Easier development of polymer structure prediction tools.
Project description
Flatland
Easier development of polymer structure prediction tools
The development and debugging of modern neural network approaches for protein structure prediction has the potential to be a somewhat complicated task. Here we provide a framework, Flatland, for simulating training datasets of the form used by current state-of-the-art systems. The central task performed is to evolve a population of polymer sequences, predict their folded structure, produce distance matrix and bond angle tensors, and calculate spectra of per-polymer compound or other polymer interaction likelihoods (emulating protein-protein and compound-protein interaction datasets). This framework intentionally performs such simulations in a highly simplified manner that does not accurately model the real biological processes. Relatedly, it provides the means to simplify the problem to an arbitrary extent - including permitting the reduction from three to two dimensions, reducing the amino acid alphabet from twenty-one to three, and simplifying characteristics of the evolutionary optimization. We provide the framework together with a collection of starting-point notebooks for training neural network models to learn from such data by way of an open source repository to kick-start the development efforts of those interested in contributing to solving this vitally important problem.
Installation
You can install the last released version of Flatland via pip
pip install cg-flatland
Or install the most recent development version of Flatland using git
git clone https://github.com/cayley-group/flatland
pip install -e flatland
Demonstrations
We're putting together various demonstration notebooks which you'll be able to access via the following:
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for cg_flatland-0.0.1-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 9584ea45bb42ddbb6a639f0d997e9a915a7f26ff884d682a12b1cbe49286831a |
|
MD5 | 4a888592e8eeb4928058ff3674de407e |
|
BLAKE2b-256 | d11c46e4299ed552ed61c7fca51873c51c3bb9289bd67e3de3965d7b511c2755 |