Segmentation fault

[xiaoyezao]

I have six bam files and corset crashes with Segmentation fault after finishing reading the first bam. Is it because of inadequate memory? The job was runn on a 128G cluster.

[nadiadavidson]

Hi,

Running out of memory from one bam seems unlikely with 128G. Can you please send me some more details about the run such as version of corset, size of bams? number of reads, number of transcripts? what command you used to run corset?

Have you checked that the bam files is not corrupted? Does corset run if you give it just one file?

If the bam file is bad, an alternative to using bams is to generate equivalence class counts with salmon, which you could try. There are some instruction on how to do this on the wiki.

Cheers, Nadia.

[luoxun-xl]

I have six bam files and corset crashes with Segmentation fault after finishing reading the first bam. Is it because of inadequate memory? The job was runn on a 128G cluster.

I came across the same problem and situation using Corset v1.09, and I don't know why. However, when I degraded the Corset version from 1.09 to 1.07 using conda, the error disappeared.

[tzaquin]

I have six bam files and corset crashes with Segmentation fault after finishing reading the first bam. Is it because of inadequate memory? The job was runn on a 128G cluster.

I came across the same problem and situation using Corset v1.09, and I don't know why. However, when I degraded the Corset version from 1.09 to 1.07 using conda, the error disappeared.

Same here

[nadiadavidson]

Hi, Thanks all for reporting this. There were some major changes to the data structures introduced into corset in version 1.08, so this may have introduced a bug as well. Version 1.07 is as good in accuracy, so there is no issue reverting to this. If anyone has a small reproducible example, they are welcome to email it to me: [email protected] and I will look to see what's causing the segmentation fault. Cheers, Nadia.

[luoxun-xl]

Hi, Thanks all for reporting this. There were some major changes to the data structures introduced into corset in version 1.08, so this may have introduced a bug as well. Version 1.07 is as good in accuracy, so there is no issue reverting to this. If anyone has a small reproducible example, they are welcome to email it to me: [email protected] and I will look to see what's causing the segmentation fault. Cheers, Nadia.

Today, I use dmesg to print the kernel ring buffer, and I found that Corset had a corresponding message as following,

[Jun27 14:50] corset[2509720]: segfault at 28 ip 00007fb22b5ed101 sp 00007ffce49168c0 error 4 in libc-2.28.so[7fb22b567000+1bc000]

It seems the segmentation fault is related to the dynamic link library libc-2.28.so.

$ file /lib64/libc-2.28.so
/lib64/libc-2.28.so: ELF 64-bit LSB shared object, x86-64, version 1 (GNU/Linux), dynamically linked, interpreter /lib64/ld-linux-x86-64.so.2, BuildID[sha1]=aceb5a6a8e8000a295b4024b2754d117433411ef, for GNU/Linux 3.2.0, not stripped

However, when Corset v1.07 was used, the error disappear. Hope it helps you

[asan-emirsaleh]

Hello! Very promising tool! When I ran it on a dataset I am working on I have encountered the same issue on files from Kallisto (two-stage processing with -r true-stop). Downgrading to 1.07 did the job. Another thought was to repack bam data with samtools v1.2. So having corset reads files (obtained from .bam files by using corset v1.07) I can proceed with clustering on corset v1.09.