Hardware requirement for bacterial NGS analysis

[trexbob]

I'll throw my 2 cents in... from recent experience. Less cores might be slower but not enough RAM can really ruin your day. More RAM is better than faster RAM, if you're on a budget.

[GenoMax]

@chariko If your budget is maxed out at the specs you posted not much you can do but move forward. A dual socket system would be nice but will add significantly to your budget.

As Sven suggested remember to choose a reasonable RAID configuration (i.e. no RAID 0/1) and a backup solution of some kind.

[sklages]

mapping -> many cores
denovo assembly -> lots of RAM

If you want to do both, buy quite a few cores, maybe a dual processor version of the CPU you mentioned, and a lot of RAM. 64GB should be enough for even the ugliest bacterial genomes (of the 5Mb-class). Work with fast HDDs (RAID) and/or SSDs for I/O intensive tasks.
If you work with metagenomic data you could need even more RAM.
Think about a backup solution.

My 2p.

[chariko]

Hi all,

...I am trying to figure out what would be a good hardware for our ngs analysis needs (Illumina) and I hope some of you can help me to choose the best option.
We would like to analyze bacteria genome data from the beginning to the end. That means, alignment, variant detection, annotation, gene expression quantification etc.

I have been reading some posts with similar needs but in my case I am working with small genomes so I suppose it is not necessary such a strong hardware as when you work with large genomes.

Anyway, I think this hardware will be quite enough to perform my analysis but I would like to know your opinion about it.

16 cores
32 GB RAM
10 Tb hard drive

Any other suggestions would be appreciated.

Thanks,

chariko

Do you think it is necessary to have a Xeon processor? I have read a lot of posts telling that one of the most important things is RAM and not as important a high number of cores. According to that, I have redesigned my future computer with the following characteristics:

INTEL CORE i7 4930K 3.4 GHZ SK2011 12MB (6 cores)
64 RAM DDR3 2133 KINGSTON HYPERX BEAST CL11 XMP
16 Tb WD 7200 RPM Hard Drive

As I asked before, do you have any recomendations?

Thanks in advance

[rhinoceros]

At first, the genomes to be analyzed will have a size around 2.5 million of pb. Do you think 32 GB RAM will be enough?

It's not only about the size of the genome, but what kind of reads you have (saying Illumina is not very specific) and how many, and what algorithm you use. I can't really say. We used to do metagenomic assemblies on 40 GB RAM and it really wasn't enough. Our new cluster has > 256 GB RAM in each node. But then, I never did any single genome assemblies, so I really don't know.

edit. Might be a good idea to also get a second 10 TB hdd for some RAID configuration so that you don't lose all your data if the drive happens to fail..

[chariko]

Thanks a lot for your answer rhinoceros,

I was thinking about Intel hyper-threading hardware.

At first, the genomes to be analyzed will have a size around 2.5 million of pb. Do you think 32 GB RAM will be enough?

Also, in the future we will have to sequence entire genomes (not novo). Just in case the novo sequenciation would be necessary to be performed, do you think the hardware will be enough to analyze it?

Thanks in advance

[rhinoceros]

Is the CPU AMD or Intel? Intel has hyper-threading, so number of 'efficient' threads is cores x 2. I'm not sure if 32 GB RAM will be enough for assembly of larger prokaryote genomes, but of course this depends on what algorithms you use and how many reads you have and whatnot. I'd also get a 512 GB SSD along with the larger HDD..

[chariko]

Hi all,

I am trying to figure out what would be a good hardware for our ngs analysis needs (Illumina) and I hope some of you can help me to choose the best option.
We would like to analyze bacteria genome data from the beginning to the end. That means, alignment, variant detection, annotation, gene expression quantification etc.

I have been reading some posts with similar needs but in my case I am working with small genomes so I suppose it is not necessary such a strong hardware as when you work with large genomes.

Anyway, I think this hardware will be quite enough to perform my analysis but I would like to know your opinion about it.

16 cores
32 GB RAM
10 Tb hard drive

Any other suggestions would be appreciated.

Thanks,

chariko