dengzac
commented
3 years ago Hi Daan,
(1) Yes, there is currently a hardcoded limit of 100kb because in our experiments, contigs that reached this length were likely to be inaccurate. I will make this value user-configurable.
(2) Currently, in paired mode, bowtie is used on the entire contig instead of just its edges, which causes memory to be proportional to contig length. I will look into using the --maxins parameter to reduce the amount of contig that is aligned.
(3) Usually most of the time is spent waiting for bowtie alignment results, so I don't see any easy optimizations at the moment.
DaanJansen94
Hi,
I'm running ContExtender on our bioinformatic pipeline after using MetSPAdes and a 1 kb cutoff on the contigs. I have a few questions regarding the tool:
(1) Is it correct that the maximum length of the contig is limited to 100 kb? I saw the following lines a few times already, but I didn't read it in the paper.
Is there a way to go beyond 100 kb?
(2) I am using fastq files with ~ 20 M PE reads & a fasta file with a few hundred contigs (ranging from 1 kb - 150kb). If I run the "extender_wrapper" with the option "--enable-pair-constraint" the job always crashes, but if I run it without "--enable-pair-constraint" it does seems to work. I obtain the following error:
/var/spool/torque/mom_priv/jobs/50666468.tier2-p-moab-2.tier2.hpc.kuleuven.be.SC: line 26: 80356 Aborted (core dumped)
My guess is that this is a memory problem, since I can run the "--enable-pair-constraint" on only a few small contigs. Is it possible that is isn't scalable to this input, or can I twist it in a way that it would work?
(3) The tool takes a lot of time to run, especially with this amount of reads & contigs, could I twist it in a way to speed it up but still retain accuracy?
Thanks for the help!
Kind regards,
Daan