Entering edit mode
7.6 years ago
pepijn.kooij
•
0
I just deployed abyss 2.0.0 and tried to run the abyss-pe pipeline, using both my own data as well as the provided test data, but the resulting unitig.fa, contig.fa and scaffolds.fa files are empty. The verbose does give statistics for these files, so I'm a bit puzzled as to why these files are empty. I previously ran data with version 1.9.0, and there the files all seemed fine.
Any suggestions?
Cheers Pepijn
Interesting. Is there adequate amount of free disk space where-ever these files are being created?
Yes, more than enough. I actually tried my own data first, with two separate datasets, which I ran through a bash script. Aside from the fact I know there is more than enough space (over 400Gb free), the total amount written for for the second set is larger than the expected size for the three files of the first set, which should have been produced before starting the second set. (I hope I explain myself alright here)
Hi @pepijn.kooij,
That's odd. It's hard to say what went wrong without some more information. Could you post the ABySS log output here (or in a github gist, if you prefer)? Preferably with verbose logging turned on (add
v=-vto yourabyss-pecommand).The verbose is too long to post here, so I deposited on my github: abyss-2.0.0_test_verbose.log
Oh, I just realized: You are probably seeing the -unitigs/-contigs/-scaffolds.fa files as empty because they are symlinks to the -3.fa,-6.fa,-8.fa files.
Btw, the results on the test data set may be worse in 2.0.0 vs. 1.9.0 because I changed some of the default parameter values, in preference of larger-k assemblies. (See the
abyss-pesection of the ChangeLog for 2.0.0: https://github.com/bcgsc/abyss/releases ).I was thinking that it would be something like that, this is different from 1.9.0, am I right? However, there's still a problem. With other assemblers I have similar symlink files, and using $nano file, I get to see the sequences. When I try that with the ones generated by abyss 2.0.0, it just simply opens a new file, and it doesn't seem to recognise any of the symlink files that were generated.
Might there be a bug in the script for generating these files? Or is it somewhere on my side?
I'm not sure. What do you see when you do
ls -lin the assembly directory?It seems to look ok, which makes it even weirder...
Hmm. I can barely read it but those symlinks look a bit funny. Normally they would be relative symlinks that point directly to the files in the assembly directory, whereas your symlinks are absolute paths that have some funny-looking spaces in them.
What O/S are you running this on? Are you trying to run this on cygwin or something?
It appears that @pepijn has blacked out username and the sample names from the screenshot so that is the reason it appears that there are spaces in names.
Yes, for privacy reasons I blacked out those parts. I'm running it on Linux Ubuntu 14.04 with Biolinux 8. The symlinks look correct, but still my files appear empty. Might it be in the permission of the files? I'm very puzzled by why this is. Previously, with 1.9.0, and with other assemblers that create files with symlinks I have had no problems whatsoever, but somehow here I do.
For privacy reasons I blacked out those parts. I'm running it on Linux Ubuntu 14.04 with Biolinux 8. The symlinks look correct, but still my files appear empty. Might it be in the permission of the files? I'm very puzzled by why this is. Previously, with 1.9.0, and with other assemblers that create files with symlinks I have had no problems whatsoever, but somehow here I do