Hello it's me again. Now , I was traying to train my own classifier wih SILVA 132, but my terminal gave me an error and I don´t know wtha’s happen. I hope you can help me once again.
The first image is about the log information and de second is my terminal 
Hello Vicky,
That first warning is from an old classifier. You can get a new one from the Qiime2 2022.8 data resources page.
Both errors (the log and the terminal screenshot) is from running out of memory. It takes a lot of RAM to train a classifier, which is why the premade ones are so helpful. Do you have a computer with more RAM you can use for training?
Wich size of RAM is ideal for trainning a classifier? For the nonce I don’t have another computer, but I’m waiting for access to a cluster. 
OK that's perfect!
I would recommend 16 GB. That's what I have on my desktop and I can run most things. (It's hard to estimate exactly how much RAM is needed because this depends on the size and complexity of the database and input data set 
)
For some tasks, more RAM is needed, which is when clusters / servers are helpful.
When running qiime feature-classifier classify-sklearn, you could
set --p-reads-per-batch 2000 and see if it is able to run. That is 10x less reads per batch, which should reduce ram usage, maybe enough to run it on your current 12 GB machine.
This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.
