galaxyproject
BUSCO gff file(s) not captured in output when Augustus used, no errors as stderr suppressed
Despite >90% of BUSCOs being matched in a genome, the buso_output.gff file is empty.
I'm not sure why this cat command doesn't get the contents of the .gff files. The path looks fine and the run_*/busco_sequences folder seems to be there, but error messages are redirected. https://github.com/galaxyproject/tools-iuc/blob/bc948510a50635ce98123c9e09ae09b39f1e5cee/tools/busco/busco.xml#L86
It seems that it works when MetaEuk is selected but not Augustus.
I can't submit a bug report but here is a history demonstrating the issue on some tutorial data: https://usegalaxy.eu/u/hymenium/h/busco-gff-test
We observed a similar issue, an empty gff file from Busco Galaxy Version 5.7.1+galaxy0, on Galaxy Australia. Busco. An early version of Busco, Galaxy Version 5.5.0+galaxy0, produces a valid (non-empty) gff file on the same nucleotide sequences. Both versions produce an empty gff file on protein sequences, but I don't know if gff output supported for protein inputs.
@bgruening There is an issue reported by the community for busco version 5.7.1 and it has been fixed for v5.8.0 (see https://gitlab.com/ezlab/busco/-/issues/751) . In addition to that, the size parameter in element_line() of ggplot 3.4.0 is deprecated (see https://github.com/wilkelab/cowplot/issues/190) . size is now replaced by linewidth. BUSCO uses this script , generate.py (https://gitlab.com/ezlab/busco/-/blob/master/scripts/generate_plot.py?ref_type=heads) to generate R code. In the bioconda recipe of busco (https://github.com/bioconda/bioconda-recipes/blob/3d499753c59a546e0e8843b65cd3a88b451cd1fb/recipes/busco/meta.yaml#L2) , r-ggplot2 is >=2.2.1. Therefore, use r-ggplot2 = 2.2.0 (ie see the comment in generate.py) or replace size with linewidth in the generate.py
The complete error message of busco 5.7.1 on Galaxy
