Hi, Our cluster is being use right now by a group that runs a lot of tblastx jobs and those jobs are quite long. We were wondering if there is a way to "accelerate" each job. Our cluster has 8 dual-processor Athlon MP2200 nodes= , each with 2 GB RAM and 120 GB HD. Right now, each of the nodes runs up to two jobs and everything is under the supervision of SGE. We are thinking about creating a custom DB created from nt that would held all of nt translated with 6 frames and have the users use blastx on this. We would en= d up with 6 times the numner of sequences but we would not be hit by the overhead of translating each of the sequence from nt. Is this making sense? Sylvain =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D Sylvain Foisy, Ph. D. Directeur - operations / Project Manager BioneQ - Reseau quebecois de bio-informatique U. de Montreal / Genome-Quebec Adresse postale: Departement de biochimie Pavillon principal 2900, boul. =C9douard-Montpetit Montr=E9al (Qu=E9bec) H3T 1J4 Tel: (514) 343-6111 x.4066 Courriel: sylvain.foisy@bioneq.qc.ca =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D