SEQanswers

Go Back   SEQanswers > Bioinformatics > Bioinformatics



Similar Threads
Thread Thread Starter Forum Replies Last Post
BLAST server SDPA_Pet Bioinformatics 8 10-31-2013 03:31 PM
Geneious Server 1.5 Now Available Geneious Vendor Forum 0 10-10-2011 07:03 AM
NGS platform: To buy or not to buy? Melissa General 3 03-05-2009 06:15 AM

Reply
 
Thread Tools
Old 02-17-2017, 10:40 AM   #1
FairEnough
Junior Member
 
Location: Milwaukee

Join Date: Feb 2017
Posts: 4
Question I need to buy a server

In my lab, we have two server already. We usually use DNASTAR to make assembly or mapping on the second one. And we use GoldenHelix and Partek on the first one. Because we do not know a lot about Linux, so we all use windows system. I by my self has install a virtual Ubuntu to use some software. Now my boss ask me to order a new server/workstation. The budget is about 10K and we hope we can get more powerful machine than we already have. As a freshman in this realm, I come here to ask everyone's advice. Thank you!
The first one:
CPU: i7-4770K 3.5GHz 8 core
RAM: 32G
DISK: ATA Samsung SSD 840 SCSI Disk Device
Intel Raid 0 Volume SCSI Disk Device
Display: NVIDIA Geforce GTX 760
IDE ATA/ATAPI controllers:Asmedia 106x SATA Controller
ATA Channel 0
ATA Channel 1
Storage controller: Intel(R) Desktop/Workstation/Server Express Chipset SATA RAID Controller

The second one:
CPU: Dual Intel Xeon E5 2670, 16 core each
RAM: 96G
DISK: DELL PERC H310 SCSI DISK Device
Display: Standard VGA Graphics Adapter
IDE ATA/ATAPI controllers: Standard AHCI 1.0 Serial ATA Controller
ATA Channel 0-5
Storage controller: PERC H310 Adapter
FairEnough is offline   Reply With Quote
Old 02-17-2017, 10:59 AM   #2
GenoMax
Senior Member
 
Location: East Coast USA

Join Date: Feb 2008
Posts: 6,781
Default

What is the intended application for the new server/workstation?
GenoMax is offline   Reply With Quote
Old 02-17-2017, 11:34 AM   #3
FairEnough
Junior Member
 
Location: Milwaukee

Join Date: Feb 2017
Posts: 4
Smile

Quote:
Originally Posted by GenoMax View Post
What is the intended application for the new server/workstation?
Will also be like DNASTAR, GoldenHelix, R, Notepad++, NextGEne and mach1 or minimac and so on. I have suggest my boss to install a Unix system.
FairEnough is offline   Reply With Quote
Old 02-17-2017, 11:48 AM   #4
GenoMax
Senior Member
 
Location: East Coast USA

Join Date: Feb 2008
Posts: 6,781
Default

Does not sound like your needs include NGS data analysis. If that is the case you will be fine with either of those configs.
GenoMax is offline   Reply With Quote
Old 02-17-2017, 11:48 AM   #5
gaberudy
Junior Member
 
Location: Bozeman, MT

Join Date: Sep 2010
Posts: 5
Default

Hi, I'm head of engineering at Golden Helix so can answer with some experience on at least that use case.

There are three things to trade-off resources at a server config:

- CPU (cores and speed)
- Memory
- Disk

I didn't see your disk config, bug given the potential applications I think you need a RAID5 with 3x 4TB drives ~10TB of usable space, which takes up a large chunk of your budget already.

If you are doing alignment etc you may need the 96GB of RAM, but you won't for our software. 16GB should be plenty. But maybe high-mem is a requirement which fixes you there as well.

On CPU, the big question is whether you can parallelize a lot of analysis across many cores. I would suspect you can some, but a lot of things will be single-threaded bound and it makes sense to go for a faster GHz vs core count. So a 4 or 6 core (which is really 8-12 hyperthreaded) would be fine with a high GHz.
gaberudy is offline   Reply With Quote
Old 02-17-2017, 12:13 PM   #6
FairEnough
Junior Member
 
Location: Milwaukee

Join Date: Feb 2017
Posts: 4
Default

Quote:
Originally Posted by gaberudy View Post
Hi, I'm head of engineering at Golden Helix so can answer with some experience on at least that use case.

There are three things to trade-off resources at a server config:

- CPU (cores and speed)
- Memory
- Disk

I didn't see your disk config, bug given the potential applications I think you need a RAID5 with 3x 4TB drives ~10TB of usable space, which takes up a large chunk of your budget already.

