search

RosettaCommons / protein_generator

Joint sequence and structure generation with RoseTTAFold sequence space diffusion
https://huggingface.co/spaces/merle/PROTEIN_GENERATOR
MIT License
263 stars 43 forks source link

The start_num argument for the primary inference pipeline does not work. #9

Closed anirudhbhashyam closed 1 year ago

anirudhbhashyam commented 1 year ago

The Problem

The repository was cloned and the environment installed on a machine with the following configuration: Component Config
OS Ubuntu 20.04.5 LTS
RAM 200 GB
GPU NVIDIA NVIDIA A100-SXM4-40GB

While running the following scaffolding command (the PDB was downloaded before hand): 

python ./inference.py --pdb ./7x9e.pdb --out ./results/7x9e_scaffold --contigs 20-24,E813-825,20-24 --start_num 37 --num_designs 2

the designs that are created do not start from 37 but from 0. As seen in stdout below:

Screenshot (81)

The Fix

This seems to be a problem in how the sampler class is initialised which calls reset_design_num instead of setting the design_num attribute from args (where start_num by default is 0). The pull request is here.

0merle0 commented 1 year ago

just merged your pull request, thanks!