Making connections between genotype and phenotype has become more important than ever in recent years. Whole exome sequencing will be even more inexpensive as more and better techniques are developed, but in cancer, many of the genes responsible for transformation of normal cells are known. Current software for measuring the impact of non-synonymous single nucleotide polymorphisms is designed for analysis of all proteins. We present a protein impact predictor, PTENpred, that is trained on and meant to predict phenotypes related to a single important protein, PTEN. PTEN is implicated in most cancers and is also connected to germline disorders including Cowden syndrome and autism. Our predictor is designed for use by biologists, clinicians, and laymen interested in interpreting the possible effect of a novel PTEN mutation.
Here, PTENpred is implemented as a Python script. You need Python 2 to run it. If you're on a Mac or on Linux, you probably already have Python, but if you're on Windows, you should go to the Python Software Foundation to find a distribution.
To run PTENpred, first clone and go into repository in your working directory with
git clone https://github.com/seanjohnite/PTENpred.git
cd PTENpred
If you're going to be working with a lot of different Python projects, I would recommend setting up a virtual environment for PTENpred. This will isolate PTENpred packages from other packages you might have installed to limit conflicts. Start with
pip install virtualenv
virtualenv PTENpredenv
These commands will install the virtualenv package to your system and create an interpreter for PTENpred.
source PTENpredenv/bin/activate
will "turn on" the Python version in your terminal. If everything worked, your command prompt should look like this:
(PTENpredenv) computer-name:PTENpred user$
Now, you can install the packages required to run PTENpred just for the PTENpredenv virtual environment.
pip install -r requirements.txt
will install all of the required packages for PTENpred.
cd predict into the folder where the scripts are, and run PTENpred with the following command syntax:
./classifyMutation.py [-c CAT_SPLIT] VARIANT
CAT_SPLIT is a number that designates the way certain groups of mutations are categorized in the training data for each predictor. This is 2 by default, but can also be 22, 3, or 4. VARIANT is a required argument which is the identity of an amino acid change. This consists of the wild type amino acid in one-letter form, the codon position, and the variant amino acid in one-letter form. Try L70P to test it out. L70P designates a change from leucine (L) to proline (P) at the 70th codon. Here's a sample run:
./classifyMutation.py L70P
PTENpred will give the possible categories for a specified prediction as well as the predicted category.
You may also need to run the script once with the command line argument -mk True. This will remove and restore the predictor in the way that your system understands it. You can see all possible command line arguments for the script by running ./classifyMutation.py -h.