BioCluster
A website for the Ashworth Cluster Computing Initiative(s)
CONDOR Clusters
Condor is a software suite that allows a controlling computer (the head node) to send jobs to a series of slave nodes. It is very richly featured, and includes careful control of job priorities and so on. It runs on LINUX and other flavours of UNIX computers.
Setting up CONDOR
Dan Gaffney has written a short guide to setting up a CONDOR cluster. It is available as a pdf file here
GALAXY
The galaxy cluster is a 16-node freestanding cluster. It is mainly used for large volume BLAST search jobs, Hidden Markov model searches (e.g. Pfam), and the like. Each node has a dedicated hard disk on which are copied all the databases and resources (e.g. programmes) needed for running jobs submitted by users via the head node.
Originally funded by NERC, it is mainly used by the Sequencing Service in fulfilling NERC sequencing commitments, and by the Blaxter lab for comparative genomics. We maintain the "standard" sequence databases on the cluster, such as UniRef, etcetera, as well as custom databases.
To use galaxy you will need
a login on the head node
your databases or programmes on the head node
Contact Ann Hedley in rm 354 for information.
HAL
This is a dynamic cluster, which has a few static (permanent) slave nodes, and will be joined by dynamic slave nodes as we figure how best to do this. The dynamically present nodes will be available to the cluster when they are idle. As we cannot guarantee the operating system environment on the slave nodes, jobs for this cluster may need to be specially tailored. In any case, jobs that require the transfer of large amounts of data (e.g. BLAST-formatted databases >20 Mb) are not suitable for this cluster.
Contact Ann Hedley in rm 354 for information.
cluster name |
what sort of cluster | number of processors and speed | what it is good for | contact for access to cluster |
galaxy |
a static, Linux-based CONDOR cluster | 16 x 2.4 GHz | large volume BLAST jobs, especially those requiring comparisons to large databases such as UniProt or NCBI_nr | Mark Blaxter or Ann Hedley (rm 354) |
hal |
a dynamic, Linux-based CONDOR cluster with variable membership | currently 7 cpus | simulations, large volume BLAST jobs against small databases | Mark Blaxter or Ann Hedley (rm 354) |
taxInspector |
an 8-node (16-process) LINUX-based MPI cluster | 8x dual core 3.4 GHz | MrBayes-type phylogenetic analyses | Martin Jones (rm 354) |
Xgrid |
dynamic Apple Xgrid based cluster | whatever is available; currently ~14 GHz across 5 machines | simulations, large volume BLAST jobs against small databases | Mark Blaxter (rm 355) |
WOOLHOUSE |
SUN Solaris Grid Engine | ? | ? | Mark Woolhouse |
KNOTT |
LINUX-based CONDOR cluster | ? | ? | Sara Knott |