Having too many short contigs

[chtsai0105]

Description of bug

I launched a job two weeks ago and it finally completed today. It is a soil metagenome sample which have been preprossed by fastp/bfc (I used bfc instead of the built-in bayesHammer since I stuck at subclustering hamming graph like this case #703) and still have 100M paired reads after trimming.

The spades stuck at Running Bulge remover this time for a very long time. After it was done today I checked the number of contigs by grep \> contigs.fasta | wc -l and there are 11,955,485 contigs! I haven't check the length distribution yet but I noticed there are tons of contigs only 55 bp in length. I guess writing those short contigs causes the I/O bottleneck which causes the slowdown of the entire process. But maybe there's another reason that causes the slowdown.

I'm also wondering if that is possible to integrate a min length filter to prevent reporting all these small contigs. If it is truly a I/O bound issue then maybe that can improve the speed.

spades.log

spades.log

params.txt

params.txt

SPAdes version

3.15.5

Operating System

Linux-4.18.0-348.12.2.el8_5.x86_64-x86_64-with-glibc2.28

Python Version

3.11.0

Method of SPAdes installation

conda

No errors reported in spades.log

• Yes

[chtsai0105]

Maybe I was not totally right. There are tons of contigs shorter than 1000 bp, especially around 125 bp.