Producing correlation matrix in mothur

[ritikadev]

Hi,

I am working with a large dataset in Mothur (around 350 samples, each with approximately 10,000 reads), and I need to generate a correlation matrix for the OTUs. I attempted to use the SparCC command in Mothur to calculate correlations, but the process seems to stall indefinitely when I run it on my university’s HPC. Despite allocating sufficient computational resources (e.g., high memory and multiple cores), there’s no progress, and no correlation matrix is produced.

Has anyone else faced similar challenges with SparCC in Mothur for large datasets? If so, is there a workaround to successfully run SparCC or any alternative approach within Mothur that might work for large datasets? I’m looking for a way to efficiently handle this without needing to completely switch to external tools, if possible.

Also potentially looking into corr.axes? Not sure how this works or if it can give me what I need.

[pschloss]

Hi -

How many OTUs are you working with? That’s probably more important than the number of samples or the number of reads. Also, where did you get your shared file from - mothur?

It’s been a very long time since I used sparcc (probably since we put it into mothur), but I recall it taking a very long time to run. I’m not certain that sparcc is necessary/warranted if your data have been subsampled to a consistent number of reads. You might try otu.association for correlation between OTUs

The corr.axes function calculates the correlation between an OTUs abundance and the position of samples on the x- and y-axes from an ordination plot.

Pat