Clearcut Memmory Allocation Error

Hi all,

I’m trying to build a tree with Clearcut through Mothur, but I get the following error:
mothur > clearcut(fasta=nc_bact_unique.filter.fasta, DNA=T)

Clearcut: Memory allocation error in NJ_compute_dmat()
Clearcut: Error computing distance matrix
Clearcut: Failed to build distance matrix from alignment.

I get the same error when I run clearcut directly, so I’m not sure where the problem is. I’ve tried using both the silva and core databases to build my alignment, and I’ve tried using a distance matrix instead of an alignment, but nothing has worked. These are bacterial 16s pyrosequencing samples. I’ve used this command using archaeal, fungi, and a different alignment of bacterial samples before without any problems.

Has anyone seen this particular error or can someone give me some advice? I’m kind of at a loss. Thanks!

-Damon

Weird. Since you’re using 454 data, these errors are probably because your dataset is too large for your computer. Since you get the same problem when you use clearcut directly, I think the problem is on their end and so you’d probably be best served by contacting the clearcut developers. Let us know what you turn up.

I’ll see if I can contact the clearcut developers. This data set is a little larger than the others I’ve been working with, but it’s still strange. I watch my system resources and it doesn’t even get close to using up the maximums before shutting down. I have a linnux system with 16 gb of memory and 8 processors. I use more resources building distance matrices and multiple processors than clearcut accessed. Maybe it’s a built-in maximum? If I learn more, I’ll post a response. If anyone else has any advice or suggestions, please post. Thanks!

-Damon

how big is the distance matrix in gb? getting alignments and calculating distances is actually pretty memory light. also, note that the number of processors won’t matter since clearcut is a single processor operation.

I might try FastTree outside of mothur and then read the resulting tree. FastTree can handle huge datasets and you can take advantage of your multiple processors.

To follow up on jsahl’s comment - the next release of mothur (soon!) will be able to read fasttree files without a problem.