SEQanswers

Go Back   SEQanswers > Bioinformatics > Bioinformatics



Similar Threads
Thread Thread Starter Forum Replies Last Post
NGS platform: To buy or not to buy? Melissa General 3 03-05-2009 07:15 AM
In Sequence: Preparing for ABI Buy, Invitrogen Stresses Importance of Second-, Third- Newsbot! SOLiD 0 07-29-2008 01:30 PM

Reply
 
Thread Tools
Old 07-08-2009, 08:40 AM   #1
polsum
Member
 
Location: Texas

Join Date: May 2009
Posts: 32
Default Which Computer to Buy?

Hi, Our lab is planning to buy a computer that would be dedicated primarily to the bioinformatic analysis. We will be doing lots of deep-sequencing data analysis including Blasting, reference genome mapping etc.

I would like to request the experts here to please suggest a basic configuration for the computer, particularly the processor, OS, RAM, graphics card. We would prefer to use Windows OS but if someone really prefers Linux, no problems.

Thanks a lot in advance
polsum is offline   Reply With Quote
Old 07-09-2009, 06:41 AM   #2
nilshomer
Nils Homer
 
nilshomer's Avatar
 
Location: Boston, MA, USA

Join Date: Nov 2008
Posts: 1,285
Default

Quote:
Originally Posted by polsum View Post
Hi, Our lab is planning to buy a computer that would be dedicated primarily to the bioinformatic analysis. We will be doing lots of deep-sequencing data analysis including Blasting, reference genome mapping etc.

I would like to request the experts here to please suggest a basic configuration for the computer, particularly the processor, OS, RAM, graphics card. We would prefer to use Windows OS but if someone really prefers Linux, no problems.

Thanks a lot in advance
Linux, 32GB RAM, dual-quad AMD, 2TB of disk, and a low powered GPU.

The reasoning for Linux is that academic bioinformatic programs use Linux, where you can always run vmware (or in some cases wine) to run windows program.

Some NGS applications need >16Gb of RAM, with a lot of CPU. Also the data coming off Illumina/SOLiD data requires a lot of storage, so 2TB is a must.
nilshomer is offline   Reply With Quote
Old 07-09-2009, 08:42 AM   #3
polsum
Member
 
Location: Texas

Join Date: May 2009
Posts: 32
Default

Quote:
Originally Posted by nilshomer View Post
Linux, 32GB RAM, dual-quad AMD, 2TB of disk, and a low powered GPU.

The reasoning for Linux is that academic bioinformatic programs use Linux, where you can always run vmware (or in some cases wine) to run windows program.

Some NGS applications need >16Gb of RAM, with a lot of CPU. Also the data coming off Illumina/SOLiD data requires a lot of storage, so 2TB is a must.
Thank you very much for the reply. Is there a significant advantage in having dual-Quad AMD over dual-quad Xeon processor?
polsum is offline   Reply With Quote
Old 07-09-2009, 10:59 AM   #4
nilshomer
Nils Homer
 
nilshomer's Avatar
 
Location: Boston, MA, USA

Join Date: Nov 2008
Posts: 1,285
Default

Quote:
Originally Posted by polsum View Post
Thank you very much for the reply. Is there a significant advantage in having dual-Quad AMD over dual-quad Xeon processor?
I was just thinking cost. The Nehelem processors are the best (20% faster for my multi-threading applications).
nilshomer is offline   Reply With Quote
Old 07-09-2009, 05:41 PM   #5
allo
Member
 
Location: Davis, CA

Join Date: Jul 2009
Posts: 15
Smile Memory, memory, memory....

Quote:
Originally Posted by nilshomer View Post
Linux, 32GB RAM, dual-quad AMD, 2TB of disk, and a low powered GPU.
Memory will depend on the size of the genomes that you are working with. With 32 GB of memory you can do bacterial genomes up to 10 Mb, assuming that you have ~10 million 35bp single-end reads. If you want to do genomes in the 30-50 Mb range you may need 64-128 Gb of memory and for plants you a looking at $ 1 TB $ . Also, you have to plan on how are you going to store your data and the always essential backups and consider that this type of machines run better in temperature-controlled rooms, they like it cold .
allo is offline   Reply With Quote
Old 07-09-2009, 08:05 PM   #6
nilshomer
Nils Homer
 
nilshomer's Avatar
 
Location: Boston, MA, USA

Join Date: Nov 2008
Posts: 1,285
Default

Quote:
Originally Posted by allo View Post
Memory will depend on the size of the genomes that you are working with. With 32 GB of memory you can do bacterial genomes up to 10 Mb, assuming that you have ~10 million 35bp single-end reads. If you want to do genomes in the 30-50 Mb range you may need 64-128 Gb of memory and for plants you a looking at $ 1 TB $ . Also, you have to plan on how are you going to store your data and the always essential backups and consider that this type of machines run better in temperature-controlled rooms, they like it cold .
With reference genome mapping, for example Human, 32GB should suffice. For assembly of larger genomes, I agree, you need a pretty beefy computer but I don't think this was the original intention.

Nils
nilshomer is offline   Reply With Quote
Old 08-04-2009, 08:16 PM   #7
Torst
Senior Member
 
Location: The University of Melbourne, AUSTRALIA

Join Date: Apr 2008
Posts: 275
Default

Quote:
Originally Posted by nilshomer View Post
The Nehelem processors are the best (20% faster for my multi-threading applications).
The Intel i7 (Nehalem) processors have another advantage. The SSE2 SIMD pipeline is full width, whereas in previous Core2 CPUs it was half width. This means that i7 can execute SSE2 instructions in a single cycle. This would make SSE2-aware software such as SHRIMP much faster.
Torst is offline   Reply With Quote
Old 08-04-2009, 10:48 PM   #8
nilshomer
Nils Homer
 
nilshomer's Avatar
 
Location: Boston, MA, USA

Join Date: Nov 2008
Posts: 1,285
Default

Quote:
Originally Posted by Torst View Post
The Intel i7 (Nehalem) processors have another advantage. The SSE2 SIMD pipeline is full width, whereas in previous Core2 CPUs it was half width. This means that i7 can execute SSE2 instructions in a single cycle. This would make SSE2-aware software such as SHRIMP much faster.
The funny thing about compilers is that at full optimization (say gcc -O3), they do a pretty good job of automatically doing vectorization, loop optimization, as well as taking advantage of specific instruction sets. If you really want optimized code, you could go as far as buy Intel's own compiler, which in some cases can produce another 20-30% faster code.

Nils
nilshomer is offline   Reply With Quote
Reply

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 11:35 AM.


Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2019, vBulletin Solutions, Inc.
Single Sign On provided by vBSSO