These are chat archives for dereneaton/ipyrad

5th
May 2017
leclearnm
@leclearnm
May 05 2017 15:46
Hi @dereneaton @isaacovercast
I am having some issues with the bootstrapping function in tetrad using version 0.5.15
Specifically I am getting a huge range in time to calculate individual bootstrap replicates on my 12 datasets generated from 3 similarity cutoffs and 4 min taxa settings from ipyrad. One dataset ran each replicate in under a minute, while most others required more than 10 minutes per replicate, causing timeouts.
I am running this analysis on a remote cluster, and using the following launcher script
#
#         <------ Setup Parameters ------>
#
#SBATCH -J tetrad-kept-85               # Jorb name
#SBATCH -n 1                       # Tasks per node (processors), 24 per node maximum
#SBATCH -N 1                       # Number of nodes (computers)
#SBATCH -p development            # Queue names: normal, large, development, gpu, vis
#SBATCH -o tetrad-kept-85.o%j            # Screen output filename
#SBATCH -e tetrad-kept-85.e%j            # Error filename
#SBATCH -t 2:00:00                # 48h normal, 24h large, 1h dev
#SBATCH --mail-user=leclearnm@utexas.edu
#SBATCH --mail-type=begin
#SBATCH --mail-type=end
#          <------ Account String ----->
# <--- (Use this ONLY if you have MULTIPLE accounts) --->
##SBATCH -A
#------------------------------------------------------
export OMP_NUM_THREADS=48

ipcluster start --n=48 --profile=ipyrad --daemonize && sleep 120
tetrad -j analysis_tetrad/kept85m8.tet.json -c 48 -b 100 --ipcluster &>tetrad.log
ipcluster stop --profile=ipyrad
Isaac Overcast
@isaacovercast
May 05 2017 15:50
@leclearnm Can you update to the newest version (0.6.19), we are constantly updating and upgrading stuff, so there's a strong chance the new version will work better.
leclearnm
@leclearnm
May 05 2017 18:40
@dereneaton @isaacovercast I have updated and rerun the analysis. For two different data sets with the same number of taxa (57) the average time to calculate a bootstrap is an order of magnitude longer for the matrix that has fewer SNPs