distance matrix to large to store into RAM

Hi there,

I have a large distance matrix (2,8 Gb) that doesn’t store into RAM. The read.dist command gives an error after about 2 Gb is stored. Using a subset of my sequences works just fine, so it is really the size of the matrix.
I have already done everything I could think of to reduce the size of the matrix (unique.seqs, screen.seqs after aligning, filtering and running unique.seqs again and finally setting the cutoff to 0.1). Is there anyway I can still perform the downstream analysis, besides switching to another PC?



I assume you are trying to cluster your distance matrix? If so, you can use the hcluster command which doesn’t store the distance matrix in RAM. http://www.mothur.org/wiki/Hcluster