gene prediction programs for prokaryotes

Question

[Am.A]

I run Orphelia and MetaGeneMark the results are a set of predicted genes.

How to compute accuracy of these programs( number of True positives, number of false positives and number of false negatives)?

Answer 1

[5heikki]

Unless you have rnaseq or proteomics data, you really can't, can you? To get an idea of prediction accuracy, you could run said programs on some refseq genomes and then compare the predictions to the actual annotations.

p.s. I'm a big fan of prodigal

[Am.A]

Yes I use data sets that contain 400bp reads from E. coli genome. Is there a script to compare the predictions to the actual annotations?

[Juke34]

There is ParsEval (http://parseval.sourceforge.net)