Protty: A pretty simple tool for protease prediction

Protty utilizes profile Hidden Markov Models (HMMs) constructed from MEROPS database to predict putative proteases in query protein sequences

Installation

git clone https://github.com/ArtemF42/protty.git
cd protty
pip install -e .

# Not available at the moment
pip install protty

Building profile HMMs

Running protty-build

Note: protty-build uses Clustal Omega to perform multiple sequence alignment and requires it to be installed. By default, Protty assumes clustalo is in your PATH. If this is not the case, you should specify the --clustalo parameter

protty-build [options] $DATABASE

Output

Once the process is complete, the profile HMMs will be available in $DATABASE/profiles. The easiest way to merge them into the database is to use cat

cat $DATABASE/profiles/*.hmm > $DATABASE/merops.hmm

Advanced usage

The protty-build pipeline consists of 3 major steps:

1. Downloading MEROPS data
2. Filtering raw FASTA files
3. Building profile HMMs

Use --skip option if you want to manually control the pipeline. For example, the command below will only download data from the MEROPS server

protty-build --skip 2,3 $DATABASE

Scanning query sequences against the database

Running protty-scan

protty-scan [options] $DATABASE/merops.hmm proteins.faa

Output

By default, protty-scan generates two files named predicted_proteases.tsv and predicted_proteases.faa, located in the working directory. Use --tsv and --faa options to change the default behavior