Understanding the roles of agents and humans in virtual lab

[tnnandi]

Hi,

Thank you for the excellent framework, which I’m sure will lead the way for many multi agent frameworks for protein and RNA biology.

I’d appreciate it if you could clarify a couple of questions I have after exploring and running the run_nanobody_design python notebook:

1. Currently, are team members being added manually rather than being dynamically obtained from the team selection step? Was this done to prevent the stochasticity of choices after the final round, and could it potentially be included dynamically after some experimentation with the prompts?

2. After running the entire "run_nanobody_design.ipynb" notebook, it seems that the final outputs are the python codes for running the nanobody design tools, i.e., ESM, AF, and Rosetta (written in discussions/(esm alphafold, rosetta) directories).
   I couldn't find the code with the prompts for the agents for running these tools, and was wondering if we are supposed to execute these steps manually using the scripts generated by the agents, or is it expected that we extend the framework you provided to build our own agents (or reuse some of the existing ones) for these steps, to design the nanobodies?

Thanks