Open olingerc opened 10 years ago
Can you show us the full command? I just tested this with Velvet and VelvetOptimiser and I cannot reproduce the error.
Upon further inspection, I can confirm this is a bug in VelvetOptimiser. I was running VelvetOptimiser from a git branch where I have made a number of changes to the code (mentioned in a previous issue report), and this is one of the things I fixed long ago but forgot about it.
The other solution is to just replace dashes with underscores and you won't get this issue.
Thanks for the reply, I was scratching my head what else I could have done wrong in my pipeline. Changing only the line I mentioned caused the expected coverage optimization at the end to not start at all, so my one line fix was a bit naïve. I am glad you fixed the problem. Is it safe to work with master? Or shall pick out the related commits? Replacing dashes with underscores is of course an option.
You should use this branch for the bug fixes (I haven't touched the master branch yet): https://github.com/sestaton/VelvetOptimiser/tree/topic/reuse_sequences
I should mention that currently the dashes are replaced with underscores because I never had a situation where it was critical. So, if you specify LTCF-1-10
as the prefix, the prefix would become LTCF_1_10
in the output files. My main concern was getting things to work but I do think it should produce the expected output (with dashes), so I'll look into that when I get a chance. That should be an easy fix, but if you want to get things running right away you could always change the names back manually after the program runs, if that is an option.
When using the option -p with a dash in the string, I get strange error messages about wrong category numbers. For example with LTCF-1-10:
After a lot of trial and error I could trace it back to dashes in the prefix. By putting single quotes around the prefix in line 546 I got rid of the error:
I think the problem lies in velvet itself, but am not sure.