mixer.py fit2 hanging on calc_unified_bivariate_pdf

[EleniFri]

Hi,

I have successfully run ld, snp, fit1 and test1 with SLURM, but when I try to run the mixer.py fit2 command, the script runs normally for 7 hours and then appears stuck for the rest of the 24h on the first line of calc_unified_bivariate_pdf in the log file. My script is:

!/bin/bash

SBATCH --job-name=qqMDDSCZfit2

SBATCH --time=24:00:00

SBATCH --partition=pi_polimanti,scavenge,day,week

SBATCH --cpus-per-task=48

SBATCH --mem-per-cpu=5G

SBATCH --nodes=1

SBATCH --array=1-20

SBATCH --requeue

module load miniconda cd /home/userid/ conda activate mixer cd mixer/ module load GCCcore/10.2.0 module load Boost/1.71.0-GCC-10.2.0 CMake

python3 precimed/mixer.py fit2 \ --trait1-file prepped_sumstats/MDD_noMHC.csv \ --trait2-file prepped_sumstats/SCZ_noMHC.csv \ --trait1-params-file fit1/MDD.fit.rep${SLURM_ARRAY_TASK_ID}.json \ --trait2-params-file fit1/SCZ.fit.rep${SLURM_ARRAY_TASK_ID}.json \ --out fit2/MDD_SCZ.fit.rep${SLURM_ARRAY_TASK_ID} \ --extract ~/1000G_EUR_Phase3_plink/1000G.EUR.QC.prune_maf0p05_rand2M_r2p8.rep${SLURM_ARRAY_TASK_ID}.snps \ --bim-file ~/1000G_EUR_Phase3_plink/1000G.EUR.QC.@.bim \ --ld-file ~/1000G_EUR_Phase3_plink/1000G.EUR.QC.@.run4.ld \ --lib src/build/lib/libbgmg.so \ --qq-plots \ --power-curve

##############################################################################################

MDD_noMHC.csv and SCZ_noMHC.csv were prepared using the sumstats.py command and fit1 ran smoothly. I added --qq-plots and --power-curve because when I tried to do the figures for the univariate models, an error was produced suggesting I added those two flags.

I also tried the same code adding --threads 48, but nothing changed.

Any ideas what the problem could be?

Thank you

[biomguler]

Hi Eleni, I have the same issue; did you find any solutions? Best, Murat