Seeking Advice for Building a Computer for Qiime2


I am a student majoring in microbiology, and I am looking to purchase a computer for microbial ecology analysis. I'm facing some difficulties because I lack extensive knowledge in this area, but I will provide the information I have in order to seek advice.

I plan to perform microbial ecology analysis using Qiime 2 on Ubuntu Linux, starting from raw Illumina MiSeq 2x300 paired-end sequencing fastq data, with approximately 15 to 30 samples, above 7GB in total size.

My questions are as follows:

    • What are the recommended hardware specifications, including CPU, RAM, GPU, SSD, and any other relevant components, for efficiently handling microbial ecology analysis tasks?
  • Which component(s) of the computer's hardware specification are most crucial for optimizing microbial ecology analysis in Qiime 2?
    • Does the computer's hardware specification have a significant impact on the depth of analysis or the quality of results obtained from Qiime 2?
    • In general, how many hours or days should I expect to invest in processing a dataset of around 7GB with 15 to 30 samples using Qiime 2?

I would greatly appreciate any advice or recommendations regarding these questions.
Thank you in advance for your assistance and insights.

1 Like

Welcome to the forum!
I think this topic is quite covered already and one can find relevant topics by using "search" option on the forum, for example:

I would go for at least 16 GB of RAM and SSD from 256 GB. The higher you can get the better it will be. With CPU I would go for at least 6 if you want to run something in multithreaded mode (with a cost of RAM).

RAM and SSD capacity - You can easily run out of RAM or storage for temporary files while running some heavy tasks or large datasets. But 30 samples should be acceptable.

I guess it either can handle it or not, rather than the quality of the analyses will be dependent on the machine specs,

It depends on the most important component of any bioinformatical analysis - the one between the chair and the monitor. If you don't quite know yet what to do, it can take months, or a couple of days/weeks if you are experienced in the field and experimental design is simple.


From the first link @timanix shared in his reply...

...and within @jwdebelius 's reply therein about using a virtual machine...

I just wanted to reiterate the point that a virtual machine will be your cheapest option if you only plan on doing this for a single experiment. I can't imagine any 16S analysis that would require more than 10s of dollars if (THIS IS A BIG "IF") you have already set up everything you need, and the thing you are doing is renting the components you need to do the job. But two major things with that "IF":

  1. The original poster, @Miseon already stated that he is looking to purchase a computer for analyses, so my notion of renting a virtual machine is kind of besides the point. If you start from the standpoint of requiring some kind of physical machine in your presence, this whole argument is moot.
  2. Even if you were amenable to working through a virtual machine setup, there is an added learning curve to using these resources, and that's on top of what you'll have to spend learning, you know, the whole microbial ecology thing. Don't make school take longer than it has to.

If you do want to consider a virtual machine approach, just remember that you can go (at least) two distinct ways:

  • Through your university resources. If you haven't already, find out if your school has some sort of shared computing resource. Here's one at Northern Arizona University, for instance. Here's a smaller one at my alma mater too! There may be a way for any student like yourself to access such a resource at your university, and if that is true in your case, it means you can purchase a much more stripped down laptop, and instead rely on the university machines to do the heavy lifting. I'd advocate anyone doing research to learn about how to leverage these resources, because at some point in your growing career, you are likely going to run into the problem of requiring a bigger machine that you don't want to pay the cash to build :moneybag: ...
  • You can also go with a non-university virtual machine, accessed either through a commercial vendor like Google, AWS, or Microsoft. As mentioned above, renting a computer from one of these providers is not a 1-step task, and you'll need to learn how to install QIIME on that platform (or run it via Docker), and you'll want to learn about how drives are mounted and how to transfer those data of the rented disk to your local device... but if you do go that route, you can do a lot of repeatable research for a lot less than the cost of a laptop.