I am currently working on an assembled eukaryotic genome. The file size is 2.5 GB. I have read that people use genscan for de-novo gene prediction. Currently i am unable to figure out how as genscan does not work well with multifasta and large files. I need to produce a .gff file. I used a perl script that was available i.e run_genscan_gff.pl. But that script also encounters the same problem of large file. How does everyone else do this prediction using Genscan?? Any help would be useful right now.