Often, we want random access to a collection of (possibly tokenized) sequences that may be too large to hold in memory. numpy can do this via a memmapped array. Is there an equivalent for the sequence arrays used in bionumpy?