sbatch, below), and I am not getting any apparent errors in the .out or .err files. However, when the analyses "ends" the .cov and covse matrices are full of -nan instead of numbers, and the vertices and edges files are not produced. The .llik and .treeout files are produced but empty. 

Can someone please give me an insight on what is going wrong? Clearly is something early in the process, but I am not seeing any error messages that would help to troubleshoot. Can the problem be in the formatting of my input file, and if so, do you have any reccomendation of specific things to check to start troubleshooting?

Thank you very much in advance!

All the best,

Laura

---
#here I parallelized runs for different m values, but I am having exactly the same issues if just doing a run for a single m.

module load gcc/12.1.0
module load treemix/1.13
echo "treemix was loaded"

treemix_parallel () {    
local n=$(( $1 - 1 ))
treemix -i ./treemix_data.gz -o ./mig"$1" -root pictus -k 1000 -m "$1" -noss
echo "treemix has finished for m="$1""
}

export -f treemix_parallel

/apps/default/bin/parallel --delay 0.2 -j 12 treemix_parallel ::: {1..6}



syreeta

unread,
Jun 25, 2025, 8:22:25 AMJun 25
to structure-software
hi laura,
Did you figure out the problem? I have also got this multiple times.

best to you
Reply all
Reply to author
Forward
0 new messages