Closed FrostFlow13 closed 1 year ago
Hello, The bug occured when an haplotype was very divergent from the assembly: HairSplitter tried to polish the assembly with the reads, but failed to map the reads on the assembly. I added a reassembly module for the cases where the reads are too divergent. This should also improve the phasing of the large insertions/deletions/inversions. A new version has been pushed and released :-)
Wonderful - thank you so much! I'll give it another try again today (one with multiploid and one without again) and see how it goes.
Sorry for already having another bug to report! I was trying to run Hairsplitter today after the new update (one with the multiploid command, one without, both using multithreading). Hairsplitter ran so much faster this time around, and none of the previously problematic steps seemed to have issues!
However, there seems to be a new issue with STAGE 6.
Running (this one was without multiploid)
Resulted in
And running (this one was with multiploid)
Resulted in
Both appear to be the same error, so I don't think it's the multiploid argument doing anything, but it's definitely not something I encountered previously, and this is the same dataset I ran last week, the only difference being multithreading (but I'm not certain it's multithreading at fault here). I did look at the commits and noticed that
src/tools.cpp
was changed to allow an exit() during polishing, but given that it didn't have a problem during the Minimap2 step (which was also given an exit()), I don't know why racon would end up having an issue.An oddity I just noticed during this - the
variants_filtered.vcf
file never seems to have much added to it, even in my successful runs of Hairsplitter. All it seems to have is:Lastly, I don't know if it's helpful, but it looks like Hairsplitter tried to make a![image](https://github.com/RolandFaure/Hairsplitter/assets/105405804/649cbad2-38af-43a4-95f7-5c1b25ee7937)
/tmp/
directory and atrash.txt
file in my assembly folder (i.e. in the/fs/ess/PAS1802/ALW/2023_06_15-MAY1376_TLOKOs_LongRead/1376/2_flye_assembly-keephap-25kb-2/
directory). I don't know if it has done that every time and simply deleted it later, or if this is a new bug and it simply happened to leave the files there because Hairsplitter died before it could remove them. There are also a lot more files in the proper/8_hairsplitter/tmp/tmp/
directory than there were previously, like it wasn't deleting the files as it was running: