Server Requirements for long read data analysis
2
0
Entering edit mode
18 days ago

Hello everyone,

I would like to ask what are the minimum to very good server specifications for setting up a server suitable for long read data analysis. I have only analyzed short read sequencing data before, so requirements such as good GPU are not much of a problem. If your lab also uses long read data, I appreciate your inputs or suggestions. I appreciate all your help!

longreadserver longreadanalysis serverrequirements Server • 512 views
ADD COMMENT
2
Entering edit mode
18 days ago
Michael 55k

Like always, it depends on what you want to do and at what scale. The most resource-intensive step determines the minimum requirements of the system. Do you want to do the basecalling of ONT data? Then, the system requirements for dorado will shape GPU support and vRAM requirements of your system, including the necessity of CUDA (nvidia) support. (Indeed from the description it looks like I could do the basecalling on a MacBook Air) https://dorado-docs.readthedocs.io/en/latest/ Otherwise, it may be assembly with Flye or whatever you want to use. If you already have a system for short read analysis (with a minimum of 128-256GB RAM) and enough storage, it may be sufficient to add CUDA/GPU support for the bassecalling step to it. Also, if you have a machine already, it is worthwhile just trying out some standard NF-core pipelines (https://nf-co.re/genomeassembler/1.0.1/) with a public dataset of the size and organims you will be dealing with.

ADD COMMENT
0
Entering edit mode

EPI2ME workflows available from ONT for sake of completeness: https://nanoporetech.com/products/analyse/epi2me

A number of pre-made NF workflows are available: https://epi2me.nanoporetech.com/wfindex/

ADD REPLY
2
Entering edit mode
17 days ago
  • Basecalling - https://github.com/Kirk3gaard/2025-Crowdsource-GPU-basecalling-stats Note that apple silicon does not seem to work well these days. Go for an nvidia card
  • server or workstation ? Depends on your budget, consumer cards like the 4090 etc are a lot cheaper than the data center A100 H100 etc. 24GB GPU Ram is enough for basecalling, but insufficient for dorado-correct (32GB+)
  • CPU - go for AMD for the number of hyperthreaded cores, they work very well
    • RAM - as much as possible, min 128 GB, more is better

And yes, please say what applications you have in mind, and try some public datasets.

ADD COMMENT

Login before adding your answer.

Traffic: 2962 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6