while attempting to assemble a metagenomic sample with megahit with the option meta-large, I noticed that i get two very different results for running megahit with the same parameters on two different clusters.
The only difference in the running parameters come from the fact, that different number of cpus and memory are available on cluster 1 and 2: megahit used 256 threads and 0.9 of 504 Gb memory on one system, and 80 threads and 0.9 of 95 GB memory on the other system.
This results in
number of contigs: 660003 vs 357694
largest contig: 125048 vs 51345
N50: 1196 vs 969
(For more stats see the attached Quast report. I will also attach the megahit options.json)
I already saw that megahit is not deterministic, bit this seems to be to vast a difference. I also checked, that it is the same version of megahit; both installations are megahit 1.2.9
Do you have any idea what may have caused this? And i how can i proceed, so that i can have a trustworthy assembly?
Hello,
while attempting to assemble a metagenomic sample with megahit with the option
meta-large
, I noticed that i get two very different results for running megahit with the same parameters on two different clusters. The only difference in the running parameters come from the fact, that different number of cpus and memory are available on cluster 1 and 2: megahit used 256 threads and 0.9 of 504 Gb memory on one system, and 80 threads and 0.9 of 95 GB memory on the other system.This results in
(For more stats see the attached Quast report. I will also attach the megahit options.json)
I already saw that megahit is not deterministic, bit this seems to be to vast a difference. I also checked, that it is the same version of megahit; both installations are megahit 1.2.9
Do you have any idea what may have caused this? And i how can i proceed, so that i can have a trustworthy assembly?
Thank you very much in advance
Cluster2-options.txt HPC-options.txt report.txt