How to use Glimmer to predict orf from Solexa contigs

[anyone1985]

A genome about 6M has about 340 solexa contigs. I'd like to predict genes of each contig with the Glimmer. I don't know to do it from reference genome as training data or if i can use the long-orf to get training data from each contig. I wander if anyone do the same job and how i should do.

[Mark]

Yes, it is possible to do it using the long-orfs program, but I often juct go to NCBI and retrieve genes from their taxonomy browser of a closely related strain or species and use that to do my training. Be sure to use the -r flag when using build-icm. In my experience it helps a lot.

[anyone1985]

Yes, glimmer with the training data is perfect. However, the contigs are parts of the genome. The training data maybe not suitable for the contig. Or I should do the training for every contig?

Yes, it is possible to do it using the long-orfs program, but I often juct go to NCBI and retrieve genes from their taxonomy browser of a closely related strain or species and use that to do my training. Be sure to use the -r flag when using build-icm. In my experience it helps a lot.