Fungal genome annotation

[Fitoedu]

Hi,

I have sequenced the genome of a fungus with very little information, no EST or protein are aviable.
For the gene prediction i used Augustus whit two different training models, of this program and give me different numbers of genes. As I can validate this?
There is the possibility of combining the two lists of genes?
Anyone knows a better program, I have also used GeneMark and does not work very well.

I am desperate to validate my list of genes.
Thank you very much.

[Hobbe]

Gene prediction has a random part to it, so even if you run the same program twice with exactly the same parameters, you would probably get a different amount of genes. If you try different training organisms, you are almost certain to get different numbers of genes.

Augustus and Genemark are the two programs that have worked best for our fungal genome. You can combine gene models using Evigan, or if the differences are small, you could try visualizing in ,e.g., IGV and then just cutting and pasting in the GFF files.

Also, I am not really certain what you are trying to do. You cannot validate a gene without having some sort of transcript or rnaseq data. Without this any gene model will just be theoretical. If a gene model is 100% similar to genes from several other organisms, and you have reason to believe that the gene in question is conserved, you would be pretty safe in trusting you gene model, but apart from that you need transcripts or something similar.

[shashankgupta]

if i have a fungal genome (approx 50 MB or more), does above mentioned program works ?