Thank you for articulating my point better than I could myself about the 'cascade of dependencies'.

It seems to me that these platforms/packages are essentially trying to build something almost-as-complex as a genome from scratch, and that trying to avoid those cascades is paramount. However, is this possible in practise, and if so, what would help streamline the process?

Would a generalisable package solution help with this, and what's limiting development of such a solution? Is it because we don't know what such a solution should/ought to look like?