Memory considerations

Hi Mehrbod,

I read your replies on: How much RAM do I need to run the DADA2 program?:

I have related question to ask.

My laptop have the following specs:

  • Intel Core i5-7200U @ 2.50GHz

  • Sockets 1, Cores 2, Logical processors 4

  • 500GBHDD

I would like to install Qiime2 (Native through miniconda) . I have sent my samples for amplicon sequencing (Illumina HiSeq 2500). 2 samples for 16s and 2 samples for 18S (so total 4 samples).

May I know if my laptop able to process the data? Or should I upgrade RAM? If yes, how much is better?

I do not sure yet what function I will be used, but basically I want to run raw data that we receive from company that run our sample. Maybe we will analyze for OTU and taxonomic assignment. I am sorry, because I am still new in this. .

Besides that, I also will use ARB software to analyze sanger sequecing results. (Less than 100 samples.

Can I just install all the software in my computer?

I plan to upgrade RAM to 8GB or 12GB/16GB.

What your opinion based on your experience? Can I still use my laptop for other basic task (internet browsing, MS office) while running the softwares?

I hope you can help me. Thank you.

Hi @ilhamimani

This should probably be posted as a new topic so one of the developers maybe want to re-categorize this.

What is the operating system you are using? You mention a native install so I would guess this is a mac or linux system? If that’s not the case consider that some of your processing power and memory will be spent to run QIIME2 via a virtual machine. You also want to make sure you are running a 64bit OS otherwise the increase in RAM isn’t going to do anything for you. That out of the way, to answer a few of your questions.
Your current system as you describe it is not particularly fit for most datasets. The 2 big steps that are memory and processing intensive are the denoising (DADA2/deblur) steps and the training of a classifier. There’s lot of examples on the forum of people running out of memory trying to perform these steps on machines with better memory and processing power than the one you mention. That being said you are trying to handle only 4 samples, so for the denoising step it might be that for 4 samples what you have is enough to get by. You will probably have better luck with Deblur as far as memory requirements go. Just try it and see what happens! :pray: My guess however is that when its time to train your classifier you will certainly run into memory issues.

Again, there are a lot of factors at play here but I have heard recommendations on the forum of dedicating at least 8GB of ram. I think that’s a good safe start, though I have been able to handle some small datasets with 6GB as well. Never tried with any less…

That depends again on how much of your system you dedicate to these tasks. If you set your system to fully focus on these tasks using the --p-n-thread 0 option (uses all available cores) then I wouldn’t run anything else. Given that you only have 4 cores to start with, I wouldn’t run much else on the computer personally.

Sorry I actually have no idea about this question, someone with experience on your software should be able to answer your question. As long as you’re not running ARB and Qiime2 simultaneously I’m guessing you’ll probably be ok, but don’t quote me on that :stuck_out_tongue:

In summary, if you can, I would recommend trying to get access to a more powerful machine or at the least upgrading your ram as you mentioned, especially if you ever plan on analyzing more than 4 samples at a time.

Hope that answers some of your questions!

4 Likes

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.