How can I assembly big genome by abyss with 128 Gb ram

Question

[Mbillah]

I tried to get 140 million read assembly by abyss-pe, But memory could not load it(sparsehash fatal error). As a result I divided my fastq file and created two contig file (70 million reads). is there any quality problem? and now I want to make this contig file one. how can I do this? Can I continue to work with the combined contig file?

[colindaven]

I wouldn't divide up the fastq.

Soapdenovo2 might be more memory efficient for you.

Minia certainly will be, but assemblies are poor in my experience.

Good luck.

If you're really struggling and those tools don't work I could perhaps run assemblies for you if you gave me the exact commands to run.

[h.mon]

Although Minia alone really produces too fragmented assemblies, the GATB-Minia-Pipeline results in better assemblies and keep memory usage low.

Answer 1

[jean.elbers]

ABySS offers a bloom filter (https://github.com/bcgsc/abyss#assembling-using-a-bloom-filter-de-bruijn-graph), so you could redo the assembly using the bloom filter to reduce memory usage.