search

kcleal / dysgu

Toolkit for calling structural variants using short or long reads
MIT License
98 stars 12 forks source link

PacBio CLR reads failing with Nanopore setting #35

Closed AQWixom closed 2 years ago

AQWixom commented 2 years ago

Based on another issue posted here, I am running Dysgu with PacBio CLR reads using the Nanopore model. The specific error that it is outputting is:

Exception ignored in: 'dysgu.assembler.topo_sort2' ValueError: Graph contains a cycle. Please report this. n=10, w=10, v=0. Node info n was: 10, 0, 0, 32 ValueError: Graph contains a cycle. Please report this. n=22, w=22, v=0. Node info n was: 6, 1, 0, 32

This occurs thousands and thousands of times using both the run and call subcommands.

Any suggestions?

kcleal commented 2 years ago

Hi @AQWixom,

Thanks for reporting this. What was the mapper and reference genome you used?

You can bypass the error by setting the option contigs False, although the accuracy might be slightly lower.

AQWixom commented 2 years ago

Hi @kcleal

I used pbmm2 and hg38.

I will give that a shot and see if it will finish running through. Thanks!

kcleal commented 2 years ago

I think v1.3.10 should now work properly with pbmm2 - there were some issues with X/= style cigar strings that pbmm2 uses. Closing this for now, but please re-open if the issue persists!

AQWixom commented 2 years ago

The previous version did finish going through. I have updated to the newest version (side note: may want to include a numpy version in the requirements.txt, the numpy version that worked previously does not work with 1.3.10 and running the pip numpy install command when its already installed doesn't update it). I will let you know if there are problems. Thanks!