Ram Required For Ray-1.4.0

Question

[Farhat]

Is there any information on how much RAM is required for Ray to run? I am using Ray-1.4.0 and everytime it crashes after swapping. I am trying to assemble about 30M 100 bp long paired Illumina reads. Is there any way in which RAM usage can be minimized?

Any other suggestions on assembling a Eukaryotic genome?

[Brad Chapman]

The Ray de-novo assembler project page: http://denovoassembler.sourceforge.net/

Answer 1

[Martin A Hansen]

There is a -show-memory-usage switch to Ray.

How to assemble a eukaryotic genome is a whole different question that has been discussed elsewhere.

[Farhat]

I am aware of that switch. I was wondering if there is a formula or benchmarks that could give expected RAM usage given number of reads, length of reads, coverage etc.

Answer 2

[Pablo]

If I remember correctly, somebody told me that they were using 1T to 1.5T to get a genome assembly in a day (yes, that is 1TB of RAM).

Answer 3

[Rob Syme]

I can help with some early tests on fungal genomes (which might not be helpful for you) assembled with Ray version: 1.6.0.

Assembled on 8x8 Intel Xeon 2.93 GHz - Nehalem series cores

---

Genome #1: Estimated genome size ~38MBp. 20 million single-end reads (sorry - oldish data) Assemblies ran in about half an hour.

• k=47: peak memory usage was 29941MB
• k=39: peak memory usage was 30211MB
• k=31: peak memory usage was 19147MB

---

Genome #2: Estimated genome size ~ 40MBp. 43.5 million paired-end reads (21.8 million pairs)

• k=27: peak memory usage was 13364MB

I'll try and update later with more memory usage from the second set of reads and some

[Farhat]

Thanks! That gives some estimates of the RAM involved though I am dealing with an ~1GB genome.