Zachary Kurtz

@RitaBogorad thanks for the report. You can run in multicore mode (single machine) by supplying `ncores` to `icov.select.params`. This is documented on the [README](https://github.com/zdk123/SpiecEasi#pulsar-parallel-utilities-for-model-selection). If you're on a cluster you...

Can you filter OTUs and try again? I'm not sure where the internals may be locked up and there doesn't seem to be any error message.

That's very odd.. Can you try running the internal function `SpiecEasi:::.phy2mat` on the phyloseq and otu_table? I would expect both to spit out the same underlying data matrix. Thanks.

thanks for pointing that out @fabiodepa Treating the OTUs/features as the columns is standard in statistics / R packages, but I should add that to the troubleshooting guide on the...

@andrewd789 That's somewhat surprising, but I also wouldn't expect runtime to be linear w.r.t lambda.min.ratio. How long does the run take when pulsar.select=FALSE? This should give you an estimate for...

No, not as currently implement - the `spiec.easi` functions expects a count table.

Yes in principle - but not implemented yet. I'll add this as a potential feature in the next release.

try dropping the `lambda.min.ratio`? It seems the final network is not actually getting fit and there actually an empty graph in the `plot_network` call.

The output matrix indeed does not retain OTU names. However, the input order is always retained whether you start with an OTU table as a matrix or phyloseq object. On...

what's wrong with filtering the output matrix - `se1.corr` here?