Segmentation fault with --Ivcf argument

[wdecoster]

This seems similar to #31

sniffles runs, happily printing out progress, and at the end I get:

                [...........]
		# Processed reads: 6170000
		# Processed reads: 6180000
		# Processed reads: 6190000
	Switch Chr chrY
		# Processed reads: 6200000
Print:
Finalizing  ..
/bin/bash: line 1: 35826 Segmentation fault      (core dumped) sniffles --mapped_reads ngmlr_alignment/anon.bam --vcf sniffles_genotypes/anon.vcf --threads 12 --Ivcf sniffles_combined/calls.vcf 2> logs/sniffles_genotype/anon.log

No vcf is produced, but I can see a "anon.vcf_tmp_genotype" binary file in the output directory.

Unfortunately, I cannot share my input data. The alignment is done using ngmlr, the vcf for --Ivcf is made using Survivor after using sniffles without --Ivcf (which did not lead to this problem).

What can I do to try to debug this?

[fritzsedlazeck]

Can you check if you have the latest version? I know this always sounds stupid. I have made several smaller corrections over the code , but did not form a newer release.

Also can you check for permissions and if that folder exists: "sniffles_genotypes"
Just trying to cross out the obvious.

thanks
Fritz

[wdecoster]

I'm using sniffles version: 1.0.7 (from bioconda). That appears to be the latest version on bioconda, but may not include these corrections you made? I can install from GitHub and try again.

The target folder exists and permissions are okay.

[fritzsedlazeck]

Yes please. I made quite some corrections especially in that part.
It should be easy to install.

Let me know how it goes.
Thansk
Fritz

[wdecoster]

Alright, I will keep you posted on the results.

Installation from source is indeed no problem - our old server with outdated OS has given me quite some headaches with other software.

[fritzsedlazeck]

If you have conda installed you can update your cmake version if it is too old.
e.g.: conda install -c anaconda cmake

[wdecoster]

Our cmake appears to be fine for Sniffles, and the installation went fine without issues (unlike what I'm used to - thanks!)

[fritzsedlazeck]

customer service ;)

[wdecoster]

Absolutely great service.

While I'm rerunning "the workflow" I also changed the reference genome for ngmlr so it will take a while before I can report how things go with the latest version - so no need to stay awake ^_^

[wdecoster]

Oh well, that took even longer than expected, but now I can confirm that there is no longer a segmentation fault in this version. Perhaps it's worth creating another release to allow updating the bioconda version?

[fritzsedlazeck]

Yes. I wil soon make a new release and then update the bioconda version. Thanks for getting back to me. Let me know if there are other issues or suggestions.
Thanks
Fritz