No progress from flye-repeat after two weeks about flye HOT 9 CLOSED

Hi,

Yes, the graph construction algorithm might sometimes be slow on highly repetitive genomes, and it is not trivial to make it run in parallel. Given the size of the genome and high repeat content, the computation might take a while. Unfortunately, it is hard to predict how long it might take, since it varies greatly from one dataset to another. The largest genome that we tested so far was ~3G, so 26G genome might have some unexpected bottlenecks.

from flye.

It sounds like it is unlikely to finish any time soon in this case so I think I'll need to investigate other options. I have done an assembly using canu but it seems to only use about half the available data. The repetitiveness is killing all the various tools I've tried unfortunately. Thanks for getting back to me though.

from flye.

You might also try MARVEL (https://github.com/schloi/MARVEL) - it looks like it was originally designed to assemble very large genomes.

from flye.

Yep, I've got that and it was on my list to try after I read the paper on the Axolotl assembly they did but they had 32x PacBio coverage so we're looking at possibly getting some more data. Theirs was also a lot less repetitive than ours. I'm hoping lots more longer reads could resolve the problems. Ilumina de-novo assembly with SOAPdenovo2 produces a lot of contigs (54 million last run) so that's not a great deal of help either. MaSuRCA got stuck (repeats again) and was likely to take 2000 days to finish. Tough one this.

from flye.

Closing this year-old issue. Flye should now be better with respect to the resource usage (still does not scale to axolotl though).

from flye.

How does the -t option behave? I have requested several CPUs in the SLURM script, but did not use the -t option, and flye is only using one CPU on the server node. Would you recommend using the option?

How long would you expect flye to run with a 50m genome, and 10 CPUs and 90g mem per CPU requested?

from flye.

@spe238 -t option sets up the number of threads, which is one by default - please do use it. Some estimates of memory usage/running time are given in the FAQ/Manual.

from flye.

Thank you!

from flye.

Thank you!

from flye.