I'm trying to assign taxonomy and am having trouble figuring out whats going wrong. I downloaded the Silva classifier from qiime2, (https://data.qiime2.org/2023.2/common/silva-138-99-nb-classifier.qza) as a zip file, but when I try to run the command I immediately get the error message "killed". The computer should have ~100 GB left, so I don't think its a memory issue?
Does it refer to the disk storage (SSD, hard drive) or RAM (operative memory)? For example, RAM is crucial for taxonomy assignment and from my experience Silva can require 32 GB or even more of RAM. How much of RAM have you available for the run?
It depends on your machine and how you are running qiime2. If you are using virtual machine, check in the settings how much of RAM is available for VM.
If you don't have enough of RAM on the entire machine, it can be upgraded mechanically (I ordered 32 Gb of RAM to be physically added to my laptop), or you can run analysis on more powerful machine, or you can access HPC, or remote cluster for heavy tasks.
Ok great, thank you so much! Is it normal to be so limited? This is a pretty new computer and I don't have much downloaded other than miniconda, qiime, and microsoft.
RAM is not depended on the amount of software installed, it is a memory that can be allocated for performing various computational tasks. To perform heavy tasks, such as taxonomy classification, you need to find a way to run it on a stronger machine (with higher amount of RAM).
Another option is to try to use GreenGenes classifier, which is trained on a lesser amount of reads and thetefore requires less amount of RAM.
This is the script I'm trying to run. I have $NCORES = 10, is that potentially why I'm running into issues with RAM? I can also include a lower --p-reads-per-batch
I'd recommend only using 1-2 cores with feature-classifier classify-sklearn, for reasons outlined here. As you've noted, you can combine this with --p-reads-per-batch.