troubleshooting an empty output

[devonorourke]

Good afternoon,
I'd like to be able to lift over the annotations from an older assembly (of a mammalian genome) to my new assembly, and I was hoping the software you've developed would do the trick. While I managed to run the program without error, the resulting .gff file for the target is empty, and the unmapped.txt file contains every gene found in the original assembly. Given that these assemblies came from the same species, I wasn't sure what the first steps in troubleshooting might be.
Thanks for any advice you can offer

[agshumate]

Hi,
The first place i would check is in the intermediate_files directory. There should be a fasta file with your gene sequences and a .sam file from the minimap2 output. Check to make sure that the .sam file has gene alignments. If not, try deleting the .mmi and re-running liftoff from the beginning

[devonorourke]

Thanks for the quick response @agshumate ,

For my first attempt at this, the job initially ran out of memory, so I ended up just restarting it. I didn't look into which files were created in that first failed attempt after restarting - I just relaunched the script. However, following your advice, it looks like the .mmi file was created for the target genome assembly, but was empty:

-rw-r--r-- 1 dro49 cluster 438K Dec 18 11:28 mylu_HiC.fasta.fai
-rw-r--r-- 1 dro49 cluster    0 Dec 18 11:28 mylu_HiC.fasta.mmi
-rw-r--r-- 1 dro49 cluster 2.0G Dec 18 11:22 mylu_HiC.fasta

I deleted that .mmi file, as well as the .fai file. I then restarted the same script, and it started to generate the kind of mapping data I was hoping for 😄. I think as far as my initial issue is concerned, your suggestion of deleting the .mmi and rerunning was exactly the solution - thanks! 👍

This sounds a bit silly, but I wonder if it would be of any use to add a flag like --force to delete any of those temporary files and restart from the alignment step to avoid this kind of quirk. Glad you solved my issue though - thanks again

[agshumate]

glad this was the solution! adding a flag like --force is a good idea. others have encountered the issue as well