Differences

This shows you the differences between two versions of the page.

--- automatically_fixing_gene_models [2024/07/16 10:54] – [Running fix_genes_with_false_introns.py] 134.190.232.164
+++ automatically_fixing_gene_models [2024/07/16 11:07] (current) – [Running fix_genes_with_false_introns.py] 134.190.232.164
@@ Line 111: / Line 111: @@
 </code>
-And that's it! It's not the fastest script in the world but should take anywhere between a couple of hours and a day, depending on the size of your genome and BAM file. The time consuming part of the script is assessing the coverage level at each part of the genome, thus, the deeper your coverage, the more time it will take.
+And that's it! It's not the fastest script in the world but should take anywhere between a couple of hours and a day, depending on the size of your genome and BAM file. I'd recommend testing the script out on a smaller GFF3 file, for example one that contains features of only one contig, or a subset of a contig.
-Each newly created gene in the output GFF3 file will have in its source field ''fix_genes.py''. So, if you'd like to just see the newly created genes, you can extract them using ''awk '$2=="fix_genes.py"' <output_GFF3_file> > only_new_genes.gff3''
+Each newly created gene in the output GFF3 file will have in its source field ''fix_genes.py''. So, if you'd like to just see the newly created genes, you can extract them using ''awk '$2=="fix_genes.py"' <output_GFF3_file> > only_new_genes.gff3''. I'd recommend loading the original GFF3, along with the entire updated GFF3, and a GFF3 with just the new genes in IGV. This will allow you to see whether the newly updated genes actually make sense.
+Hopefully they do! If they don't, contact me (Joran) or try to see if you can update the code yourself. So far I've tested the script on Ergobibamus cyprinoides (which has a relatively simple genome structure) and Meteora (which is a bit more complicated). I already had to adjust the code quite a bit to yield sensible results with Meteora so chances are it may not work very well with another genome.
+Remember that this script will not always yield correct genes. It is very simple and looks for just the longest ORFs in a defined region (while respecting supported introns, of course). Sometimes a shorter version of an ORF actually corresponds to a gene. It's up to you to detect these and curate them by hand. This script just does a lot of heavy lifting for you, but won't be perfect.