Question: How does Supernova downsample reads when given the
--maxreads=N argument? Does it randomly subsample, or just input the first N reads?
supernova --maxreads argument randomly subsamples with a fixed seed. For a given seed, the random numbers generated are reproducible, which is generally a good thing for bioinformatics applications. To get a different random subsampling, you need to change the order of the reads or the input files. Supernova does not currently have an option for changing the seed.
For more information please see Supernova Assembly Process.