chaplin 0.0.3

A predictor of protein-chaperon interaction

---

About chaplin

chaplin is a deep learning-based predictor of protein-chaperon interaction that works from cDNA sequence. The tool can therefore handle different DNA encodings of the same protein. It can also be used to optimize a protein to interact (or not interact) with chaperons only implementing synonimous mutations.

If you use chaplin in your research, please consider citing:

Installation

Package installation should only take a few minutes with any of these methods (pip, source).

Installing chaplin with pip:

We suggest to create a local conda environment where to install chaplin. it can be done with:

conda create -n chaplin

and activated with

conda activate chaplin

or

source activate chaplin

We also suggest to install pytorch separately following the instructions from https://pytorch.org/get-started/locally/

pip install chaplin

The procedure will install chaplin in your computer.

Installing chaplin from source:

If you want to install chaplin from this repository, you need to install the dependencies first. First, install PyTorch separately following the instructions from https://pytorch.org/get-started/locally/.

Then install the other required libraries:

pip install numpy scikit-learn transformers

Finally, you can clone the repository with the following command:

git clone https://github.com/grogdrinker/chaplin/

Usage

the pip installation will install a script called chaplin_standalone that is directly usable from command line (at least on linux and mac. Most probably on windows as well if you use a conda environemnt).

Using the standalone

The script can take a fasta file or a sequence as input and provide a prediction as output

for a full list of options, use the command

chaplin_standalone -h

if you want to predict the probability of a protein cDNA, just call the standalone in predict mode with the DNA sequence as argument. The script can also take a fasta file as input instead of a single seuqence

as example

chaplin_standalone --command predict  ATGGAAGATGCTAAAAACATCAAGAAGGGTCCGGCT

or, for multiple sequences, do

chaplin_standalone fastaFile.fasta

if you want to find the optimal cDNA enocding for a protein, just call the standalone in optimize mode with the amino acid sequence as argument.

chaplin_standalone --command optimize AWESAMEPRTEINSEQENCEASINPT

this will give a cDNA sequence encoding for the input protein with a maximized probability of interacting with chaperons. If you want to optimize it to not interacting with chaperons, you can use the option --target_optimization, which takes as argument the probability of interaction the protein should be optimized to.

Using chaplin into a python script

chaplin can be imported as a python module

from chaplin.optimize import optimize
from chaplin.predict import run_prediction

seq = "MKYLLPTAAAGLLL"
s = optimize(target_seq=seq)
print("optimized protein cDNA:",s)
p = run_prediction(["ATGGAAGATGCTAAAAACATCAAGAAGGGTCCGGCT","ATGCGAGCAGCA"])
print("cDNA prediction:",p)

Help

For bug reports, features addition and technical questions please contact [e-mail]