Dear all,
Mothur failed to read my distance matrix, I receive the following message:
terminate called after throwing an instance of ‘std::bad_alloc’
what(): std::bad_alloc
[ERROR]: [ERROR]: [ERROR]: [ERROR]: [ERROR]: Abort
Is it related with the capabilities of the computer? or is it an other kind of problem?
I am working with 180 000 sequences.
Thank you for your answer! I don’t know what to do now.
regards,
Tamara
You’ve run out of memory. Are you following along with our Costello data analysis example? I’d be surprised if you really had 180,000 unique, high quality sequences…
Pat
Thank you very much for your answer Patrick! I am just starting with mothur! my initial dataset is 210 000 seqs for 36 samples, from a river.
This memory problem can be solved by using a server or a cluster?
Concerning my data:
I trimmed: mothur > trim.seqs(fasta=Zen.fna, oligos=Zen.oligos, qfile=Zen.qual, minlength=123, maxlength=520, maxambig=0, maxhomop=8, qaverage=20, bdiffs=1, pdiffs=2)
After unique and align:
mothur > summary.seqs(fasta=Zen.trim.unique.align)
Start End NBases Ambigs Polymer
Minimum: 1044 1056 2 0 1
2.5%-tile: 1044 3613 152 0 4
25%-tile: 1044 6418 321 0 5
Median: 1044 8411 380 0 5
75%-tile: 1044 9822 402 0 5
97.5%-tile: 1046 13854 484 0 6
Maximum: 43115 43116 520 0 8
And then I made screen.seqs by lengths and chimera slayer (9000 sequences were removed) and then filter, read.dist and cluster and there it ran out of memory.
I also saw that those are many “uniques”, but what can I do?
Thank you again!
Tamara
Well welcome to the mothur family…
Unfortunately, qaverage=20 doesn’t do much (if anything). I’d strongly encourage you to parallel the Costello Analysis and use qwindowaveage=35 and qwindowsize=50 in trim.seqs (http://www.mothur.org/wiki/Costello_stool_analysis). This will significantly reduce the error rate and the number of unique sequences, which has the effect of decreasing the memory requirements.
Hope this helps,
Pat