I’m trying to use the classify.seqs() function from mothur using a non-standard reference database obtained from OTUX - https://web.rniapps.net/otux/index.html. I initially tried running the code on locally installed version of mothur on my system containing 32GB ram and 7GB of additional swap, and had an error stemming from exceeding allocated RAM and swap space. I had then increased the swap space to 86GB and the code runs slowly but smoothly without problems and consumes 24GB of the 86GB allocated swap space. So it probably required ~56GB of RAM. The authors of OTUX also claim that they used a 64GB RAM server for their computation, hence it likely explains it.
The problem is, swap space is considerably slower than RAM and also the processing power in my consumer grade PC is limited. I was hoping to utilize the computational power of the galaxy servers to get this done faster. But because of the memory intensive processing that is required, galaxy returns an error “Memory allocated exceeded” at the exact very step of the process that my previous trials with less than sufficient swap used to get stuck at. So, is there any way I can continue to use Galaxy for my analysis?
NOTE: I’m using V3-V5 database of the OTUX