Closed LironShv closed 5 years ago
Hi Liron,
So there are quite a few things here, lets try to get them working one-by-one. First is Augustus/BUSCO. Can you confirm that the $AUGUSTUS_CONFIG_PATH variable is set to the properly installed version of Augustus? You want to avoid the system using a failed version of your training parameters, those are stored in $AUGUSTUS_CONFIG_PATH/species/species_name
, so to remove a failed set of parameters, remove the proper folder, i.e. you can see what is in there with funannotate species
command.
I'm surprised that the Augustus conda version didn't work on your system, on Linux it is usually fine.
The second problem might be tblastn
, it looks to have failed during your BUSCO run and hence resulted in only finding a few BUSCO models, realistically you should be finding more like ~1000. Multithreading in tblastn
is still broken as far as I know, I've tried to write the code to adapt and default to 1 cpu for this step, but not sure that worked in your case. The easiest solution is to remove the tblastn
from the PATH and install an older version, likely all the way back to 2.2.31.
Take advantage of the funannotate test
module, which will test your system install. You can run funannotate test -t busco -c 12
to test the busco training method using 12 cpus.
After you get Augustus install fixed and validate that you are passing the tests in funannotate test
then look at the docs to see how to use RNA-seq data. You should not be passing RNA-seq reads directly to funannotate predict
, instead you need to run these through the funannotate train
script to generate a transcriptome assembly, etc. Then using those results you run funannotate predict
.
Hi,
Thanks for your quick reply and your advice, I will have another try. Much appreciate your pipeline and work!
cheers!
Hi there,
Just letting you know the cause of my error and it might be useful for other people. Although it was quite obvious in the end.
I didn't create symbolic links of the manually installed executables/scrips in augustus/scrips and augustus/auxprogs to conda/env/funannotate/bin/ . When running funannotate it combined parts of the manually installed augustus and parts of my previous conda install.
Best wishes,
L
Hello
I am running into some issues with running funannotate predict, i am using a conda installation, with manually installed genemark and Augustus (and I am a new to bioinformatics).
funannotate check --show-versions
Firs,t I ran into an issue with my Augustus version, so I installed Augustus manually.
This was my first error output:
After installing augustus from source i reran funannotate predict
I get the following error output
The funannotate-predict.log has an error i traced this back to Hash::Merge not being properly installed and changing the path to perl for all .pl files of genemark but the error presist, but still finshes:
Busco log:
These are the files i have in predict_misc
augustus.initial.training.txt busco/ busco_augustus.gff3 busco_augustus.tmp busco.evm.gff3 busco.evm.proteins.fa busco.final.gff3 busco_genemark.gff3 busco_predictions.gff3 busco_proteins/ busco_proteins.gff3 buscos.bed busco.training.gb busco.training.gb.test busco.training.gb.train busco_transcripts.gff3 busco_weights.txt EVM_busco/ genemark/ genemark.evm.gff3 genemark.evm.gff3.bak genemark.gff genemark.temp.gff genome.softmasked.fa gmhmm.mod hints.ALL.gff hints.all.sort.tmp hints.all.tmp hints.P.gff protein_alignments.gff3 repeatmasker.bed scaffold.sort.order.txt scaffold.sort.rename.txt transcript_alignments.gff3
Augustus.log
I have been trying for a while to find out what is going wrong. Do you have any suggestions?
Best wishes,
Liron