How to actually run it?

[dirkwalther]

Hi, could you please add a short description as to how to run it? I assume that it will be possible to take an nt-sequence set and obtain the embeddings. How? What about different sequence lengths? Padding? What character? And if I have two sets of nt-sequences that I want to classify, can I use your "classification.py"? Thanks, Dirk

[saberhq]

Hi Dirk, Thanks for your message and for checking out ntEmbd!

You're absolutely right — the goal is to take a set of nt-sequences and generate embeddings, but the current GitHub version is an early alpha and missing key features like the embed sub-parser. A stable release with better docs has been delayed a bit due to some personal commitments.

I will update the embd subparser ASAP and let you know here (ETA: end of April). Using that, you will be able to input a set of nt-sequences and generate their embeddings. For downstream analysis you can use any classification methods. A simple ensemble approach as noted in classification.py should work.

I will keep in touch shortly. Cheers,
Saber