diamond seems to run out of memory

Question

[Ric]

Hi, I ran diamond on a 128GB ram computer with a 2.8Gb genome. However, it appears it ran out of memory. Would there be a way to reduce the amount of memory?

$ diamond makedb --in GenbankAA.fasta --db diamond -p 8
$ diamond blastx --threads 1 -q genome.softmasked.fasta --db diamond -o diamond.matches.tab -e 1e-10 -k 0 --more-sensitive -f 6 sseqid slen sstart send qseqid qlen qstart qend pident length evalue score qcovhsp qframe
...
Searching alignments...  [0.602s]
Deallocating buffers...  [0s]
Clearing query masking...  [0.247s]
Computing alignments... Error: std::bad_alloc

Thank you in advance

Answer 1

[buchfink]

Diamond is limited in processing that long of a query sequence in blastx mode. One thing you can try is to use the frameshift mode (option -F 15). Otherwise, you probably should extract ORFs from the genome and search those.

[Ric]

1. Thank you, -F 15 worked. What does frameshift mode mean and why 15 has been chosen?
2. How would you extract ORFs from the genome?

[GenoMax]

You will need to predict genes and then extract the sequences.

[Ric]

Thank you.

[buchfink]

1. This mode is able to compute alignments that contain a change in the reading frame. 15 is the penalty and only a recommended value. The mode was also designed to handle longer sequences.
2. There are tools for doing this, I'm sure google can help.

[Ric]

I created for each of the 19 chromosomes a FASTA file. Each of the 19 FASTA files is around 150Mb. Is frameshift mode (option -F 15) a good value?

[buchfink]

I have not tested it on this kind of data but you can surely try.