10 likes | 115 Views
Cadaques A cluster for our bioinformatic needs. Txema Heredia 1,2 , Ángel Carreño 1,2 , Carles Perarnau 3 , Carlos Morcillo-Suarez 1,2 , and Arcadi Navarro 1,2,4
E N D
Cadaques A cluster for our bioinformatic needs Txema Heredia1,2, Ángel Carreño1,2, Carles Perarnau3, Carlos Morcillo-Suarez1,2, and Arcadi Navarro1,2,4 1 IBE Institut de Biologia Evolutiva UPF-CSIC, Barcelona, Spain. 2 Instituto Nacional de Bioinformática, Spain. 3 Unitat de Suport Tecnològic a Projectes de Recerca, UPF, Barcelona, Spain. 4 Institució Catalana de Recerca i Estudis Avançats (ICREA). Catalonia, Spain. arcadi.navarro@upf.edu INTRODUCTION • Nowadays, the amount of biological data available has increased in such a way that the “friendly” analysis you were used to run have become several week-long monsters which no longer can be faced by your desktop or laptop computers. • It is expected that the number of analysis needed will increase, so we need a solution to be able to run them in a proper way. • That solution is CADAQUES, our cluster system. • It is widely used. Total time consumed so far: 540,349 hours, i.e. 61.7 computing years in only 2.3 real years!! TECHNICAL SPECIFICATIONS • 11x IBM XM 21 blades (same technology as Marenostrum) resulting in: • 16x Intel Xeon E5345 Quad-core @ 2.33GHz CPUs, which allow to run up to 64 single-core jobs • 192 Gb of memory (4 blades with 32 Gb and 4 blades with 16 Gb) • 30 Tb of disk • Sun Grid Engine queue system OPEN TO EVERYONE • The cluster is open to any member of the IBE. • If you are interested in using it, send me an email to txema.heredia@upf.edu, and I will create you an account. • The scientific director of the cluster is Arcadi Navarro, so pester him if you are in a hurry. • QUEUES SYSTEM • In order to submit a job to the cluster, instead of running it directly, you have to submit it to the queue system. This allows the cluster to distribute the jobs in a fair and efficient way. • Fairness. The queue system has a fairness feature that, instead of scheduling the jobs “first in first out”, it distributes it among all the users, preventing a single user to monopolize all the job slots. • High allocation system. The queues system tries to allocate the jobs according to the available cluster resources in a given moment. This allows little demanding jobs to slip through bigger ones, decreasing your waiting time and increasing effectively the cluster resource usage. SOFTWARE • WHAT DO I NEED? • A computer (Windows, Linux and Mac are welcome). • An ssh connection software (Putty for windows, or Linux & Mac • system’s built-in). • An internet connection. • Some Linux usage skills. Don’t panic! It’s easy. • Operative System: Linux CentOS 5.0 (RedHat) Rocks Cluster Distribution • The following bioinformatics software are currently installed: • Haplotype Estimation & Analysis • Fast Phase • Phase • Haploview • LDhat • Phylogenetics • MrBayes • Paml41 • Whole Genome Association Analysis • Plink • Population Genetics • Clumpp • Structure • Cosi • Simcoal2 • ihs • Sweep • Xpehh • Programming languages available in the cluster: • R • Perl • Bioperl • Python • C • Java • Php • MPI • Open MP • Sequence Analysis & Manipulation • Hmmer • Staden • TrimAl • GBlocks • Microbiology • Dotur • Mothur • S-libshuff • Sons • Treeclimber • Sequence Assembly • Caftools • Gap5 • Mira • Sequence Alignment • Blast • T-coffee Time gained by using the cluster • DATA • 30 Tb of disk storage. • Mysql server to store your databases. • Currently hosting a series of public databases: • HapMap • UCSC Genome Browser • Ensembl • Samba server, so you can access your data easily, and use it as a remote backup system. • Web repository. • … but anything can be installed under demand. Feel free to ask!