Frame2seq for protein sequence design
Project description
Frame2seq
Official repository for Frame2seq, a structured-conditioned masked language model for protein sequence design, as described in our preprint Structure-conditioned masked language models for protein sequence design generalize beyond the native sequence space.
Colab notebook
Colab notebook for generating sequences with Frame2seq:
Setup
To use Frame2seq, install via pip:
pip install frame2seq
Alternatively, you can clone this repository and install the package locally:
$ git clone git@github.com:dakpinaroglu/Frame2seq.git
$ pip install Frame2seq
Usage
Sequence design
To use Frame2seq to generate sequences, you can use the design
function.
from frame2seq import Frame2seqRunner
runner = Frame2seqRunner()
runner.design(pdb_file, chain_id, temperature, num_samples)
Arguments
pdb_file
: Path to PDB file.chain_id
: Chain ID of protein.temperature
: Sampling temperature.num_samples
: Number of sequences to sample.save_neg_pll
: Whether to save the per-residue negative log-likelihoods of the sampled sequences.verbose
: Whether to print the sampled sequences and time taken for sampling.
Citing this work
@article{akpinaroglu2023structure,
title={Structure-conditioned masked language models for protein sequence design generalize beyond the native sequence space},
author={Akpinaroglu, Deniz and Seki, Kosuke and Guo, Amy and Zhu, Eleanor and Kelly, Mark JS and Kortemme, Tanja},
journal={bioRxiv},
pages={2023--12},
year={2023},
publisher={Cold Spring Harbor Laboratory}
}
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
frame2seq-0.0.5.tar.gz
(86.4 MB
view hashes)
Built Distribution
frame2seq-0.0.5-py3-none-any.whl
(86.4 MB
view hashes)
Close
Hashes for frame2seq-0.0.5-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 8856b75a31cd3835a8104db75370cc9f0e0bf66b8573aad92810c2fea0057f35 |
|
MD5 | 786107c784c3c9255d7f86c83aee3e10 |
|
BLAKE2b-256 | 5d8ab89d6776e59bfe32f6b77e87a2b29e59cbffd8cbc8c00b5fbdf0cf2e1ece |