unrecognized error in the abyss assembly software

Question

[xuchuan]

Dear all,

I am using the abyss2.0 for genome assembly (kmer=93). The software terminates with errors like this:

"

`/home/genome/R2.trim.fastq': discarded 6014989 reads shorter than 93 bases

`/home/genome/R2.trim.fastq': discarded 2650626 reads containing non-ACGT characters

`/home/genome/R1.trim.fastq': discarded 1765369 reads shorter than 93bases

`/home/genome/R1.trim.fastq': discarded 139412 reads containing non-ACGT characters

[e9b2f3191bc9:00024] Read -1, expected 43688, errno = 1

.....

[e9b2f3191bc9:00024] Read -1, expected 43688, errno = 1

[e9b2f3191bc9:00028] Read -1, expected 43688, errno = 1

[e9b2f3191bc9:00022] Read -1, expected 43696, errno = 1

The minimum coverage of single-end contigs is 1.30526.

The minimum coverage of merged contigs is 3.01333.

Consider increasing the coverage threshold parameter, c, to 3.01333.

***DistanceEst: /biosoftware/conda/lib/libgomp.so.1: version `GOMP_4.0' not found (required by DistanceEst)*

abyss-map: /biosoftware/conda/lib/libgomp.so.1: version `GOMP_4.0' not found (required by abyss-map)****

Mateless 0

Unaligned 0

Singleton 0

FR 0**

RF 0

FF 0

Different 0

Total 0

**abyss-fixmate: error: All reads are mateless. This can happen when first and second read IDs do not match.

*make: * [abyss_kmer93-3.dist] Error 1*

"

I thought it was the lack of GOMP_4.0 causing the error. But Abyss continues to run after reporting the lack of GOMP_4.0, and subsequently report the "All reads are mateless" error. So I am wondering what causes the stop of abyss? The GOMP_4.0 or matelss reads?

Here is my fastq header example

R1 file:

@K00370:34:HMFCVBBXX:2:1101:31558:1279 1:N:0:ATTCCTTT

R2 file:

@K00370:34:HMFCVBBXX:2:1101:31558:1279 2:N:0:ATTCCTTT

Thanks advance for your help.

[benv]

I'm not a Conda user, but it appears that you have a version mismatch between the installed versions of ABySS and gcc (libgomp is a part of gcc that implements multithreading.) In other words, ABySS was compiled with some other version of gcc (libgomp) than what is installed in Conda.

I noticed that similar issues with Conda/gcc have been posted: https://github.com/jupyter/docker-stacks/issues/177 https://github.com/dmlc/xgboost/issues/1786

Maybe looking over those issues will help.

Personally, if I had this kind of problem, I would probably either: (i) fiddle with LD_LIBRARY_PATH, in order to point ABySS to the correct version of libgomp (ii) forget about Conda and just compile ABySS from source using configure, make, make install

This type of issue (applications finding/loading the wrong version of a shared library) is very common on Linux, but it takes some learning/practice to troubleshoot the problems skillfully. This looks like a good introductory article about the subject, if you want to learn more:

http://innovationsts.com/?p=1042

[lieven.sterck]

ABySS stops due to the 'mateless' read issues which (as far as we can see) are in their turn most likely caused by the GOMP_4.0 issue . If abyss-map does not work it can not align the reads, resulting in missing (and/or empty) files eg. the dist files causing abyss to stop