If you are doing alignment etc you may need the 96GB of RAM, but you won't for our software. 16GB should be plenty. But maybe high-mem is a requirement which fixes you there as well.

On CPU, the big question is whether you can parallelize a lot of analysis across many cores. I would suspect you can some, but a lot of things will be single-threaded bound and it makes sense to go for a faster GHz vs core count. So a 4 or 6 core (which is really 8-12 hyperthreaded) would be fine with a high GHz.

Sorry for not indicating the disk room. The first one is 250GB for SSD and 4TB for RAID 0. The second one is 3x 8TB drives ~21TB of usable space.
FairEnough is offline   Reply With Quote
Old 02-17-2017, 12:23 PM   #7
FairEnough
Junior Member
 
Location: Milwaukee

Join Date: Feb 2017
Posts: 4
Default

Quote:
Originally Posted by GenoMax View Post
Does not sound like your needs include NGS data analysis. If that is the case you will be fine with either of those configs.
We still have some problems when doing the mapping of the whole genome sequence data mapping. We think possibly owing to the RAM. Besides, when doing imputation, the RAM always ran out.
FairEnough is offline   Reply With Quote
Old 02-18-2017, 07:12 PM   #8
gringer
David Eccles (gringer)
 
Location: Wellington, New Zealand

Join Date: May 2011
Posts: 819
Default

For extra "burst" memory, you can use SSD drives (with ideally a PCIe or NVMe interface) as swap, which will allow you to do assembly without completely hosing a desktop computer.
gringer is offline   Reply With Quote
Old 03-30-2017, 05:09 PM   #9
n0rr
Junior Member
 
Location: Tokyo

Join Date: Jul 2010
Posts: 2
Default

https://arxiv.org/abs/1703.10231
A $4,000 Workstation for Mammalian Genome Assembly with Long Reads
"Long-read sequencing has enabled the de novo assembly of several mammalian genomes, but with high cost in computing. Here, we demonstrated de novo assembly of mammalian genome using long reads in an efficient and inexpensive workstation."
n0rr is offline   Reply With Quote
Old 04-05-2017, 12:59 AM   #10
gringer
David Eccles (gringer)
 
Location: Wellington, New Zealand

Join Date: May 2011
Posts: 819
Default

Yes, Canu is a wonderful piece of software; very nice for long-read assembly.
gringer is offline   Reply With Quote
Old 04-05-2017, 11:49 AM   #11
Timofey Skvortsov
Junior Member
 
Location: UK

Join Date: Apr 2017
Posts: 3
Default

We just got ourselves a new Supermicro-based workstation:
  • 2xE5-2620v4
  • 128GB DDR4 RAM
  • 1TB SSD+2x4TB HDD
  • GeForce GT 730
Less than 4,000, including a 24" IPS monitor.

For 10K USD you can buy an very powerful workstation. As was already mentioned, you will need to adjust the specifications to match your needs. For example, if you aim to do assemblies of large data sets, you want to have as much RAM and storage space as possible, CPU speed and # of cores will be less important. If you mainly run multithreaded programs that benefit from increased CPU clock speed, then invest in better processors. SSD/NVMe will be a good addition, but I would prioritise RAM amount over larger SSDs (a single 512GB SSD should be enough for the system disk). If you work with NGS data sets, having lots of backed-up storage space (HDDs) would never harm.

As for the OS, we use Ubuntu on the workstation and CentOS on our HPC, but if you feel more comfortable with Windows or the majority of the programs you use are Windows-only, then stick with Windows. You will be able to run Linux and Windows on the same machine in a dual-boot configuration, or run Linux in a virtual machine or use Docker. The performance overhead of virtualisation is not that dramatic.
Timofey Skvortsov is offline   Reply With Quote
Old 06-04-2017, 04:32 AM   #12
macarima
Junior Member
 
Location: Seoul, Korea

Join Date: Aug 2014
Posts: 4
Default

I think you need more RAM.
I have a workstation with more RAM and less CPU power.
  • 2x E5-2640 v4
  • 512GB DDR4 RAM
  • 400GB SSD + 2x3TB HDD
  • GTX 1070 for calculation with CUDA

They were just 9,000 USD. I didn't have any problems from TSA projects to WGS projects with genome size under 800 mb.
macarima is offline   Reply With Quote
Reply

Tags
hardware, rnaseq, workstation

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off




All times are GMT -8. The time now is 08:00 PM.


Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2018, vBulletin Solutions, Inc.
Single Sign On provided by vBSSO