Originally posted in the old forum by Lisa on 30.04.2013 - 10:37
I am annotating a yeast species and find that Augustus predicts too many introns.
How can I reduce the number of predicted introns?
Reducing the number of introns
Moderator: bioinf
Re: Reducing the number of introns
by Mario on 30.04.2013 - 10:41
There is an easy trick to reduce the number of predicted introns:
If you do not use hints already create an empty hints file with
Then copy and edit the hints parameters:
The intron line should then look like this
The number at the position where 0.6 stands is a factor (malus) that applies to introns that are not supported by evidence from hints. In the case of an empty hints file that is all introns. Reduce this nonnegative number as long as you get too many introns, increase it when you get too few. The default is 1.
Run augustus like this
There is an easy trick to reduce the number of predicted introns:
If you do not use hints already create an empty hints file with
Code: Select all
touch hints.gff
Code: Select all
cp config/extrinsic/extrinsic.cfg extrinsic.punishintrons.cfg
Code: Select all
intron 1 0.6 M 1 1e+100
Run augustus like this
Code: Select all
augustus --species=saccharomyces --hintsfile=hints.gff --extrinsicCfgFile=extrinsic.punishintrons.cfg genome.fa
Re: Reducing the number of introns
by Karina H on 20.03.2014 - 07:49
comparison of output Vs already known structures
is there a dataset where the number of introns etc are verified experimentally where # introns is well establishedd
so that malus value can be altered until current results match known results.
Also, are there scripts within latest AUGUSTUS release that help make this comparison (new results Vs old/expected results) and in batch mode for gene predictions on whole chromosome scale?
comparison of output Vs already known structures
is there a dataset where the number of introns etc are verified experimentally where # introns is well establishedd
so that malus value can be altered until current results match known results.
Also, are there scripts within latest AUGUSTUS release that help make this comparison (new results Vs old/expected results) and in batch mode for gene predictions on whole chromosome scale?