few long genes predicted by augustus after the training with busco genes
Hi,
I am using BUSCO to train Augustus. I installed with anaconda busco version 5.4.4
which has augustus version 3.5.0
. I used the following command line:
busco -i genome.fa -l basidiomycota_odb10 -o speciesAnalysed -m genome -c 16 --long –augustus
After the training of Augustus, I noticed that the new trained model is not present in the list of Augustus species. So, I copied the folder of my species of interest inside the busco folder /home/miniconda3/envs/busco5.4.4_env/config/species. After that, I predicted genes in my genome with the command:
augustus --gff3=on ../01_repeat_annotation/genome.fasta.masked --extrinsicCfgFile=extrinsic.BUSCO_ speciesAnalysed.cgf --hintsfile=introns.hints.gff --species=BUSCO_ speciesAnalysed --allow_hinted_splicesites=atac --softmasking=on --outfile= speciesAnalysed.predictionAugustus.gff3 --errfile=augustus.err
However augustus predicted just few long genes. I checked the etraining_err.log and I found two types of errors:
- gene r1.m1.g3.t1 transcr. 1 in sequence Serin_5_570979-575609: Terminal exon doesn't end in stop codon. Variable stopCodonExcludedFromCDS set right?
- gene r1.m1.g1.t1 transcr. 1 in sequence Serin_3_2013793-2025190: coding length not a multiple of 3. Skipping..
Instead using one of the Augustus default species I got more predicted genes.
How can I get more predicted genes?
Thanks in advance,
Concetta