2
$\begingroup$

I know this is a somewhat general or vague question, but I’m interested in your opinions. I must build a desktop for general bioinformatics activities with human genomes. I will work with Python and R libraries, I will also need to install denovo sequence assembly software, alignment tools, variant viewers/browsers, sql, among other software. I don’t have much experience customizing hardware that will be dedicated to bioinformatics and genome data analyses. I’m pretty sure I work with linux and/or iOS. What is the recommended RAM, processor speed, HD capacity, graphics cards, etc. I appreciate your opinions on the matter. Thank you.

$\endgroup$
  • 1
    $\begingroup$ You mention sequence assembly software. For the specs of the machine it is important to know what you will be assembling. Are you assembling genomes or transcriptomes? And will you be assembling human genomes/transcriptomes or other organisms? $\endgroup$ – winni2k Feb 4 at 7:52
3
$\begingroup$

The RAM requirements for some bioinformatics analyses like assembly can be quite high (in the hundreds of Gigabytes). My recommendation is to get a fast laptop. Something with an i7 quad-core processor, 16 GB of RAM, and 1TB of storage should do. Do your analyses on this laptop. If you run out of RAM or storage, then use a cloud service to get access to bigger machines or buy an external hard drive. Once you have a feeling for what your actual needs are, start looking at a dedicated rack server solution in your local institution's server farm or again, in the cloud.

You don't mention any analyses that clearly benefit from a GPU, but as you get more familiar with your analyses, perhaps that will become useful as well.

$\endgroup$
  • $\begingroup$ Thank you. I believe your advice is quite reasonable, especially since I’m not sure yet what species I’ll be working with. In all likelihood, I’ll have access to server based software for large sequences, but the more I can do locally the better. $\endgroup$ – L. Aparicio Feb 4 at 11:51
  • $\begingroup$ I am glad you feel that way. If you think this answer is a good one and is the best answer to your question, then I recommend you accept this answer. $\endgroup$ – winni2k Feb 4 at 13:31
  • $\begingroup$ Yes. I realize how vague my question is. I don’t have much information on what resources I’ll have access to yet, but I was asked to give some specs so my PC can be configured accordingly. My best answer at this point is to be provided with something that can easily be upgraded if needed. Thanks again. $\endgroup$ – L. Aparicio Feb 4 at 13:35
2
$\begingroup$

If you deal with genomics NGS data then heavy RAM will be essential. I don't think you will get this from a reasonable priced laptop.

If you are using a Deep Learning based algorhithms or Bayesian or maximum likelihood heavy duty processors are needed.

I would go for an allround machine. Two hard drives:

  • 0.5 to 1 Tb SSD (I use 1TB),
  • 2 to 4 Tb Stata data storage,
  • 4-core (light) or 8-core, i7 (10-core is common now),
  • 16GB-32GB RAM.

Important: For your purpose if you are doing de novo NGS assemblies use:

  • ALOT of RAM.
  • Sufficient harddrive - make sure this is backed up

If you want evidence why RAM is important simply look at the threads here and on Stackexchange which read ... de novo NGS assembly PC has stopped working?(!).

The following threads give an insight into the sort of problems RAM creates

The following looking like a particularly bad case:

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.